University of Southern California Dissertations and Theses

Impacts of building performance on occupants' work productivity: a post occupancy evaluation study

(USC Thesis Other)

Impacts of building performance on occupants' work productivity: a post occupancy evaluation study

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

IMPACTS OF BUILDING PERFORMANCE ON OCCUPANTS’ WORK
PRODUCTIVITY

A Post Occupancy Evaluation study

By

Aishwarya V Swamy

A Thesis Presented to the
FACULTY OF THE USC SCHOOL OF ARCHITECTURE
UNIVERSITY OF SOUTHERN CALIFORNIA
In partial fulfillment of the
Requirements of degree
MASTER OF BUILDING SCIENCE

MAY 2021

Copyright [2021] Aishwarya V Swamy
ii

ACKNOWLEDGEMENTS

The completion of this study would not have been possible without the constant supervision and
inspiration from my committee chair Professor Joon-Ho Choi. I would like to express my sincere
gratitude to my other committee members Professor Marc Schiler and Professor Douglas Noble
for their invaluable inputs. I would also like to thank Professor Karen Kensek for her constant
support and encouragement.
Thank you to my fellow classmates, especially Rushita, Muntaseer, Vidya and Zhequan for their
help. I must thank MBS alumni Nikhita Sapuram for her guidance throughout my time at USC.
Thank you also to Brandt Bradley for his technical assistance.
I am forever grateful to my parents and family for their love and support without whom this would
not have been possible. Lastly, thank you to my friends, Pranav, Akshay, and Neha for helping me
achieve this goal.

iii

CONTENTS

ACKNOWLEDGEMENTS ......................................................................................................... ii

LIST OF TABLES ....................................................................................................................... vi

LIST OF FIGURES .................................................................................................................... vii

ABSTRACT .................................................................................................................................. ix

CHAPTER 1 INTRODUCTION ................................................................................................. 1
1.1 Indoor Environmental Quality (IEQ) .................................................................................... 1
1.2 Post Occupancy Evaluation (POE) ....................................................................................... 2
1.2.1 Types of POE ................................................................................................................. 2
1.2.2 Advantages of POE ........................................................................................................ 3
1.2.3 Limitations of POE ........................................................................................................ 3
1.2.4 Methods of conducting POE .......................................................................................... 4
1.3 Humans as an integrated source in built environment .......................................................... 5
1.4 CMU Data set ....................................................................................................................... 6
1.5 Summary ............................................................................................................................... 6

CHAPTER 2 LITERATURE REVIEW ................................................................................. 7
2.1 IEQ Indicators ....................................................................................................................... 7
2.1.1 Thermal quality indicators ............................................................................................. 7
2.1.2 Indoor air quality indicators ........................................................................................... 9
2.1.3 Visual indicators .......................................................................................................... 11
2.1.4 Acoustic indicators....................................................................................................... 12
2.2 POE in an office environment ............................................................................................. 13
2.3 Importance of integrating IEQ with occupant survey ......................................................... 15
2.3.1 Absenteeism in offices due to psychological stressors and poor IEQ ......................... 15
2.3.2 Occupant health issues due to poor IEQ ...................................................................... 16
2.3.3 Impact on work productivity ........................................................................................ 17
2.3.4 Thermal preference/ visual preference based on human factors .................................. 19
2.4 Impacts of architectural design parameters and POE ......................................................... 21
2.5 Research Objectives ............................................................................................................ 22
2.6 Summary ............................................................................................................................. 22
iv

CHAPTER 3 METHODOLOGY .............................................................................................. 23
3.1 Overview ............................................................................................................................. 23
3.2 Technical Attributes of Building Systems (TABS) ............................................................ 24
3.3 Indoor Environmental Quality measurement (NEAT) ....................................................... 24
3.4 Occupant survey (COPE) .................................................................................................... 25
3.5 Statistical analysis ............................................................................................................... 32
3.5.1 Correlation analysis ..................................................................................................... 35
3.5.2 Simple regression ......................................................................................................... 38
3.5.3 2 Sample t test .............................................................................................................. 39
3.5.4 ANOVA ....................................................................................................................... 39
3.5.5 Data Mining ................................................................................................................. 40
3.5.6 Decision tree ................................................................................................................ 41
3.5.7 Statistical software ....................................................................................................... 41
3.6 Summary ............................................................................................................................. 42

CHAPTER 4 ................................................................................................................................ 43
4.1 Data Overview .................................................................................................................... 43
4.1.1 Based on Human factors- age and gender .................................................................... 43
4.1.2 Based on Spatial factors – location and enclosure of workstations. ............................ 44
4.2 Measured IEQ data ............................................................................................................. 46
4.3 Occupant responses ............................................................................................................. 51
4.3.1 Comparison by age ...................................................................................................... 53
4.3.2 Comparison by gender ................................................................................................. 59
4.3.3 Comparison by occupation .......................................................................................... 63
4.3.4 Self-reported productivity ............................................................................................ 67
4.4 Summary ............................................................................................................................. 72

CHAPTER 5 ................................................................................................................................ 74
5.1 Statistical analysis ............................................................................................................... 74
5.1.1 Impacts of IEQ factors on work productivity .............................................................. 74
5.2 Analysis by different occupant group ............................................................................. 76
5.2.1 Gender .......................................................................................................................... 76
5.2.2 Age ............................................................................................................................... 79
5.2.3 Workstation enclosure ................................................................................................. 81
5.2.4 Workstation location .................................................................................................... 83
v

5.3 IEQ Ranking order .............................................................................................................. 86
5.3 Advanced analysis using Machine Learning ...................................................................... 87
5.3.1 Self-reported productivity ............................................................................................ 88
5.3.2 Satisfaction with the indoor environment .................................................................... 94
5.4 Summary ............................................................................................................................. 99

CHAPTER 6 .............................................................................................................................. 101
6.1 Discussion ......................................................................................................................... 101
6.2 Limitations and future work .............................................................................................. 102
6.3 Conclusion ........................................................................................................................ 103

REFERENCES .......................................................................................................................... 107

APPENDIX ................................................................................................................................ 111

vi

LIST OF TABLES

Table 3.1 - Thermal quality and variables measured (IEQ), surveyed (COPE) and observed
(TABS) .......................................................................................................................................... 29
Table 3.2 - Visual quality and variables measured (IEQ), surveyed (COPE) and observed
(TABS) .......................................................................................................................................... 29
Table 3.3 - Air Quality and variables measured (IEQ), surveyed (COPE) and observed (TABS)
....................................................................................................................................................... 30
Table 3.4 - Acoustic Quality and variables measured (IEQ), surveyed (COPE) and observed
(TABS) .......................................................................................................................................... 30
Table 3.5 - Spatial Quality and variables measured (IEQ), surveyed (COPE) and observed
(TABS) .......................................................................................................................................... 30
Table 3.6 - General information of occupants .............................................................................. 31
Table 3.7 - NEAT Variables measured ......................................................................................... 32
Table 4.1 - Distribution of data based on human factors .............................................................. 43
Table 4.2 – Variation results based on age. .................................................................................. 58
Table 4.3 - Variation results based on gender ............................................................................... 62
Table 4.4 - Variation results based on occupation ........................................................................ 66
Table 5.1 - Independent variables considered for decision trees .................................................. 88
Table 5.2 Summary of the decision tree results ............................................................................ 99
Table 6.1 - Comparison between values set by STANDARDS and measured vales (Based on
human factors) ............................................................................................................................ 104
Table 6.2 - Comparison between values set by STANDARDS and measured values (based on
spatial factors) ............................................................................................................................. 104

vii

LIST OF FIGURES

Figure 2.1 - Thermal parameters of the thermal comfort factor (Blyussen 2009) .......................... 8
Figure 2.2 - Mechanical components form one of the sources of indoor air pollutants ............... 10
Figure 2.3 - Visual parameters of indoor space (Blyussen 2009) ................................................. 12
Figure 2.4 - Sound parameters of an indoor space ........................................................................ 13
Figure 2.5 - The Web based tool developed by CBE (Zagreus, Huizenga and Arens 2004) ....... 14
Figure 2.6 - Thermal comfort satisfaction amongst the 215 buildings surveyed (Huizenga,
Abbaszadeh and Zagreus 2006) .................................................................................................... 20
Figure 3.1 - Methodology diagram ............................................................................................... 23
Figure 3.2 - Instrument cart developed at CMU ........................................................................... 25
Figure 3.3 - Occupant survey (page 1) .......................................................................................... 27
Figure 3.4 - Occupant survey (page 2) .......................................................................................... 28
Figure 3.5 - Image of occupant survey data .................................................................................. 33
Figure 3.6 - Image of measured IEQ data compiled on Excel ...................................................... 34
Figure 3.7 - Data distribution and analysis steps .......................................................................... 34
Figure 3.8 - Formula to identify Pearson coefficient (Young 2017) ............................................ 36
Figure 3.9 - Negative correlation between two variables (Iuliano 2019) ..................................... 36
Figure 3.10 - Positive correlation between two variables (Iuliano 2019) ..................................... 37
Figure 3.11 - Null correlation between two variables (Iuliano 2019) ........................................... 37
Figure 3.12 - The formula to derive Spearman's coefficient (Young 2017) ................................. 38
Figure 3.13 - Data mining types and methods (Rokach 2008) ..................................................... 41
Figure 4.1 - Distribution of participants by age and gender ......................................................... 44
Figure 4.2 – Number of open and closed workstations ................................................................ 45
Figure 4.3 – Number of Perimeter and Interior workstations ....................................................... 45
Figure 4.4 – Distribution of temperature in degree Celsius at 1.1 m ............................................ 47
Figure 4.5 – Distribution of temperature in degree Celsius at 0.6 m ............................................ 47
Figure 4.6 – Distribution of temperature in degree C at 0.1m ...................................................... 48
Figure 4.7 – Distribution of Carbon-dioxide levels ...................................................................... 48
Figure 4.8 – Distribution of relative humidity .............................................................................. 49
Figure 4.9 – Distribution of screen illuminance ........................................................................... 50
Figure 4.10 – Distribution of worksurface illuminance ................................................................ 51
Figure 4.11 – Rose chart of survey questions ............................................................................... 52
Figure 4.12 – Distribution of survey responses ............................................................................ 53
Figure 4.13 – Responses distributed by six age groups ................................................................ 54
Figure 4.14 – Rose chart of responses distributed by three age groups ........................................ 55
Figure 4.15 - Interval plot of question 1 to question 7 based on gender ....................................... 55
Figure 4.16 – Interval plot of question 8 to question 17 based on gender .................................... 56
Figure 4.17 – Interval plot question 18 to question 29 based on gender ...................................... 56
Figure 4.18 – Rose chart of survey responses based in gender .................................................... 59
Figure 4.19 – Interval plot of question 1 to question 7 based on gender ...................................... 60
Figure 4.20 - Interval plot of question 8 to question 17 based on gender ..................................... 60
Figure 4.21 - Interval plot of question 18 to question 29 based on gender ................................... 61
Figure 4.22 – Frequency chart based on occupation ..................................................................... 63
Figure 4.23 – Interval plot of question 1 to question 7 based on occupation ............................... 64
Figure 4.24 - Interval plot of question 8 to question 17 based on occupation .............................. 65
viii

Figure 4.25 - Interval plot of question 18 to question 29 based on occupation ............................ 65
Figure 4.26 - Interval plot of question 28 based on age group ..................................................... 67
Figure 4.27 - Interval plot of question 28 based on junior and senior age group ......................... 67
Figure 4.28 - Interval plot of question 28 based on workstation location and enclosure ............. 68
Figure 4.29 - Interval plot of question 28 based on survey responses .......................................... 69
Figure 4.30 - Interval plot of question 28 based on survey responses .......................................... 69
Figure 4.31 - Interval plot of question 28 based on survey responses .......................................... 70
Figure 4.32 – Interval plot of question 28 based on seasons ........................................................ 71
Figure 4.33 – Productivity levels based on seasons and grouped by age and gender ................... 72
Figure 5.1 - Forward regression .................................................................................................... 75
Figure 5.2 - Forward regression .................................................................................................... 76
Figure 5.3 - Confidence interval plot based on gender ................................................................. 77
Figure 5.4 - Confidence interval plots of temperature at 1.1m and gender .................................. 78
Figure 5.5 - Confidence interval plot based on age groups. ......................................................... 80
Figure 5.6 - Confidence interval plots of temperature at 1.1m and age groups ............................ 81
Figure 5.7 - Confidence interval plots based on workstation enclosure ....................................... 82
Figure 5.8 - Satisfaction levels of three survey questions with respect to enclosure of
workstations .................................................................................................................................. 83
Figure 5.9 - Confidence interval plots based on location of workstations.................................... 84
Figure 5.10 - Comparison of location with three survey questions .............................................. 85
Figure 5.11 - IEQ factors ranked according to age group ............................................................. 86
Figure 5.12 - IEQ factors ranked according to junior gender ....................................................... 87
Table 5.1 - Independent variables considered for decision trees .................................................. 88
Figure 5.13 - J48 tree of self-reported productivity based on occupant survey alone .................. 89
Figure 5.14 - J48 tree of self-reported productivity based on occupant survey and two spatial
factors ............................................................................................................................................ 90
Figure 5.15 - J48 tree of self-reported productivity based on occupant surveys, spatial and human
factor ............................................................................................................................................. 91
Figure 5.16 - J48 tree of self-reported productivity based on occupant survey spatial, human
factor and measured data .............................................................................................................. 92
Figure 5.17 - J48 tree of self-reported productivity based on the entire dataset .......................... 93
Figure 5.18 - J48 tree of satisfaction with IEQ based on survey only .......................................... 94
Figure 5.19 - Satisfaction with IEQ based on survey and spatial factor ....................................... 95
Figure 5.20 - Satisfaction with IEQ based on survey, spatial and human factors ........................ 96
Figure 5.21 - Satisfaction with IEQ based on survey, spatial, human factors, and measured data
....................................................................................................................................................... 97
Figure 5.22 - Satisfaction with IEQ based on entire dataset ......................................................... 98

ix

ABSTRACT
The United States energy department’s report indicates that residences, commercial and office
buildings consume more energy than the industrial buildings and the transport sector. The building
sector accounts for nearly half of the energy used in the US. One of the ways to improve building
energy usage is through post occupancy evaluations (POE), it is an effective tool in understanding
the occupants’ needs. The approach of this study is an integrated POE where humans are
considered as valuable factors along with spatial characteristics and Indoor Environmental Quality
(IEQ). Measurable IEQ data, occupant surveys and building characteristics are collected from
1865 office workstations in 60 buildings across the United States. Statistical analysis was done
mainly to study factors affecting occupant work productivity and their satisfaction with the indoor
environment. The collected data was studied categorically in terms of human and spatial factors.
The human factors considered are age and gender and the spatial factors considered are workstation
location and workstation enclosure. Some of the findings were that women and men had similar
thermal comfort preferences in winter and summer seasons. Occupants working in perimeter
workstations and closed workstations were more satisfied than occupants working in interior
workstations and open workstations, respectively. However, the difference in the average values
is not large enough to deduce that workstation location and enclosure matter. Also, the average
lighting levels at work surfaces were more than the recommended range and occupants were very
satisfied with the same. The factors that affect occupants’ work productivity and job satisfaction
has been ranked according to statistical significance. The findings help in enhancing office
environments and occupants' work productivity.

x

HYPOTHESES
1. Amongst all the IEQ factors, thermal comfort has the most impact on work productivity.
2. Workstation location (perimeter or interior) matters when it comes to perceived Indoor
Environmental Quality (IEQ).
3. Occupants working in open plan and closed plan layouts have different job satisfaction
levels.

Key words: Post-occupancy evaluation, indoor environmental quality, self-reported
productivity, occupant satisfaction.

1

CHAPTER 1 INTRODUCTION

This chapter begins by explaining the terms Indoor Environmental Quality (IEQ) and Post
Occupancy Evaluation (POE). It further goes on to explain the methods, types, advantages, and
limitations of POE. It introduces the Carnegie Mellon University dataset and why humans are a
vital source of information for evaluating built environment.
1.1 Indoor Environmental Quality (IEQ)
The term indoor environmental quality can be defined as the conditions of an indoor space. Indoor
environment has always been given enough importance, but it was only in the 19
th
century that
people became aware that diseases and illnesses could spread through bad or contaminated air.
They also started believing in the germ theory of disease which states that most diseases are caused
due to the presence of micro-organisms. This led to discussions on having well-ventilated spaces
and ventilation rates that would help in preventing the spread of the illnesses. Ventilation rates are
a topic of discussion even today.
Gradually, thermal comfort was recognized as an important aspect of an occupant’s overall
comfort. The ancient Greeks and Romans observed that natural and artificial light could be used
to treat minor illnesses. They also considered ways to control noise and noise pollution in the
overall application of acoustics to a space. By 1970s, factors that impacted the indoor environment
were identified.
The indoor environment of a space can broadly be assessed through four aspects, i.e., thermal
quality, visual or lighting quality, indoor air quality and acoustic quality. In addition to this, spatial
qualities, ambience and other factors like personal preferences, age, gender, and clothing levels
have an impact on the occupant’s overall comfort. These factors directly influence one’s state of
2

health and well-being. Poor indoor environmental conditions can bring about varying levels of
infections, allergies, and diseases amongst occupants (Blyussen 2009)
Human beings spend 90% of their time indoors and therefore it is necessary to analyze the factors
that affect them and their productivity. The impacts of IEQ factors on occupants’ health,
well-being and productivity are an important field of research. It has also been demonstrated that
IEQ is directly related to sick building syndrome (SBS).
1.2 Post Occupancy Evaluation (POE)
Post occupancy evaluation (POE) is a broad, organized process of assessing the conditions or
performance of a building once it has been occupied. The process could be used for various
analyses during a building’s life cycle. It could be used to re-set or re-structure certain building
performance variables that can help in achieving better performance results. The idea of
conducting a POE revolves around utilizing building occupants to understand their needs and
preferences. POE is an activity conducted usually to understand if the building is performing
according to the initial set expectations and if the users are satisfied with the same.
The process can help in reviewing design ideas for better implementation in the future. User
surveys can be used to check if the original criteria meet the actual performance. POE can help
facilities management services to learn about comfortable temperature range (Federal and
Facilities Council 2002)
1.2.1 Types of POE
POEs could broadly be classified into three types based on level of investigation:
Indicative POE- This type of POE typically highlights the pros and cons of the performance of a
selected building. Only a small group of people are interviewed along with expert walk throughs
in and around the space. The aim of this POE is mostly to indicate major concerns in the
3

performance of a building (Blyussen 2009). The time required to conduct this type of POE is
usually one or two days (Smithgroup n.d.).
Investigative POE- This POE goes one further than indicative POE. It seeks to identify the reasons
behind the issues and effects of the performance of a building. It also has a particular agenda
associated with it (Blyussen 2009) The time required could vary from one to two months. It
involves literature review and user questionnaires (Smithgroup n.d.).
Diagnostic POE- This is a comprehensive POE. It integrates occupant satisfaction, surveys, and
physical attributes of the immediate environment. This approach requires more time but provides
better results (Blyussen 2009). A Diagnostic POE could take several months to complete. There
are several methods used to conduct a diagnostic POE, such as facility observations, detailed
questionnaires, staff observations and technical measurements (Smithgroup n.d.).
1.2.2 Advantages of POE
Post occupancy evaluations are beneficial to the building industry by providing data on building
performance. The results and findings can help in refining existing standards and policies. It can
assist design makers, architects, and engineers in their future work. Building owners can make
better decisions about the operational cost being incurred. The process acts as a positive feedback
loop where issues or concerns could be identified and rectified. Surveys will help the facility
management services to adjust the existing conditions that might even help in energy and cost
savings (Blyussen 2009).
1.2.3 Limitations of POE
Although there are many benefits of conducting a POE, there are about the same number of
barriers. Building users form an important source of information needed to assess building
performance. However, there is rarely a100% response. A full turnout of occupant surveys will
4

provide an accurate analysis. The resources needed to conduct a POE could sometimes be
expensive. The conventional POE process takes time and is usually labor intensive. However, the
web-based tool developed by the Center for the Built Environment at University of California,
Berkeley (CBE) saves time and resources. Certain skills are required to conduct a POE and not all
organizations have in-house staff to perform one. It is useful for architectural firms to conduct
POEs of the places that they have designed, however the cost is usually not included in their fee
(Hadjri and Crozier 2009) Most importantly, not all POEs have an integrated approach and
sometimes do not consider physical measurements and occupant surveys to correlate data
(Blyussen 2009).
1.2.4 Methods of conducting POE
POEs are usually conducted by handing out questionnaires to building users, by interviewing them,
having expert walkthroughs and site visits. However, these traditional methods take considerable
time and labor. New technologies such as using web-based tools have begun to replace the
traditional methods. Web based tools and questionnaires do not require as much time and can be
circulated quickly. It also reduces utilization of resources. Electronic data can also be converted to
different file types and is easier to import/ export to different platforms. In the US, two federal
agencies started using web-based tools and electronic data in the early 2000’s. The Center for the
Built Environment at University of California, Berkeley (CBE) has collaborated with General
Services Administration (GSA) to conduct a few web based POEs. Using GIS and web-based
feedback to understand existing building conditions like ventilation, thermal comfort, etc., could
inform facilities management staff and the required changes could be accommodated to improve
user satisfaction rates.
5

There is no single comprehensive set of rules or guidelines required to perform every evaluation.
However, a basic process can be explained. The initial step is to identify the target or factor that
must be evaluated. Then, to establish the sources, how much time, cost and human resources are
required, followed by understanding the design of the space, site visits and observation of the same
are conducted. Next, questionnaires and interview formats have to be prepared. Data collection
and analysis forms the next step, and the results can be used to reduce gaps (if any) between the
stated design criteria or performance goal and actual performance (Blyussen 2009).
1.3 Humans as an integrated source in built environment
Building occupants are a vital and inexpensive source of information. Surveys can be used to
restructure the mechanical system parameters and eventually form a positive feedback loop.
The output produced by a company corresponds to how productive its employees are. Higher
employee productivity can be achieved when employees are satisfied with the indoor environment.
Therefore, it is valuable to have an integrated POE to assess conditions and ensure occupant
comfort.
(Schiavon and Peretti 2011) studied ten surveys to understand the characteristics of the various
types of surveys being conducted on indoor environmental quality in office buildings. The ten
surveys were categorized based on time, questionnaires, building type, type of evaluation
conducted, number of occupants, etc. The smallest survey was conducted in an office building
which had 7 administrative offices and the largest survey conducted was the CBE (Center for the
Built Environment at University of California, Berkeley) survey of 600 buildings which had about
60,000 occupants. The study declared that there are two types of surveys: long term and immediate
evaluations. The authors discuss that it is important to have both quantitative (physical
measurements) and qualitative (occupants’ opinion) data of a building. Occupant responses also
6

depend on factors like previous experiences, psychology, ambience, etc. They suggest that surveys
are useful and are required to understand the performance of a building.
1.4 CMU Data set
A team of faculty and trained graduate students from Carnegie Mellon University (CMU) have
collected comprehensive POE data of offices in the public sector across US. The dataset consists
of measured IEQ factors, occupant survey and building characteristics of 1865 workstations in 60
buildings. The occupant satisfaction survey was modified by CMU Professors Vivian Loftness,
and V. Hartopf. The original survey was created by the National Research Council Canada in
collaboration with the US and Canadian Govt (Park 2015).
1.5 Summary
Post Occupancy Evaluation (POE) and measuring Indoor Environmental Quality (IEQ) are mostly
concurrent. To be able to make positive changes and to achieve better building performance, it is
essential to conduct diagnostic and comprehensive POEs. Although there are many diagnostic
POEs conducted, not all have studied building attributes integrated with occupant survey and IEQ.
The next chapter explains more about IEQ and POE.

7

CHAPTER 2 LITERATURE REVIEW
2.1 IEQ Indicators
Indoor environmental quality factors each have multiple influencing parameters. This chapter
begins with descriptions of thermal quality indicators, indoor air quality indicators, visual and
acoustic indicators. It is followed by an explanation of the term productivity with respect to
occupants in a work environment and with respect to buildings. Further, literature reviews explain
the importance of integrating IEQ with occupant survey. Studies show that absenteeism in offices,
poor occupant health and work productivity are related to IEQ. This is followed by studies that
show human factors such as gender and age also play a role in perceiving the indoor environmental
conditions. Architectural design parameters such as open plan office layouts might sometimes have
negative impacts on work productivity. Lastly, research objectives of this thesis are stated.
2.1.1 Thermal quality indicators
There are six parameters that collectively describe thermal comfort. They are temperature, mean
radiant temperature, air velocity and relative humidity, clo value, and metabolic rate.
Temperature: It is the indication of the degree of motion of molecules. It is usually measured in
the Celsius scale or the Fahrenheit scale. The mean radiant temperature is the average, weighted
by viewed angle, of all the surface temperatures in an enclosed space. Air velocity can be defined
as the rate at which air molecules travel or get displaced. It is measured in meter per second (m/s).
Humidity is the amount of water vapor present in air. There are two ways of expressing the amount
of water content in air; absolute humidity and relative humidity. Absolute humidity is the amount
of water content for a given volume of air. Whereas relative humidity is the ratio of the amount of
water present in air to the amount it can hold at a given temperature. It is usually expressed as a
percentage. Clo is the term used to express the thermal resistance provided by a piece of clothing.
8

At a temperature of 21 degree Celsius, one Clo can be defined as the thermal insulation needed to
keep a sitting person comfortable (Blyussen 2009). The energy balance in a person is maintained
when the heat produced is equal to the heat dissipated. This was defined by P.O Fanger, a professor
at the Technical University of Denmark. Metabolic rate represents the amount of heat generated
within the body (Maohui Luo 2018).
According to ASHRAE, thermal comfort is a state of mind that is satisfied with the surrounding
thermal environment. There are two general models used to measure thermal comfort. The most
widely used is the PMV model and the other is the adaptive model. PMV stands for predicted mean
vote. It is a scale that ranges from -3 to +3, where -3 indicates very cold conditions and +3 indicates
very hot conditions. PMV is based on four quantifiable parameters (temperature, air speed, MRT
and relative humidity) and two subject related parameters: clo value and metabolic rate (Syed
Ihtsham ul Haq Gilania 2015)

Figure 2.1 - Thermal parameters of the thermal comfort factor (Blyussen 2009)
Fig 2.1 graphically represents the quantifiable thermal quality parameters where the terms mean,
RHair – Relative humidity of the air inside
Vair – Velocity of indoor air
9

Tair- Temperature of indoor air
Tr1 to Tr6 – Mean radiant temperature of the surfaces (walls and ceiling)
There are studies that show that 21˚C - 25˚C is a comfortable temperature range for an office
environment. In the 25˚C to 30˚C temperature change, a 1˚C increase in temperature can bring
down the occupant performance by 2% (Yousef Al Horr 2016) via (O.A. Sepp€anen 2006). There
are many thermal comfort models being used across the globe by designers and practitioners.
These models are generalized and apply to uniform thermal conditions. They help in designing
mechanical zones based on the number of occupants in a building. However, they do not account
for seasonal and geographical variations and this might lead to inaccuracies while designing or
choosing HVAC systems. They also do not consider subjective factors like age, gender, and
behavior (Yousef Al Horr 2016).
2.1.2 Indoor air quality indicators
Indoor air quality (IAQ) is usually measured with respect to the amount of indoor air pollutants
present. It is dependent on rate of ventilation, rate of emission of pollutants and how concentrated
the pollutants are. Pollutants originate from chemical or biological sources. Chemical pollutants
consist of particulate matter, organic, inorganic, and radioactive gases. Biological pollutants
consist of mold, bacteria, odor, and dust. These bacteria originate from outdoor and indoor
activities like combustion, traffic, outdoor air pollution, furniture, carpets, mechanical systems,
copiers, printers, and occupant habits like smoking, respectively (Cone 1998).
10

Figure 2.2 - Mechanical components form one of the sources of indoor air pollutants
Figure 2.2 shows examples of the mechanical system components that can emit dust and
accumulated debris that pollute the indoor air. New filters also give out Volatile Organic
Compounds (VOCs) (Blyussen 2009). IAQ also depends on outside conditions, weather, building
material and structure and spatial aspects such as furniture, upholstery, and equipment. A key
factor in monitoring IAQ is the ventilation rate. High ventilation rate ensures better air quality.
Studies show that high ventilation rates typically require a lot of energy. In response to that,
researchers argue that higher ventilation rates lead to higher work productivity and benefit the
organization in terms of financial return which is much greater than cost required for operations.
11

2.1.3 Visual indicators
The parameters that describe visual quality indicators are illuminance, luminance, light intensity,
glare, views of the outside, and personal control. Illuminance is the amount of light incident on a
surface. Luminance is the amount of light bouncing from or passing through an object. Light
intensity is the amount of light being emitted by a light source. It is measured in Candela. One
Candela is equivalent to the light intensity of a burning wax candle (Blyussen 2009). Glare is a
condition where one finds it difficult to perceive objects. It can occur in buildings when there is a
direct view of the bright sky. Glare can cause mild or serious discomfort. It is a direct function of
window size and the brightness levels of the sky seen through the window (Hopkinson 1972).
Studies have also proved that having a better view of the outside and being able to control their
view of the outside increase satisfaction levels. Human beings can perceive light that falls in the
visible spectrum, whose wavelengths range between 380nm to 700nm (NASA Science 2012).
There are two sources of light, namely natural light, and artificial light. The natural light falling
on a table basically consists of four parts. It includes sunlight, diffuse daylight, light being reflected
from external surfaces such as buildings, and light being reflected from internal objects such as
furniture. The second source is artificial light coming from fluorescent lamps, LED bulbs, CFLs,
candles, halogen bulbs, etc. (Blyussen 2009)
12

Figure 2.3 - Visual parameters of indoor space (Blyussen 2009)
2.1.4 Acoustic indicators
Acoustic indicators are dependent on the quality of sound and noise and they are highly relevant
in an interior space, especially in an office environment as it can directly affect occupants and their
productivity. Noise can come from external and/or internal sources. In terms of productivity, the
effect a 1-degree Celsius change brings about in an occupant is same as the effect of a change in
noise level of 2.6 decibel. High levels of noise can disturb an occupant and can add to stress and
increase one’s blood pressure levels (Cone 1998).

13

Figure 2.4 - Sound parameters of an indoor space
2.2 POE in an office environment
Productivity is a measure of the outcome relative to the input. There are studies that indicate an
increase in yield due to good work environment yields billions of dollars yearly. The term
productivity is subjective and can have different meanings in different sectors. With respect to
occupants in a work environment, it could be measured in terms of how an individual, a team and
the organization as a whole perform. For buildings, it can be defined as the ratio of operational
cost to operational performance (Yousef Al horr 2016)
(Wargocki 2011) Describes that performance can be measured by how well the task was performed
by an individual and that productivity is both a qualitative and quantitative measure of the work
done. Productivity is a comparison of the actual work done versus the work that would have been
done in ideal conditions. P. Wargocki also mentions that it is a ratio of input to output, where input
value includes the amount spent for equipment, insurance, pay, operational costs etc. and output
value being the occupant performance.
In 2004, the Center for Built Environment at University of California, Berkeley created (CBE) an
effective and innovative tool for web based IEQ survey (Fig 2.5). It was structured in such a way
14

that it has a core set of questions that focus on the main IEQ factors and a sub-set that focus on the
issues that bother an occupant. All the users are required to rate their comfort levels from the core
set and if there is anything that is of concern, then they would have to click on that particular IEQ
factor and answer more specific questions. With the help of this tool many surveys and evaluations
have been conducted. The web link to the survey was sent to occupants in more than 70 buildings.
Zagreus et al. discuss the applications of the web tool. It was used to analyze how occupants felt
before and after moving to a new office building, to check if the client is satisfied with the design
and to co relate IEQ factors with their occupant comfort. The benefits of this tool are that it saved
time, resources and was not labor intensive. It was also translated into multiple languages to find
its use in other countries (Zagreus, Huizenga and Arens 2004)

Figure 2.5 - The Web based tool developed by CBE (Zagreus, Huizenga and Arens 2004)
15

2.3 Importance of integrating IEQ with occupant survey
(Yousef Al Horr 2016) studied more than 300 research papers on IEQ and productivity and
categorized them based on the year of publication. About half of the papers researched were
published post 2000. The review suggested that apart from the basic IEQ factors, office layout,
location, ambience, and biophilia also had an impact on occupant comfort and work productivity.
Also, the paper identified that occupant satisfaction depends on the interactions between the IEQ
factors.
Integrating Indoor Environmental Quality (IEQ) with occupant survey is also important because it
acts a step towards sustainability. Building occupants have different preferences and it is essential
to cater to their needs not only to ensure their comfort, but also to efficiently operate building
systems. By doing so, energy consumption and demand can be watched.
2.3.1 Absenteeism in offices due to psychological stressors and poor IEQ
(Danielsson, et al. 2014) studied sick leave rates among occupants in different office types in
Sweden. The office sizes were categorized into seven types which ranged from cell-offices, shared
spaces to open-plan (small, medium, and large) office layouts. About 1800 occupants’ responses
were analyzed from the Swedish Longitudinal Occupational Survey of Health (SLOSH) data set.
The number of people along with architectural features such as corridors, access to windows, and
furniture were observed for each office type. Data on functional aspects such as type of work done,
equipment and amenities were also collected. The data was analyzed using the multivariate
regression concept and it was studied as three outcomes: short sick leaves, long sick leaves and
the total number of leaves taken in a year. Some of the results were that men working in flexible
office spaces took more sick leaves than women and women working in large open-plan layouts
took more sick leaves. The study also showed that open plan office arrangements can bring about
16

higher infection rates amongst its occupants. This type also tends to have more environmental
stressors like background noise, less privacy and no access to windows and these factors affect the
rate of absenteeism.
A study by Erika Finell and Jouko Natti shows that psychological stressors have an effect on long-
term sick leaves, that is more than ten days. The authors studied the relation that poor perceived
IEQ and stressors have on employee absence in Finland. The data set consisted of interviews of
more than 16000 people which was registered with the Finnish Quality of Work Life Survey
(FQWLS) and Kela which is an insurance institution. Perceived IEQ was measured in two sets,
the first set asked occupants about dust, poor ventilation, pollutants, and temperature. The second
set was a measure of the factors from the first set. The data sets were analyzed using analysis of
variance (ANOVA) and cross tabulations. The results indicated that most of the people that
participated were women and 62% of the population from the dataset had an issue with at least one
factor from the first dataset. Gender and age did not significantly affect perceived IEQ. Employees
who had less support from supervisors and colleagues took more long-term sick leaves (Finell and
Nätti 2019).
2.3.2 Occupant health issues due to poor IEQ
Oluyemi Toyinbo from the University of Eastern Finland describes that the outdoor environment
plays a role in the IEQ of a naturally ventilated space implying that it can get difficult to filter the
particulates present in the fresh air coming from doors and windows. He says that IEQ depends on
physical, chemical, biological factors, and particle factors. Chemical factors consist of the
pollutants being emitted by carpets, equipment, and other VOCs (volatile organic compounds).
Biological factors include mold and build up that could develop because of dampness and
excessive moisture content and they can have adverse effects on the occupants. The author believes
17

that cleanliness and hygiene also have an impact on IEQ. Regular hand washing can decrease the
spread of diseases, this seems to apply to the current Covid-19 situation as well (Toyinbo 2019)

Occupants can also develop Sick Building Syndrome (SBS) and Building Related Illness due to
poor ventilation inside the building and pollutants from both indoor and outdoor sources. SBS is
the term used when occupants cannot identify the reason behind their illness. It is usually short
term and with minor symptoms. Whereas building related illness have a known cause and its
effects last longer. (EPA 1991)

It is equally important to ensure good IEQ not only in conventional buildings, but also in green
buildings. By implementing passive and low energy concepts, acceptable IEQ conditions can be
achieved in green buildings. To save energy, installing energy efficient windows that do not allow
for heat loss in winter season and heat gain in the summer season. Having a green roof could help
in heat gain and heat loss. Becker and Wang studied a green roof that allowed for 74% heat gain
in the heating season and nearly 25% heat loss in the cooling season (Toyinbo 2019). Although a
naturally ventilated space can save energy and energy costs, it might be difficult to keep occupants
comfortable as it depends on the geography, weather, and air speed of a location. A study showed
that green certified buildings have less indoor air pollutants and therefore better occupant health.
When it came to work productivity, green building occupants performed better and took lesser sick
leaves than people in conventional buildings (Toyinbo 2019).
2.3.3 Impact on work productivity
Wargocki is of the opinion that poor indoor air quality has an impact on occupant productivity and
that it leads to an increase in costs incurred. The costs for an individual could be medical expenses,
18

insurance policies, reduce in salary due to an increase in absenteeism, reduce in salary due to an
increase in sick leaves, reduce in concentration and therefore reduced productivity. The costs for
building owners could be because of an increase in capital investment, raw material like fuel and
coolants that are needed to run building systems, and electricity costs. Building owners might be
required to rent out additional spaces for operation and maintenance. These costs when added up
at the society and national level, they can bring about a decrease in the gross product. When fresh
air supply was doubled it was seen that there was a 10% decrease in building related illnesses and
sick leave applications. Also, a doubled fresh air supply rate brought about an increase in work
productivity by approximately 1.5%. Good work productivity levels can help in offsetting the costs
required to maintain the fresh air supply rate. In the US, it is estimated that a good indoor
environment can reduce costs to individuals, employees and owners and can save up to 168 billion
dollars a year (Wargocki 2011).

(Zdenka Budaiova 2015) conducted a small scale IEQ study to check the levels of indoor pollutants
in two office rooms and a reception area. The office spaces had the usual equipment like printers,
systems, copiers, and furniture. The office spaces were carpeted, and the reception area was tiled.
In total, there were three people in the two office rooms and one person in the reception area. The
IEQ conditions (thermal quality, visual, air and acoustic quality) were measured. The four people
were assigned particular tasks, they were: typing, mathematic calculations and a learning test. The
study took place over three days for 6 hours each. The survey questions asked were about IEQ,
and perceived productivity. The spaces had a slab heating system and were ventilated by both
mechanical systems and operable windows. The measured data were found to indicate that carbon
dioxide and TVOC levels were below recommended levels. Office room 1 had more particulate
19

mass concentrations than room 2 and the reception, this was because the windows in office room
1 were regularly opened whereas the windows in office room 2 were kept closed throughout the
study. It was also observed that the occupants were dissatisfied with high indoor air velocity rates
and it decreased their productivity.
2.3.4 Thermal preference/ visual preference based on human factors
The CBE-developed web-based tool was used to analyze the survey responses from 215 buildings.
Roughly 34000 people took part in the survey and answered demographic questions and subjective
IEQ questions. They were required to use a 7-point scale ranging from -3 to +3 where 0 was the
center value. The survey link was sent every 2 weeks, however the research analyzed only the
summer response. The aim of the study was to specifically understand user satisfaction with two
IEQ factors, namely thermal comfort, and air quality. The survey required users to report
productivity levels by themselves. 80% of the occupants in 23 buildings of the 215 surveyed said
they were satisfied with the thermal comfort and air quality at their workplace (Fig 2.6). Results
showed that people who were near a thermostat or had access to change the temperature were more
satisfied than those who did not. Another finding was that people who could operate their windows
were more satisfied. With respect to air quality, it was identified that people were mostly
dissatisfied because of unclean and stale air and bad odor was being generated by people, carpets,
and food (Huizenga, Abbaszadeh and Zagreus 2006)
20

Figure 2.6 - Thermal comfort satisfaction amongst the 215 buildings surveyed (Huizenga, Abbaszadeh
and Zagreus 2006)
(JoonHo Choi 2010) studied Thermal comfort amongst occupants in 20 office spaces located in a
38-floor office tower. The study was performed for a period of three years to collect data for the
winter and summer seasons. The purpose of the study was to test if gender and age have any
relevance regarding one’s thermal comfort. To perform this POE, indoor environment data had to
be collected in terms of temperature, air velocity and lighting of the space. To understand the
occupants’ preference questionnaires were handed out. To get accurate data about the quantitative
parameters a cart was developed which had a camera and sensors. This cart was positioned at
everyone’s workspace for fifteen minutes and temperature data at three different heights were
recorded. At the same time, the worker is required to answer the 25 questions from the
questionnaire. The questionnaires were collected from 402 people comprising 212 females and
190 males. The data was also grouped into two age categories: above and below the age of 40. The
21

clothing value of the 402 people was observed and the clo value on an average was 0.5 in summer
and 1.0 in winter. ASHRAE standards were considered to plot temperature data with the prescribed
comfort zone. Results showed that in summer 48% of the set point temperatures were below
comfort zone and in winter 15% of the set point temperatures were below comfort zone. Women
were more uncomfortable in the summer than men. The difference was not so significant in winter
and swing seasons. Workers in the first age category i.e., above the age of 40 were more satisfied
with the thermal environment than the second category.

A recent study by USC Marshall school shows that women’s productivity is higher at warmer
temperatures. The study was conducted in collaboration with the Berlin Social Science Center.
The test was conducted in the same center and a total of 543 people took part in it. The test was
conducted in sessions and the temperatures ranged from 61
˚
F to 91˚ F. Every session had three
tasks; the first task involved mathematics where they were asked to add two-digit numbers, the
second was a verbal task, which required the participants to form words with 10 letters and the last
was a cognitive test. The results were that even a small variation in temperature had an impact on
productivity and that women prefer warmer temperatures (Miller 2019)
2.4 Impacts of architectural design parameters and POE
(Shengxian Kang 2017) conducted IEQ surveys in 19 universities in the southern part of China to
specifically study the impact of IEQ in open plan research spaces. The survey was conducted for
6 months, 265 people took part and amongst them 231 were considered for analysis, which makes
up for an 87% response rate. The surveys were analyzed in four parts. First, occupant productivity
was studied and 58.5% of them said that they could be more productive. It followed by the overall
satisfaction with IEQ factors, occupants were most satisfied with the lighting levels. Then, IEQ
22

sub factors like noise sources, furniture, quietness, etc. were assessed. Lastly, demographic factors
like sex, age and type of work performed were taken into consideration to co relate with the first
three parts. The main finding was that acoustics plays the most important role in open plan office
spaces and that noise generated from conversations between people has a negative impact. The
study also revealed that the perceived comfort levels depend on demographic factors; occupants’
birthplace matters as people belonging to different places have different levels of sensitivity to the
IEQ factors.
2.5 Research Objectives
1. To check if women prefer warmer temperatures than men and who were more satisfied
with their job.
2. Many factors affect occupant work productivity and job satisfaction. The second objective
is to identify which factors among the thermal, visual, air, acoustic, and spatial factors are
the most important and to rank them based on significance.
2.6 Summary
The main factors that affect the indoor environment are thermal comfort, air quality, visual quality,
acoustic quality, and spatial quality. With the help of POE, we can identify the cause and effect of
any issue. POE is a broad study through which architects and engineers can formulate better design
guidelines. Issues like absenteeism, Sick Building Syndrome (SBS) can also be rectified.
Architectural features also have an impact on the occupants. To design and operate a building
better, post occupancy evaluations must be included in a building’s life cycle and regular operation
and maintenance schedules. There have been many POE and IEQ studies conducted in an office
environment. However, there are not many diagnostic POEs which have data from over 40
buildings and 1000 plus workstations.
23

CHAPTER 3 METHODOLOGY
3.1 Overview
The Center for Building Performance and Diagnostics at Carnegie Mellon University (CBPD)
collected workstation data from 60 buildings to form a large database. The data was collected from
different types of organizations such as federal offices, private offices, research centers and
commercial offices. Three types of data were collected from 1865 workstations: occupant survey
(COPE), technical building characteristics (TABS) and IEQ measurements at workstations
(NEAT). The area of all offices is less than 5000 ft
2
. The research team also created the large
database to study and co-relate data between all three datasets. This study focuses on analyzing
the data using statistical software: SPSS (by IBM) and Minitab. Figure 3.1 briefly illustrates the
methodology of the study.

Figure 3.1 - Methodology diagram
24

3.2 Technical Attributes of Building Systems (TABS)
The technical attributes of a building can be categorized under thermal quality, air quality, visual
quality, acoustic quality, ergonomic and spatial quality attributes. The purpose of recording the
building characteristics is to understand if any key features or attributes affect occupants’
satisfaction with their respective work environment. The Center for Building Performance and
Diagnostics (CBPD) at Carnegie Melon University recorded (CMU) and collected data by having
expert walkthroughs. The experts were either faculty members or specifically trained graduate
students. Every workstation was attributed a unique space ID and the IEQ stressors (such as the
presence of fans, thermostats, heater, printers, plants) were recorded. For the purpose of this thesis,
the selected building attributes are workstation enclosure (open or close), and location of
workstation (perimeter or interior). The workstations that are located within a distance of 15’ from
the exterior wall are considered to be the workstations at the periphery and the ones located more
than 15’ away are the interior workstations.
3.3 Indoor Environmental Quality measurement (NEAT)
NEAT is the National Environmental Assessment toolkit used to measure the indoor environment.
It was created at the Carnegie Melon University (Fig 3.2). It is basically an instrument cart with
sensors placed at different heights. The cart is about 1.2m in height and records spot measurements
of temperature, air speed, air quality and relative humidity. The cart was placed at the center of the
seat location of every workstation for 15 minutes. To ensure accurate measurements, recordings
are taken once the cart is attuned with the environment. Temperature sensors are placed at different
heights (0.1m, 0.6m, and 1.1m). Simultaneously, hand-held measurements of the lighting levels
and temperatures are recorded. The measurements are then exported to a database for data
classification and analysis.
25

Figure 3.2 - Instrument cart developed at CMU
3.4 Occupant survey (COPE)
Occupants were provided with hard and soft copy versions of the survey questionnaires. The
purpose of the questionnaire is to understand how occupants feel about their workspace and
whether they were satisfied or not. Occupants were required to fill the survey form while the IEQ
measurements were being recorded by the instrument cart at their respective workstation. The
Center for Building Performance (CBPD) and diagnostics at CMU has modified the questionnaire
generated by the National Research Council Canada (NRCC). The original survey is a two-page
document and contains 25 questions regarding indoor environmental performance and occupant
satisfaction. Questions regarding odor, satisfaction with temperature during fall, winter, spring and
26

summer, and cleanliness were added by CBPD. The questionnaires were given out to occupants in
60 buildings (Fig 3.3-3.4).
The questionnaire has 29 questions, out of which 21 questions ask how the occupant feels about
the indoor environment such as overall air quality, temperature, background noise and job
satisfaction. The occupants answered these questions using the seven-point scale ranging from -3
to +3 with 0 being ‘Neutral’, -3 being ‘Very Dissatisfied’ and +3 being ‘Very Satisfied’. The
occupants were also asked specific questions such as their age, gender, education, and job type.

27

Figure 3.3 - Occupant survey (page 1)
28

Figure 3.4 - Occupant survey (page 2)

29

Table 3.1 - Thermal quality and variables measured (IEQ), surveyed (COPE) and observed (TABS)

Table 3.2 - Visual quality and variables measured (IEQ), surveyed (COPE) and observed (TABS)

Thermal
Quality
IEQ data Occupant Survey TABS
Air temperature Q3. Temperature in your work area Window controls
1.1 m 3a. Winter
0.6 m 3b. Spring Perimeter or interior
0.1 m 3c. Summer
3d. Fall
Surface temperature ˚C Q14. Air movement in your work area
Exterior
Interior
Floor

Ceiling

Relative Humidity (%)

Air speed (ft/min)
Visual
Quality
IEQ data Occupant Survey TABS
Illuminance on monitor
with light Q1. Light on desk for paper-based tasks
Q4. Aesthetic appearance of your office
Perimeter or
interior
Illuminance on
keyboard with light
Q6. Level of visual privacy within your
office
Q10. Light for computer work View
Illuminance on work
surface with light Q11. Glare on computer screen
Q12. Glare from light fixtures Open or closed
workstation Glare Q13. Glare from daylight
Q16. Access to a view from workstation
luminance Q18. Overall quality of lighting
30

Table 3.3 - Air Quality and variables measured (IEQ), surveyed (COPE) and observed (TABS)

Table 3.4 - Acoustic Quality and variables measured (IEQ), surveyed (COPE) and observed (TABS)

Table 3.5 - Spatial Quality and variables measured (IEQ), surveyed (COPE) and observed (TABS)

Air Quality
IEQ data Occupant Survey TABS
Carbon dioxide (ppm) Q2. Overall Air quality in your work area Open or closed
Q2a. Odor
CO (ppm) Q4a. Cleanliness
Q14. Air movement in your work area
Acoustic
Quality
IEQ data Occupant Survey TABS
Noise criteria Q9. Amount of background noise Workstation size
dBA
Q19. Frequency of distractions from
others

Q7. Amount of noise from background
conversations
Open or closed
workstation

Q5. Level of acoustic privacy for
conversations in your work area

Spatial
Quality
IEQ data Occupant Survey TABS
Q20. Degree of workspace enclosure in
your work area by walls and screens.

Q8. Personal workstation size to
accommodate visitors and materials
Workstation size

Q15. Your ability to alter physical
conditions

Q17. Distance between you and other
people

Q5. Level of acoustic privacy for
conversations in your work area
31

Table 3.6 - General information of occupants

Tables 3.1 – 3.5 illustrate the selected building attributes, questions related to the specific IEQ
factors and measured data. Table 3.6 illustrates the occupant specific questions from the
questionnaire. Most of the NEAT variables consists of continuous, spot measurements and are
matched with occupant surveys. Some variables such as spatial factors and satisfaction levels
cannot be quantified and hence do not have spot or continuous measurements. The SI unit for the
variables is also given in table 3.7.

Subject related
information
Occupant Survey
Q22. Age (age group)

Q23. Gender

Q24. Job category

Q25. Education level

Q26. My department is a good
place to work

Q27. I am satisfied with my job

32

Table 3.7 - NEAT Variables measured

3.5 Statistical analysis
Data analysis is a stage that provides us with the support which could either be in support of the
hypotheses or against it and sometimes it might not lead us to any conclusions. The field of
statistics has many tools and methods that helps researchers in data analysis (Young 2017).
Population is the term used synonymously with the dataset in hand and a sample is a set of the
population. A sample could be selected randomly in a large dataset to represent the population.
Some occupant responses had missing values and their entire response set was not considered for
IEQ factor Variables measured Unit
Spot
measurements
Continuous
measurement
Survey
Thermal Quality
Temperature ˚C
• • •
Relative Humidity %
• • •
Air Quality
CO2 ppm
• • •
CO ppm
• • •
Light Quality
Illuminance lux
• - •
Glare
- • - •
Visual Quality
Glare
- • - •
Access to a view
- - - •
Ambience
- - - •
Acoustic Quality Noise level dBA
• - •
Spatial Quality
Enclosure,
workstation size,
workstation location
- - - •
Overall satisfaction
Job and satisfaction
with environment
- - - •
33

data analysis. For example, if an occupant chose not to answer some questions from the survey,
the answered questions were also not considered, this was done to maintain the accuracy of the
results. The cleaned dataset has about 1860 workstation data, NEAT and TABS data. To begin
with, cross sectional data analysis and grouping is done. For further analysis, SPSS an interactive
software program for statistical analysis is used. The analyses performed are 2 sample t-test,
analysis of variance (ANOVA), co-relation analysis and regression. Usually, the traditional value
considered to understand data relations for statistical significance is 95% (0.05 p value).

Figure 3.5 - Image of occupant survey data
34

Figure 3.6 - Image of measured IEQ data compiled on Excel

Figure 3.7 - Data distribution and analysis steps
To provide an overview, the data will first be grouped based on human and spatial factors followed
by illustrations of measured IEQ data (air velocity, temperature, relative humidity, carbon di-oxide
35

levels, worksurface lux levels, etc.). The next step is to study occupant responses, each question
from the survey will be studied through histograms and box plot diagrams. The data will then be
analyzed through simple regression, correlation analysis, and a one-way analysis of variance
(ANOVA). These methods have been explained in the following sections. Through these analyses,
the following impacts will be studied,
1. The impacts of age,
2. Impacts of gender,
3. Impacts of workstation location,
4. Impacts of workstation enclosure,
This will be followed by identifying patterns in the large data set used in this study through
decision tree.
3.5.1 Correlation analysis
Co-relation analysis is done to understand the relationship between two or more factors. This
method was found by Sir Francis Galton in 1877 and in 1896, Karl Pearson further developed the
concept and introduced the Pearson coefficient. Karl Pearson’s method is used to identify linear
relations between two data sets (Iuliano 2019).
3.5.1.1 Pearson correlation coefficient
The Pearson coefficient gives us two important details, the strength between two variables and
how they are associated with each other. It has a boundary, usually between -1 to +1. This means
that the coefficient value lies between negative 1 and positive 1 (-1 ≤ p ≥ +1). Say, there are two
variables A and B and the relationship between them, that is the p value is found to be -0.6. This
suggests that the relation between the variables is inversely related. If a value is close to 0 it means
36

that the two variables have a weak relation and if it is close to -1 or +1 it means that they have a
strong relation (Young 2017). The coefficient can be found by the formula in figure 3.8.
Figures 3.9-3.11 graphically represent the negative, positive, and null correlation between two
variables X and Y.

Figure 3.8 - Formula to identify Pearson coefficient (Young 2017)

Figure 3.9 - Negative correlation between two variables (Iuliano 2019)

37

Figure 3.10 - Positive correlation between two variables (Iuliano 2019)

Figure 3.11 - Null correlation between two variables (Iuliano 2019)
3.5.1.2 Spearman’s rank coefficient
The Spearman’s rank coefficient was proposed by Spearman in 1904, it was adopted from his
colleague’s coefficient, which is the Pearson’s correlation coefficient. It is to measure how strong
38

an association is between two variables. It is mostly used when Pearson’s correlation coefficient
leads to an undesirable outcome. While Pearson’s coefficient assumes that the relation between
two variables is linear, this does not. It does not make any assumptions regarding the frequency
distribution of the variables (Kossowski 2011).

Figure 3.12 - The formula to derive Spearman's coefficient (Young 2017)
3.5.2 Simple regression
One of the most important methods in analysis is the regression analysis. It is a tool used to identify
any relationship between a dependent variable and an independent variable. This method needs at
least two variables for identifying relations. Most of the regression analysis methods utilize a
simple regression or linear regression model. In such a model, the dependent variable can be
expressed in a linear combination with respect to the independent variable. There are three reasons
as to why this is the most commonly used model. Firstly, it is a method that is followed and
understood by many researchers. Secondly, many statistical processes have relationships that can
easily be approximated with the help of a linear regression model and lastly, they provide us with
tractable conclusions.
Simple regression methods are primarily used to establish a descriptive relationship between the
dependent and independent variables (Young 2017). Say, establishing a relationship between Body
Mass Index (BMI) of people with their age and gender. The second use is to estimate the value of
coefficient of a theory. These coefficients act in support or against the theory, they give us the
magnitude of the evidence. Other uses of this method are prediction (predicting how one variable
39

would respond with another) and variable screening (identifying the significant variables in a set
of independent variables). While both regression and correlation analyses provide us with the
relation between two or more variables, regression identifies causes and effects of the relationship
and correlation just gives the value of association between the variables (Young 2017).
3.5.3 2 Sample t test
T tests was first performed by William Gosset in 1908 who worked at a Dublin brewery. He
identified ways to study the means of a single group and means of two independent groups and he
named them one-sample T test and two-sample T test (Hess 2017). The two-sample t test is also
called as the independent samples t-test. It is the usual method used to compare means of two
groups (Stehlik-Barry 2017). The test also has certain assumptions, that the data from both samples
is continuous, has a normal distribution and is independent of each of other (Hintze 2007). The
two-sample t test basically is used to check for the null hypothesis which states that the means of
the two samples are equal if not, it proves the alternative hypothesis which states that the means
of two samples are not equal. For example, the null hypothesis for one test could be that the mean
of the incomes of two groups (male and female) is the same and the alternative hypothesis would
be that the mean of the incomes of the two groups is not the same.
3.5.4 ANOVA
A one-way analysis of variance (ANOVA) is a statistical analysis method that is performed to
check for significant results between the means. ANOVA is used when the means of two or more
groups have to be compared. This test has a dependent variable with which two or more
independent variables are studied (Statistics 2018). The procedure followed allows us to test if the
groups or samples belong to one large population or different populations (Minitab 2020). Apart
from the dependent and independent variable, there could sometimes be a third variable called as
40

a covariate and, in that case, the ANCOVA (analysis of covariance) analysis is used (Statistics
2018).
3.5.5 Data Mining
With the advent of technology and an increase in the use of computers, there is a sharp growth in
the information and the data we have. In order to make complete sense of the database it is vital to
consider data mining methods. Data mining is the process of going through or sifting through data
to explore potentially useful unknown patterns and relations. Some researchers are of the opinion
that knowledge discovery in databases (KDD) is the same as data mining, while some others think
that data mining is an integral part of KDD.
Data can be mined in two ways: one is verification-oriented (used to verify data or one’s
hypothesis) and the other is discovery-oriented. The latter uses two methods to look for patterns:
prediction and description methods. The prediction method is used to predict values and to derive
new and unseen patterns (Rokach 2008). Prediction method is also known as supervised learning
which aims at finding relations between independent and dependent variables. Description method
focuses on data operation.
The verification method involves the traditional statistical ideas like t-tests and analysis of variance
(ANOVA). This does not involve a lot of data mining as it is concerned mainly with testing a
known hypothesis, whereas descriptive methods select a hypothesis in search of patterns (Rokach
2008).
Further, the Prediction method has two models: classification and regression. Fig 3.13 shows the
Data mining taxonomy flowchart. It also shows some of the data classification techniques and
one of the classification techniques is decision tree which is used in this study.
41

Figure 3.13 - Data mining types and methods (Rokach 2008)
3.5.6 Decision tree
Decision trees are classification flow-charts. They have nodes, branches, and leaves which denote
something. A node stands for an attribute that has been tested, the outcome of the test is represented
by the branch and the leaves represent a class label (Bhaskar N. Patel 2012) (Kumar 2013).
This model provides us with the observations of an item and results about its target value (Kumar
2013). It uses an algorithm which aims at building a tree with homogeneous leaves, it tries to
organize the leaves to make it as homogeneous as possible. The end product is easy to analyze and
does not require any specific knowledge.
3.5.7 Statistical software
SPSS is a statistical software developed by International Business Machines (IBM). It has a wide
range of features. The software is used to understand relations between variables, groups, and
42

behavior. Hypotheses can be tested, and assumptions can be validated. It includes a variety of data
analytics tools such as regression analysis, decision trees and ANOVA. Predictive, prescriptive,
and machine learning methods can be used to get the desired outcome. The software also has
procedures for descriptive statistics, numerical prediction, and group identification. The latest
version is IBM 27, and it is used in this study to analyze occupant surveys, IEQ and building
attributes in relation with one another (IBM n.d.). Minitab is also a statistical software that aids in
data analysis and data visualization and is quite similar to SPSS. It helps in identifying patterns,
discovering, and predicting results in datasets.
3.6 Summary
This chapter explained the terms NEAT, TABS and Occupant Survey. The dataset is compiled on
Excel for statistical analysis. Concepts like t – tests, ANOVA, simple regression, correlation
analysis and decision tree classification technique are explained. The p value considered for
statistical significance is 95% or 0.05. Closer the value to 0, stronger the relation between
variables. Decision tree is an advanced data mining model which will be used to find patterns in
large data sets. SPSS, and Minitab will be used for data analysis. Chapter 4 gives an insight on the
data collected.

43

CHAPTER 4
4.1 Data Overview
Chapter 4 elaborates and illustrates the specifics of the data collected, including information about
the occupants, occupant responses and measured data. The survey responses and measured data
are collected from over 1800 workstations in 60 office buildings in the US. Occupant responses
are studied categorically, i.e., by age, gender, occupation, and self-reported productivity. Data
analysis was done using statistical and data visualization software. The excel database was
converted to a .csv format before exporting it to SPSS. In SPSS, variables like gender and age
were assigned values, for example, females were given a value of 1 and males were given a value
of 2 for running statistical tests. Minitab and Visual Paradigm were also used to generate results.
4.1.1 Based on Human factors- age and gender
Table 4.1 - Distribution of data based on human factors

About 1865 people responded to the survey and 1631 occupants disclosed their personal
information such as age, gender, occupation, and job-type. Occupants were required to choose
their age group from 6 groups (18-29, 30-39, 40-49, 50-59, 60-69, and 70+). The age categories
were further grouped as junior (18-29 years), middle-age (30-49 years), and senior (50-70+ years).
Age Female Male Grand Total
Junior 218 142 360
18-29 218 142 360
Mid-age 393 380 773
30-39 226 192 418
40-49 167 188 355
Senior 264 234 498
50-59 189 167 356
60-69 69 54 123
70+ 6 13 19
Grand Total 875 756 1631
44

As illustrated in table 4.1, among 1631 occupants, there were 875 women and 756 men. A total of
360 subjects were in the junior age group, 773 in the middle-age group, and 498 in the senior age
group. Figure 4.1 depicts the same data in percentages.

Figure 4.1 - Distribution of participants by age and gender
4.1.2 Based on Spatial factors – location and enclosure of workstations.
Data of the workstation features such as location (interior or perimeter) and degree of enclosure
(open or closed) were collected. Out of all the workstations, spatial features of 492 workstations
were identified by the CMU team. The IEQ measurements (RH, CO2, lux levels, temperature)
were recorded while the occupant took the survey. There were 389 (21%) open workstations which
means that they did not have any form of enclosure, and 103 (79%) workstations were enclosed
on either three or four sides. Out of the 492 workstations, 262 (53%) workstations were located in
the perimeter zone which is within 15’ from the external walls and 231 (47%) workstations were
located in the interior zone which is more than 15’ away from the external walls.
22%
30%
48%
54% 46 %
45

Figure 4.2 – Number of open and closed workstations

Figure 4.3 – Number of Perimeter and Interior workstations
It is essential to study the dataset in groups and as a whole. By studying the whole dataset, we get
an overview of the existing conditions of the buildings and occupant satisfaction range. Studying
the data in groups such as age or gender will help in arriving at accurate results and in identifying
46

the needs of the different user groups in the building. The following sections illustrate mean and
recommended values of the measured IEQ data and comparative studies of different occupant
groups.
4.2 Measured IEQ data
The IEQ variables measured at the workstations were temperature, carbon dioxide, relative
humidity, reading zone illuminance, worksurface illuminance and screen illuminance. The
temperatures at the workstations were measured at three different heights above the floor, i.e.,
0.1m, 0.6m and at 1.1m. Since the occupants at office workstation configurations are mostly seated
while working, temperature at 1.1m was considered as it is the average height of a seated person
from top of the head to floor level. If a post occupancy evaluation is to be done in an environment
where occupants are mostly standing, say a dance studio, then the temperature must be measured
at 1.6m. Distribution of temperatures in the workstation at all three heights are illustrated in figures
4.4 – 4.6. In figure 4.4, distribution of temperature at 1.1 m is indicated. The data ranges between
19 and 28 degree Celsius with a mean value of 23.5˚C and a median value of 1.7˚C. The ASHRAE
55 standards recommend that to ensure thermal comfort the temperature should range between
19.4˚C to 27.7˚C. The entire data falls under the recommended range of temperature and in figure
4.5, which shows the distribution of temperature at 0.6 m, except for a couple of values, all of the
data is under the recommended range. Figure 4.6 is of temperature distribution at 0.1m and in this
case, there are certain values (about 5%) that do not comply with the recommended range.
47

Figure 4.4 – Distribution of temperature in degree Celsius at 1.1 m

Figure 4.5 – Distribution of temperature in degree Celsius at 0.6 m
48

Figure 4.6 – Distribution of temperature in degree C at 0.1m

Figure 4.7 – Distribution of Carbon-dioxide levels
49

The CO2 levels measured at the workstations are indicated in figure 4.7. Most of the data lies in
the range of 200-1200 ppm (parts per million) and has a mean value of 658.4 ppm and a standard
deviation of 210. According to the Occupational Safety and Health Administration (OSHA), the
permissible value that one can be exposed to must be lower than 5000 ppm over a timeframe of a
week with an 8-hour workday (Mallinger 1996). Whereas until 1999, ASHRAE 62.1 suggested
that the indoor CO2 levels must be under 1000 ppm. However, some researchers still consider that
the CO2 levels must be at most 700 ppm above the outdoor CO2 levels and the maximum level to
be 1000 ppm (Persily 2020). Considering a maximum value as 1000 ppm, it is calculated that about
8% of the data has values over 1000ppm.

Figure 4.8 – Distribution of relative humidity
Relative humidity was also measured at the workstations with the help of the instrument cart. The
distribution of the values is illustrated in figure 4.8. The data is spread between 15% - 80% with a
mean value of 37%. ASHRAE 55 recommends that the relative humidity must be lower than 65%
50

to stay comfortable. Apart from a few outliers, most of the data agrees with the ASHRAE 55
recommended values. The Environmental Protection Agency (EPA) recommends that the relative
humidity levels must be between 30% and 60% to avoid mold growth (Prevention 2015).

Figure 4.9 – Distribution of screen illuminance
A frequency chart of the screen illuminance data is illustrated in figure 4.9. It has a mean value of
334.9 with a standard deviation of 250.8. Most of the data is spread between 0 and 500 lux.
According to the IES (Illuminating Engineering Society) lighting handbook the average lux levels
for a normal workstation in an open or closed office space must be 500 lux (Administration 2019).
The American National Standard Practice for Office Lighting (ANSI/IES RP-1-12), suggests that
the lux levels should fall in between 50-100 lux for screen illuminance (Moon 2016). About 7%
of the data agrees with the recommended range and the remaining 93% of the data lies beyond the
recommended range. The chart also suggests that there are a few outliers, this could be due to
excessive glare conditions.
51

Figure 4.10 – Distribution of worksurface illuminance
Figure 4.10 illustrates the distribution of worksurface illuminance. Most of the data lies in between
50 and 950 lux, the mean value is 529.6 with a standard deviation of 330. According to ANSI/IES
RP-1-12 from the American National Standard Practice for Office Lighting, the recommended
worksurface illuminance level must be between 200 and 500 lux (Moon 2016). 43% of the data
falls in the recommended range, 9% lies below and 48% of the data has lux levels over 500.
4.3 Occupant responses
The occupant survey has questions that correspond to all the IEQ factors. The building users could
answer to the survey with the help of a satisfaction scale ranging from -3 (very dissatisfied) to +3
(very satisfied) with 0 being the neutral value. However, in the excel database, these responses
were converted to a scale of 1 to 7 for analysis purposes. Simple bar charts have been generated to
identify the number of occupants and the distribution of their answers. In addition to this, interval
plot diagrams illustrate the mean and median values, these can be found in the appendix section at
52

the end. To get an understanding of all the responses, mean values of responses to all questions
have been used to produce a rose chart (figure 4.11). It essentially compares the responses of all
questions. There are seven circles that imitate the linear satisfaction scale of 1 to 7, and they are
divided into sectors based on the number of questions. The blue triangles indicate the mean value,
and the red circle indicates the overall environmental satisfaction. It is evident that occupants were
both satisfied and dissatisfied in some areas, they were least satisfied due to the noise that was
generated from people’s conversations, less acoustic privacy to have conversations amongst their
colleagues, air movement and temperature in their workstation. An important observation is that
occupants were satisfied with the visual quality in their workstation. They responded positively to
the lighting quality related questions: light for paper-based work, light for computer based work
and overall lighting quality.

Figure 4.11 – Rose chart of survey questions
53

Figure 4.12 – Distribution of survey responses
Figure 4.12 illustrates another way of looking at the overall survey response, it indicates the
percentage of people that were very dissatisfied to very satisfied. Questions 5 (level of acoustic
privacy to have conversations in your work area), question 16 (access to outside view), question 7
(amount of noise from other people’s conversations at your workstation), and question 19
(frequency of distraction from other people) have a higher percentage of people who were very
dissatisfied. Whereas questions 1 (light on the desk for paper-based activity), 10 (light for
computer-based work), 8 (size of workstation), 18 (overall lighting quality), and question 2
(overall air quality) have a higher percentage of people who were very satisfied.
4.3.1 Comparison by age
In this and the following sections, data will be presented with regard to different groups such as
age, gender, occupation, and self-reported productivity. Rose charts have been used to represent
the age groups. Figure 4.13 represents the survey responses based on the 6 age groups and figure
4.14 represents survey responses based on 3 age groups (junior, mid-age, and senior). For question
6 (level of visual privacy in your workstation), all the age groups responded similarly except for
54

the age group of 70 + and the same group was least satisfied with the access to outside view.
Acoustic quality is something that all age groups have an issue with. Temperature at the
workstations is another factor that received poor response compared to other questions. In figure
4.14 the three age groups reacted similarly to the following questions: satisfaction with the overall
lighting quality, workspace, visual privacy in your workstation, disturbance due to background
noise, job satisfaction and distance between co-workers.

Figure 4.13 – Responses distributed by six age groups
55

Figure 4.14 – Rose chart of responses distributed by three age groups

Figure 4.15 - Interval plot of question 1 to question 7 based on gender
56

Figure 4.16 – Interval plot of question 8 to question 17 based on gender

Figure 4.17 – Interval plot question 18 to question 29 based on gender
57

Interval plot charts of the three age groups are illustrated from figure 4.15 – figure 4.17, they
essentially indicate the role age factor plays in occupant satisfaction. Out of the 29 questions in
the survey, 21 questions require the occupants to rank their satisfaction levels, i.e., from 1-7. There
are slight variations in responses, the mid-age group were comparatively less satisfied than the
other two groups in twelve cases indicated by circles around the plots. The junior age group was
less satisfied in five cases, and in three other cases the senior group was less satisfied. Table 4.2
indicates statistically significant value or p - value and variations in mean values based on age.
The table shows that there are significant variations between age groups for 12 questions out of
the 21 scale-based questions. Out of the three age groups, the middle-age group was the least
satisfied, their mean values were the lowest in ten questions. Whereas the junior age group was
the most satisfied.

58

Table 4.2 – Variation results based on age.
Age
Average value Junior Mid-age Senior P value
Q1 – Visual quality 5.25 5.32 5.34 0.683
Q2 – Air quality 5.08 4.76 4.86 0.010*
Q3 – Thermal quality 4.25 3.88 4 0.004*
Q4 – Spatial quality 4.76 4.61 4.86 0.020*
Q5 – Acoustic quality 3.75 3.39 3.55 0.013*
Q6 – Visual quality 4.03 3.93 4.02 0.648
Q7 – Acoustic quality 3.73 3.45 3.41 0.021*
Q8 – Spatial quality 5.06 4.83 4.81 0.077
Q9 – Acoustic quality 4.52 4.44 4.51 0.692
Q10 - Visual quality 5.08 5.19 5.23 0.381
Q14- Air quality 4.59 4.08 4.38 0.000*
Q15 - Spatial quality 4.02 3.74 3.94 0.012*
Q16 - Visual quality 4.41 4.57 4.78 0.046*
Q17 - Spatial quality 4.72 4.75 4.86 0.439
Q18 - Visual quality 5.02 5.05 5.17 0.276
Q19 - Acoustic quality 4.14 3.83 3.94 0.023*
Q20 - Spatial quality 4.33 4.23 4.51 0.025*
Q26 - Satisfaction with workplace 5.56 5.63 5.56 0.688
Q27 - Job satisfaction 5.48 5.65 5.54 0.167
Q28 - Self-reported productivity 4.89 4.55 4.79 0.001*
Q29 - Satisfaction with IEQ 4.9 4.58 4.71 0.009*
59

4.3.2 Comparison by gender

Figure 4.18 – Rose chart of survey responses based in gender
Figure 4.18 compares the survey responses by gender. The means and satisfaction levels of both
women and men were almost the same in most cases. Women were slightly less satisfied than men
with temperature, noise generated from people’s conversations, air movement and with the overall
air quality at their workstations. Since the dataset is large, confidence interval plots will provide a
clearer understanding (figure 4.19-4.21). From the interval charts, it is observed that both the
groups have very similar responses to question 28: the environmental conditions in my work area
support my personal work productivity and question 29: I am satisfied with the indoor environment
in my work area. The interval plot of question 3: temperature in your work area, depicts a wider
gap in results. ANOVA tests in chapter 5 will be used to identify whether women prefer warmer
temperatures than men.
60

Figure 4.19 – Interval plot of question 1 to question 7 based on gender

Figure 4.20 - Interval plot of question 8 to question 17 based on gender
61

Figure 4.21 - Interval plot of question 18 to question 29 based on gender
Table 4.3 indicates that there are significant statistical variations between men and women with
four questions and in all four cases women were less satisfied than men. It is also noted that
question three (satisfaction with the temperature at your workstation) has the lowest mean values
among the four.

62

Table 4.3 - Variation results based on gender
Gender
Average value Female Male P value
Q1 - Visual quality 5.23 5.42 0.013*
Q2 - Air quality 4.70 5.06 0.000*
Q3 - Thermal quality 3.80 4.26 0.000*
Q4 - Spatial quality 4.78 4.67 0.15
Q5 - Acoustic quality 3.54 3.50 0.62
Q6 - Visual quality 4.05 3.92 0.18
Q7 - Acoustic quality 3.48 3.52 0.69
Q8 - Spatial quality 4.91 4.84 0.41
Q9 - Acoustic quality 4.53 4.44 0.29
Q10 - Visual quality 5.15 5.23 0.26
Q14- Air quality 4.16 4.45 0.000*
Q15 - Spatial quality 3.85 3.89 0.62
Q16 - Visual quality 4.52 4.70 0.11
Q17 - Spatial quality 4.83 4.74 0.27
Q18 - Visual quality 5.08 5.11 0.73
Q19 - Acoustic quality 3.97 3.89 0.37
Q20 - Spatial quality 4.41 4.26 0.10
Q26 - Satisfaction with workplace 5.65 5.54 0.13
Q27 - Job satisfaction 5.65 5.51 0.05
Q28 - Self-reported productivity 4.70 4.72 0.79
Q29 - Satisfaction with IEQ 4.67 4.72 0.59

63

4.3.3 Comparison by occupation

Figure 4.22 – Frequency chart based on occupation
There were more occupants who had a professional background constituting 40% of the workforce,
followed by managerial, administrative, technical and doctorate positions at 24%, 16%, 15% and
3% respectively. Figures 4.23 – 4.25 illustrate interval plots of the 21 scale-based questions. The
survey gave the occupants the option to choose their occupation from five options, that is
administrative, doctorate, managerial, professional, and technical. The group holding a doctorate
degree showed higher satisfaction levels than the other four groups and was followed by the
managerial group. This suggests that people at a higher post probably have more access to their
thermostat and the ability to change or modify their workstation settings. Table 4.4 indicates that
0 500 1000 1500 2000
Administrative
Doctorate
Managerial
Professional
Technical
Total
Occupation frequency chart
64

out of the 21 questions, 13 questions show statistically significant variations. All visual and
acoustic quality questions have a p value of less than 0.005. Although the different occupation-
based groups had variations in their responses to IEQ related questions, they had similar job
satisfaction and self-reported productivity levels.

Figure 4.23 – Interval plot of question 1 to question 7 based on occupation
65

Figure 4.24 - Interval plot of question 8 to question 17 based on occupation

Figure 4.25 - Interval plot of question 18 to question 29 based on occupation

66

Table 4.4 - Variation results based on occupation
Occupation
Average value Administrative Doctorate Managerial Professional Technical
P
value
Q1 - Visual quality 5.25 6.06 5.47 5.23 5.15 0.001*
Q2 - Air quality 4.48 5.14 4.93 4.97 4.77 0.001*
Q3 - Thermal quality 3.81 4.21 3.93 4.08 4.03 0.211
Q4 - Spatial quality 4.87 5.11 4.56 4.75 4.76 0.72
Q5 - Acoustic quality 3.51 4.34 3.82 3.27 3.45 0.000*
Q6 - Visual quality 3.89 4.64 4.21 3.85 3.96 0.007*
Q7 - Acoustic quality 3.54 4.40 3.70 3.31 3.38 0.000*
Q8 - Spatial quality 4.94 5.68 5.07 4.68 4.87 0.000*
Q9 - Acoustic quality 4.44 4.87 4.75 4.43 4.25 0.020*
Q10 - Visual quality 5.02 6.04 5.25 5.17 4.99 0.000*
Q14- Air quality 4.02 4.57 4.30 4.33 4.33 0.060*
Q15 - Spatial quality 3.89 4.59 3.82 3.79 3.90 0.036*
Q16 - Visual quality 4.14 5.38 4.71 4.75 4.45 0.000*
Q17 - Spatial quality 4.82 5.38 5.01 4.61 4.62 0.000*
Q18 - Visual quality 4.86 5.87 5.25 5.04 4.96 0.000*
Q19 - Acoustic quality 3.94 4.57 4.12 3.78 3.81 0.003*
Q20 - Spatial quality 4.31 4.85 4.61 4.18 4.18 0.001*
Q26 - Satisfaction with
workplace 5.50 5.53 5.60 5.60 5.64 0.872
Q27 - Job satisfaction 5.58 5.72 5.56 5.54 5.63 0.889
Q28 - Self-reported
productivity 4.76 5.00 4.71 4.61 4.72 0.449
Q29 - Satisfaction with
IEQ 4.71 4.85 4.77 4.59 4.71 0.453
67

4.3.4 Self-reported productivity

Figure 4.26 - Interval plot of question 28 based on age group

Figure 4.27 - Interval plot of question 28 based on junior and senior age group
68

Figures 4.26 and 4.27 are interval plot charts of age groups, gender, and the categorical variable
for grouping: self-reported productivity. It indicates that women who belonged to the junior age
category (18-29 years) were more dissatisfied than the men belonging to the same age group. With
the mid-age group there is not much of a variation, middle-aged men were slightly less dissatisfied
with the overall indoor environment and how that impacted their personal productivity. Whereas
men in the senior group were more satisfied with the indoor environment and were more likely to
agree that it is supporting their work productivity.
Figure 4.28 is of an interval plot of workstations located in the perimeter or interior zone and
closed and open workstations. The categorical factor considered for grouping is environment on
productivity. It depicts that the number of closed interior workstations is lesser than the number of
open interior workstations and occupants using the former workstations reported higher
productivity levels than the occupants using the latter workstations.

Figure 4.28 - Interval plot of question 28 based on workstation location and enclosure
69

Figure 4.29 - Interval plot of question 28 based on survey responses

Figure 4.30 - Interval plot of question 28 based on survey responses
70

Figure 4.31 - Interval plot of question 28 based on survey responses
Figures 4.29 - 4.31 are interval plot charts which have question 28 (the environment supports my
personal productivity) as the categorical variable for grouping and the scale-based questions as
graph variables. Responses to all questions in the three plots indicate that positive responses had a
positive impact on self-reported productivity.
The data was measured at different times of the year, figure 4.32 shows how self-reported
productivity levels varied with seasons. H indicates hot or Summer season, C indicates cold or
Winter season, and S stands for Swing season which represents Fall and Spring. The interval plot
indicates that self-reported productivity was mostly consistent throughout the seasons with no
significant changes. Further analyses were done by creating three separate datasets based on season
alone and then the effects on productivity were studied. Figure 4.33 indicates three interval plots
of work productivity based on the three seasons (Summer, Winter, and Swing) and noticeable
variations are observed. In Summer different occupant groups’ mean productivity levels ranged
71

between 4 and 5, in Winter it ranged between 3.5 and 4.5 and in the Swing season it ranged between
5 and slightly over 6 making it the most preferred or productive season.

Figure 4.32 – Interval plot of question 28 based on seasons
72

Figure 4.33 – Productivity levels based on seasons and grouped by age and gender
4.4 Summary
Chapter four discusses the number of occupants by age, gender, and occupation. The spatial
aspects of the workstation are also discussed. The distribution of measured IEQ was also plotted
as histograms to understand the mean and recommended values set by standards such as ASHRAE
and OSHA. Except for screen and worksurface illuminance levels, it is noted that the measured
data of temperatures at three heights, carbon-dioxide, and relative humidity levels have most of
the data that complies with the standard recommended values. Although most of the illuminance
data did not fall under the recommended range, the light levels were on the higher range and
reached a maximum of 1900 lux and occupants were very satisfied with the lighting quality at their
workstations. This is followed by the analyses of occupant responses as a whole and by category,
overall responses show that occupants rated higher satisfaction levels for lighting quality,
73

workspace size, and aesthetics of their workstation. The responses were then studied based on age
groups, gender, and occupation. The last section discusses self-reported productivity and its
variations played by the gender, age, workstation location, workstation enclosure and season.

74

CHAPTER 5
5.1 Statistical analysis
The previous chapter is an exploration of the whole dataset and it included some comparative
results. Chapter 5 begins by introducing the results of forward regression and discusses results that
considered the measured data at workstations such as relative humidity, carbon-dioxide, lux levels
at worksurfaces, and temperature at 1.1m. It is done so by using the two-sample-t test and ANOVA.
Further, the IEQ factors (air, spatial, thermal, visual, and acoustic quality) are ranked based on
significance for different occupant group. The last section illustrates results that used machine
learning to identify significant patterns in the dataset.
5.1.1 Impacts of IEQ factors on work productivity
Figure 5.1 is a result of running a forward regression analysis on only the survey responses and by
having Q28 (the environment supports my personal productivity) as the dependent variable.
Regression analyses must have one dependent variable and can have multiple independent
variables. The result is a hierarchy of factors based on statistical significance. It is used to generate
a set of predictors. The important column to look at is the R square change column that has been
highlighted with a red dashed rectangle. Air movement is the most important predictor with a R
square change value of 0.345 followed by three spatial quality predictors (satisfaction with
workspace, its size, and aesthetics). Thereafter, the R square changes are not quite significant.
Figure 5.2 is another result of forward regression, in this case there are additional factors such as
measured data and human factor (age, gender) added to the list of independent variables. The result
has a smaller subset of predictors than the previous figure. The first two predictors are the same
and have similar R square change values. Satisfaction with lighting quality forms the third
significant predictor followed by air quality.
75

Figure 5.1 - Forward regression

76

Figure 5.2 - Forward regression
5.2 Analysis by different occupant group
Occupant responses are studied in groups; age, gender, workstation location and workstation
enclosure to understand the preferences of building users and to explore the variations in responses
between the sub-categories in each group (for example age: junior, mid-age, and senior).
5.2.1 Gender
Figure 5.3 has three sets of interval plots which are results of two-sample-t tests. The first one is
plotted with respect to recorded carbon-dioxide values and satisfaction with air quality. It depicts
that a positive feedback was received when the CO2 levels crossed 675ppm. The plot changes
when the gender of occupants was added, women had better satisfaction levels at higher CO2 levels
77

(about 700ppm) and it was the opposite with men, they were satisfied with relatively lower CO2
levels (650ppm on an average). The second set shows satisfaction levels with respect to relative
humidity, the first plot does not imply any significant differences in responses. Women and men
had very little difference in preferences. Women were dissatisfied at about 40% relative humidity
and responded more positively at 37%, whereas men reacted negatively at 35.5% and positively at
around 37.5% relative humidity. The third set of charts is regarding the lighting levels, both charts
indicate that occupants were satisfied at higher lux levels (525 lux and above).

Figure 5.3 - Confidence interval plot based on gender
78

The data was collected through-out the year and the season it was collected/ measured in was also
noted. To understand how people responded to various temperatures in different seasons, subsets
of the data was created to study the responses for Winter, Summer and Swing (Fall and Spring).
Figure 5.4 is a set of interval plots of temperature at 1.1m and survey responses to Q3 (satisfaction
with the temperature at your workstation). Both men and women reacted similarly for temperature
changes in Winter and Summer, whereas during the Swing season their reactions varied slightly.
On average women and men were satisfied with a temperature above 23.5 degree Celsius in
Summer and Winter. Men preferred cooler temperatures than women in the Swing season, but it
is not of significant difference.

Figure 5.4 - Confidence interval plots of temperature at 1.1m and gender
79

5.2.2 Age
The second set of analyses are based on age groups (figure 5.5). The interval plots of worksurface
lux and satisfaction with the light levels for paper-based work indicate that brighter light levels are
required for senior occupants to perform their regular office tasks. The results were quite opposite
with respect to carbon-dioxide levels, the junior age group was satisfied with higher levels (above
700 ppm) of carbon-dioxide. The same pattern was seen in the third plot; relative humidity and
satisfaction with thermal quality, senior occupants were the most satisfied at lower levels (36%)
of relative humidity and junior occupants were satisfied with slightly higher relative humidity
levels (38%).
80

Figure 5.5 - Confidence interval plot based on age groups.
Figure 5.6 is a of how occupants reacted to temperatures in different seasons, the first plot does
not consider the factor of season and a strange finding to be noted here is that the senior age group
was the only age group that was comfortable at the lowest average temperature of this plot which
is 23 degree Celsius. The second plot is of the satisfaction levels during the Winter season, all
groups were comfortable at 24 degree Celsius and although there is a difference of less than one
degree, it suggests that occupants were dissatisfied at temperatures slightly above 23 degree C. In
the Summer season, the mid-age and senior group were comfortable at 24 degree C and the junior
group was satisfied at a lower temperature (in this case it is 23 degree C). In the Swing season, all
81

groups rated their satisfaction levels as positive around 23 degree C. Overall, the comfortable range
seems to be between 22.75 and 24 degree C.

Figure 5.6 - Confidence interval plots of temperature at 1.1m and age groups
5.2.3 Workstation enclosure
Apart from studying the responses based on human factors, they were also categorized and studied
based on spatial factors which is location and enclosure of the workstations. Figure 5.7 is a set of
a comparative result between open and closed workstations. With regard to temperature, occupants
in enclosed workstations preferred slightly lower temperatures (23.25 C) and there were no
significant variations with the negative to positive scale for open workstations. With respect to
CO2 levels, occupants located in closed workstations that had CO2 levels of 675 ppm were more
satisfied. It is interesting to note that the average CO2 levels at open workstations was 650 ppm
and it was the average value for both the negative and positive set of values. Closed workstations
82

have a wider range of lighting levels (450 lux to 850 lux) and occupants were satisfied at higher
light levels. Open workstations have lower lighting levels and there is no significant variation
between negative and positive responses.
Figure 5.8 is a set that compares three questions from the survey with the enclosure of the
workstations. The questions are Q28 (the environment supports personal productivity), Q27
(satisfied with the job), and Q29 (satisfied with the indoor environment). Although all three plots
indicate that occupants working in closed workstations were more satisfied than occupants
working in open workstations, the difference in the average values is not large enough to deduce
that workstation enclosure matters.

Figure 5.7 - Confidence interval plots based on workstation enclosure
83

Figure 5.8 - Satisfaction levels of three survey questions with respect to enclosure of workstations
5.2.4 Workstation location
Figure 5.9 is a comparative set of perimeter and interior workstations. The first plot shows that the
average temperature at perimeter workstations was higher than the temperature recorded at interior
workstations. There is a very small difference in the negative to positive gradient; occupants were
satisfied at the respective lower temperatures in both locations. CO2 levels were higher in the
perimeter zone and lower in the interior zone, this is because of the window and door openings in
the perimeter zone that allows for an influx of natural ventilation. Occupants situated in the
perimeter zone were satisfied with higher levels of CO2 (690 ppm) and the occupants working in
the interior zone were satisfied with 620 ppm of CO2. Here again the difference in the negative to
positive gradient (about 675-690 ppm for perimeter workstations and about 620-630 ppm for
interior workstations) is not very large. Significant variations are seen in the next plot; worksurface
84

lux, the average lux levels ranged between 400 and 450 lux in the interior spaces. Whereas the
average lux levels ranged between 550-700 lux for perimeter workstations which is significantly
higher, and users were very satisfied at these levels. The relative humidity plot indicates that on
an average, the interior workstations were slightly more humid than the perimeter ones.

Figure 5.9 - Confidence interval plots based on location of workstations

85

Figure 5.10 - Comparison of location with three survey questions
Figure 5.10 is a set of three plots that compares self-reported productivity, satisfaction with the
indoor environment and job satisfaction with the location of the workstations. Occupants located
in the perimeter zone were more productive and were more satisfied with the indoor environment
and their job than the occupants located in the interior zone. Although all three plots indicate that
occupants working in perimeter workstations were more satisfied than occupants working in
interior workstations, the difference in the average values is not large enough to deduce that
workstation location matters.

86

5.3 IEQ Ranking order
One of the research objectives was to rank the IEQ factors based on significance. The bar chart
below (figure 5.11) indicates the ranking order preferred by junior, mid-age and senior age groups
in a decreasing order of satisfaction, i.e., most satisfied to least satisfied. These charts were created
by identifying the average values of occupant responses to specific questions from the survey such
as satisfaction with air quality, and satisfaction with lighting quality. The ranking order was the
same for mid-age and senior age group, but with different average values. These two groups were
most dissatisfied with the acoustic quality followed by spatial quality. The junior age group ranked
visual and thermal quality as third and fourth respectively whereas these qualities were swapped
for mid and senior age group. Air quality was a factor that all groups were most satisfied with.

Figure 5.11 - IEQ factors ranked according to age group
JUNIOR MID-AGE SENIOR
AIR QUALITY 4.59 4.08 4.38
THERMAL QUALITY 4.25 3.88 4
VISUAL QUALITY 4.03 3.93 4.02
SPATIAL QUALITY 4 3.74 3.94
ACOUSTIC QUALITY 3.73 3.39 3.41
4.59
4.08
4.38
4.25
3.88
4
4.03
3.93
4.02
4
3.74
3.94
3.73
3.39
3.41
AIR QUALITY THERMAL QUALITY VISUAL QUALITY SPATIAL QUALITY ACOUSTIC QUALITY
87

Figure 5.12 indicates the ranking order for women and men. Occupants were most dissatisfied
with the acoustic quality and least dissatisfied with air quality. For women, thermal quality was
the second most factor that they were dissatisfied with followed by spatial, visual and air quality.
The order is quite different for men, the second factor that they were most dissatisfied with was
spatial quality, followed by visual, thermal and air quality. One common observation amongst all
five groups is that acoustic quality is the first factor and air quality is the last factor.

Figure 5.12 - IEQ factors ranked according to junior gender
5.3 Advanced analysis using Machine Learning
WEKA (Waikato Environment for Knowledge Analysis) is a data mining software; it was
developed in New Zealand at the University of Waikato (Waikato n.d.). J48 are decision trees that
help in identifying target variables or impact factors from a given list of independent variables and
a dependent variable. The following trees are basically two sets of trees, each having a different
dependent variable; the first set has Q28 “Environment supports productivity” as a dependent
FEMALE MALE
AIR QUALITY 4.16 4.45
THERMAL QUALITY 3.8 4.26
VISUAL QUALITY 4.05 3.92
SPATIAL QUALITY 3.85 3.89
ACOUSTIC QUALITY 3.48 3.5
4.16
4.45
3.8
4.26
4.05
3.92
3.85
3.89
3.48
3.5
AIR QUALITY THERMAL QUALITY VISUAL QUALITY
SPATIAL QUALITY ACOUSTIC QUALITY
88

variable and the second set has Q 29 “Satisfaction with IEQ” as the dependent variable. The list of
input variables was changed for every tree in both the sets, this was done to study how different
factors affected the output. Table 5.1 below illustrates the same.
Table 5.1 - Independent variables considered for decision trees
No Dependent variable Survey Human
factor
Spatial
factor
Survey +
Human and
Spatial factor
Entire
dataset
Independent variables
1a Q28. Environment supports
productivity
●
1b Q28. Environment supports productivity ● ●
1c Q28. Environment supports productivity ● ● ●
1d Q28. Environment supports productivity ● ● ● ●
1e Q28. Environment supports productivity ● ● ● ● ●
2a Q29. Satisfaction with IEQ ●
2b Q29. Satisfaction with IEQ ● ●
2c Q29. Satisfaction with IEQ ● ● ●
2d Q29. Satisfaction with IEQ ● ● ● ●
2e Q29. Satisfaction with IEQ ● ● ● ● ●

5.3.1 Self-reported productivity
The trees have fruits (i.e., outcomes) with the number of datasets classified into a specific response
(i.e., positive, or negative work productivity. For example, in figure 65, the value 1132/101
indicates the number of results that ended with “positive satisfaction with work productivity” and
101 is the number of results that do not meet this classification pathway. The tree results below
have certain positive values marked in red rectangles to indicate that they are significant.

89

Case 1a

Figure 5.13 - J48 tree of self-reported productivity based on occupant survey alone
The tree (figure 5.13) indicates that amongst all the survey questions, lighting quality was the most
important attribute that had a significant impact on work productivity and some of those occupants
who reacted positively to lighting quality were also satisfied with the air movement at their
workstations indicated by the value P (1132.2/101.73) which forms 60% of the responses. The
other positive fruits to be noted are aesthetics (P (97/23)), distraction from others (P (196/41)) and
satisfaction with background noise (P (67/19)), but they are not as important as the first positive
fruit. An interesting finding here was that workspace size, and distraction from others formed
negative fruits of the tree.
90

The results of the tree indicates that productivity levels can be increased by improving the negative
fruits i.e., if occupants had a bigger workspace size, had more privacy to conversations and if they
were not distracted by others (highlighted by the value N (123/21)).

Case 1b

Figure 5.14 - J48 tree of self-reported productivity based on occupant survey and two spatial factors
When human factor was added to the list of variables, it is seen that the first few attributes (lighting
quality, air movement and workspace size) remain unchanged (figure 5.14). The same positive and
negative fruits were bore in this tree as well. The only change seen is in the classification path for
distraction from others. Age groups were analyzed to show that the senior age group reacted
positively (P (10.13/0.56)) and did not think that distraction from others affected their productivity
levels, again, this is probably because they had private office spaces. while occupants belonging
to the junior age group reported that their productivity levels were affected due to distractions from
their colleagues and therefore reacted negatively (indicated by N (6.75/3.38)). The mid-age group
was concerned with the distance between them and their co-workers. However, these instances are
101
19
91

very less and therefore, not significantly affecting self-reported productivity levels. Again, the
suggestions here are to provide occupants with a bigger workspace, and privacy to have
conversations.
Case 1c

Figure 5.15 - J48 tree of self-reported productivity based on occupant surveys, spatial and human factor
Figure 5.15 is the result of having occupant responses and the two spatial factors; location and
enclosure of workstation as independent variables, and the result is the same as the first tree (1a).
The fruits also have the same associated sample sets (such as P (1132/101), and (P (97/23)). This
indicates that the spatial aspects: perimeter/ interior and open/close did not lead to significant
changes in the way people reported their productivity levels. This could also be because the dataset
did not have extreme responses. However, it is to be noted that workspace enclosure appears as
one of the factors in row six as a survey question that asks occupants about their satisfaction with
the enclosure, and it is not to be confused with the spatial factor “Workstation enclosure” that was
92

noted down by the CMU team at the time of data collection. Also, in sections 5.2.3 and 5.2.4, the
results indicated that occupants in closed workstations and perimeter workstations reported higher
productive levels than occupants in open workstations and interior workstations respectively, but
the difference in the average values was not large. This decision tree furthers supports the finding.
Case 1d

Figure 5.16 - J48 tree of self-reported productivity based on occupant survey spatial, human factor and
measured data
The fourth tree (figure 5.16) is a result of considering spatial aspects, survey responses, and human
factor. The result is exactly the same as the second tree (1b) which considered survey responses,
and human factor. This, again, suggests that spatial factor did not change the attributes or their
hierarchy. Some of the last fruits have values such as P (4/1), P (22/7), and P (2.0) which is marked
by dashed rectangles. Although these are positive fruits, they are not significant in improving
occupant work productivity levels. The pattern to be observed is that the suggestions to improve
work productivity levels seem to be repeating, i.e., by providing occupants a bigger workspace
size, privacy to have conversations. Occupants were also concerned about being distracted by
/19
93

others, this factor can probably be taken care of if they are provided with bigger workspaces so
that distance between them and their immediate co-workers increases.
Case 1e

Figure 5.17 - J48 tree of self-reported productivity based on the entire dataset
The purpose of adding or changing independent variables was to identify how the attributes
changed every time. The last tree in the first set (figure 5.17) is a result of considering the entire
dataset as independent variables which includes user responses, age, gender, location, enclosure
and, measured IEQ data such as relative humidity and lux levels. Age group and season are factors
that appear in the negative branch. However, the associated sample sets such as N (7.2/0.2), P
(13.37/2.0), P (1.03), N (11.31/2.31) and P (3.09) are smaller values and therefore not very
important to be considered to improve occupant work productivity. The main attributes and the
sample set of positive fruits still remain unchanged.

.0/ 19
94

5.3.2 Satisfaction with the indoor environment
Case 2a

Figure 5.18 - J48 tree of satisfaction with IEQ based on survey only

The same process was repeated with Q 29. Satisfaction with IEQ to find patterns in the dataset.
The first tree considers only the survey and it shows that air quality was a main attribute, followed
by light for computer work and lighting quality. It is interesting to see that overall lighting quality
formed a positive fruit (P (1295/150)) which holds good for 69% of the occupants, and light
required for computer-based tasks formed a negative fruit (N (116/24)) which holds good for 6%
of the occupants. The previous sections in chapter 4 and 5 also indicate that occupants were very
satisfied with the lighting quality at their workstations. The other positive fruit to be noted is
workspace enclosure in the positive branch (P (81/17)). Some of the negative fruits include the
questions about being able to alter physical conditions (N (65/12)), and noise due to conversations
(N (69/16)) which holds good for 3% of the population. To improve occupants’ satisfaction with
the indoor environment, they could be provided with modular workstations, or the ability to alter
95

their workspace to some extent and use more of acoustic materials to reduce the noise generated
in office spaces.
Case 2b

Figure 5.19 - Satisfaction with IEQ based on survey and spatial factor
The second tree (figure 5.19) considers survey responses and human factor, the result is quite
similar to the previous tree. Air quality, lighting quality and light required for computer work still
remain the main attributes. The only change in the classification path observed is that gender forms
the last fruits of the tree. It was expected to see gender play a crucial role, but it does not seem to
be doing so as the associated sample cases is very less i.e., (N 5.7/0.71) and P (2.29). The notable
positive and negative fruits are the same as the ones seen in the previous tree. To accomplish
significant IEQ satisfaction levels, human factor does not seem to be a critical factor to be
considered.

96

Case 2c

Figure 5.20 - Satisfaction with IEQ based on survey, spatial and human factors
The result of considering survey responses, and spatial factor (location and enclosure of
workstations) is illustrated in figure 5.20. Many researchers and studies have indicated that
workstation location matters. However, the results achieved here indicate otherwise. The tree is
the same as the first tree of the second set (figure 70), and it suggests that the two spatial factors
did not change the fruits of the trees or their hierarchy. This indicates that location, and enclosure
are not critical factors to ensure satisfaction with the indoor environment. Here again, it is to be
noted that workspace enclosure appears as one of the factors in row three as a survey question that
asks occupants about their satisfaction with the enclosure, and it is not to be confused with the
spatial factor “Workstation enclosure” that was noted down by the CMU team at the time of data
collection. Also, in sections 5.2.3 and 5.2.4, the results indicated that occupants in closed
workstations and perimeter workstations reported higher productive levels than occupants in open
97

workstations and interior workstations respectively, but the difference in the average values was
not large. This decision tree furthers supports the finding.
Case 2d

Figure 5.21 - Satisfaction with IEQ based on survey, spatial, human factors, and measured data
The result of considering survey responses, human, and spatial factors (figure 73) is the same as
shown in figure 5.21, indicating that human and spatial factors together did not change the
hierarchy of the attributes. The pattern to be observed is that the suggestions to improve satisfaction
with the indoor environment seem to be repeating, i.e., by providing occupants the ability to alter
their physical aspects of their workstation and to find ways to reduce the noise generated at their
workplace. The associated sample sets such as P (1295/150), P (163/61), N (116/24), are also same
the previous ones. Gender forms the last fruits in this tree as well, but not a critical factor that helps
in improving the indoor conditions of an office environment.

98

Case 2e

Figure 5.22 - Satisfaction with IEQ based on entire dataset
Figure 5.22 is the result of having the entire dataset in the list of independent variables. Here again
the main impact attributes remain air quality, lighting quality, and light required for paper-based
work. The associated sample instances that are marked in rectangles are also the same as previous
trees. The classification path ends differently than the other trees. Season and workspace size form
the last nodes/ fruits of the tree. However, the sample cases (N (2/1), P (8/2), P (4/1), and N (13/1))
are less and therefore not significant to be considered as a factor that could improve the indoor
environment.

Overall, from the above data mining analysis using WEKA, a pattern was identified, and it
indicated that the indoor air quality and lighting quality at workstations have positively dominated
the dataset. Despite their significance, positive work productivity is not guaranteed when the other
99

IEQ elements (spatial, acoustic, and thermal) are not perceived in a positive way. Table 5.2 is a
summary of the two tree sets, both sets identified lighting and air quality as the main attributes
impacting occupant work productivity and satisfaction with indoor environment.
Table 5.2 Summary of the decision tree results
IEQ
factor
Q28. self -reported productivity
1a 1b 1c 1d 1e
Lighting #1 #1 #1 #1 #1
Air #2 #2 #2 #2 #2
Spatial #3 #3 #3 #3 #3
Acoustic #4 #4 #4 #4 #4
Thermal #5 #5 #5 #5 #5

IEQ
factor
Q29. Satisfaction with IEQ
2a 2b 2c 2d 2e
Lighting #2 #2 #2 #2 #2
Air #1 #1 #1 #1 #1
Spatial #3 #3 #3 #3 #3
Acoustic #4 #4 #4 #4 #4
Thermal #5 #5 #5 #5 #5

5.4 Summary
Forward regression analyses highlighted certain predictors that played a significant role on one’s
productivity levels, the first few are air movement, workspace size and air quality. The survey
responses were then studied for different occupant groups. Human and spatial factors do account
for variations in one’s comfort and preferences. Occupants seated in closed workstations and
perimeter workstations had higher satisfaction levels than occupants in open and interior
workstations, respectively. The IEQ ranking order was the same for mid-age and senior group and
changed for junior group. It was also different between men and women. Decision trees were used
100

to identify patterns in the dataset. With regard to work productivity, it identified lighting quality
as the most important attribute followed by air movement and workspace size. The attributes
changed when Q29 (satisfaction with IEQ) was chosen as the dependent attribute; they were air
quality, lighting quality and light computer work.
101

CHAPTER 6
6.1 Discussion
Post occupancy evaluation is an effective tool. It is based on reviewing and analyzing occupant
responses which in return can help various users of the building. Catering to everyone’s needs is
difficult. POE is labor intensive and takes a lot of time to distribute paper-based questionnaires to
every individual. The indoor environmental conditions (air, thermal, visual, acoustic, and spatial)
at 1865 workstations from over 60 modern office buildings were studied as part of an integrated
POE (which takes into consideration the spot measurements of temperature, lighting levels,
decibel, etc.). The offices were located in research and development, public, private, and
commercial buildings. Measured data and occupant responses were studied correspondingly. The
data was correlated with human factors such as age, gender, and spatial factors such as workstation
location and its degree of enclosure. Statistical analyses show interesting relations between these
factors, measured data, and occupant responses from the survey. Also, different occupant groups
have different preferences and satisfaction levels. If adaptive environmental control is ensured at
workstations, occupants would be more satisfied with their immediate surroundings and it will
help in achieving higher self-reported productivity rates. Integrated post occupancy evaluations
are elaborate and may take six months or more to complete as they consider measured data. These
help in assessing the recommended ranges set by building standards such as ASHRAE and IESNA.
One of the sections in chapter 4 discusses the recommended range versus the measure values.
Lighting quality is a factor that the occupants were most satisfied with, but about half of the
measured data lie beyond the recommended range.
Over the last few years, digital versions of occupant surveys have been implemented where the
responses and datasets are formulated in lesser time, this makes it easier to study the data and
102

report on the exiting conditions sooner than it would take otherwise. Smart devices have been
created to ensure occupant thermal comfort; one such example is the Hyper-chair developed at the
University of California-Berkeley. This includes Bluetooth, an option to connect to a wireless
network, heating, and a cooling system that gives the user personal control. Although these chairs
cost more than a regular chair, they help in offsetting utility costs and eventually benefit the
environment (Feinn 2016).
6.2 Limitations and future work
It is likely that datasets will have some missing values either because some occupants would not
have responded to certain questions or because some measured values were not translated correctly
to the excel database. Initial results were achieved through IBM’s Statistical Package for the Social
Sciences (SPSS). However, it was simpler to get some desired graphs/ plots (confidence interval
plots and ANOVA) using Minitab. It was also uncomplicated to work with Minitab, in terms of
interoperability with Microsoft Excel.
With regard to future work, the occupant groups could be further categorized and sorted into
smaller groups such as junior women, senior men, mid-age women, etc. and their responses
corresponding to the measured IEQ data could be explored. This would shift the focus of the study
and target smaller occupant groups. Also, more detailed information of the buildings (their
location, type, size, biophilia, and number of floors) could be used for further investigations and
comparisons. Only two spatial features of workstations were considered. For further explorations,
the number of enclosures to a workstation and the partition heights could be considered. From the
results, it was seen that acoustic quality is something that was a matter of concern and decibel
levels were not considered for the tests due to technical irregularities. Since occupants were
103

satisfied with the lighting quality, a comparative study can be made between natural lighting and
artificial lighting along with its color temperature.
6.3 Conclusion
Occupant responses were collected from 1865 people and 1631 occupants informed the CMU team
about their personal information. Women formed 54% of the population and Men formed 46%.
Occupants were categorized into three age groups, almost half of them belonged to the Mid-age
group (48%), followed by the senior (33%) and junior (22%) group.
The measured IEQ values which occupants preferred or were satisfied with are tabulated to
compare with the standard or guideline values. With respect to age and gender (Table 6.1) it is
seen that most of the average carbon-dioxide values comply with the recommended values. The
average lux level recommended by IES is 500 lux and all the average light levels are much above
the recommended range and occupants were satisfied with higher levels. The relative humidity
levels and temperatures also comply with the recommended range. Table 6.2 indicates that when
data was classified based on spatial factors, the average carbon-dioxide levels were below the
minimum recommended value (700 ppm). The average light levels for interior workstations were
lower than the recommended value. The average values of RH and temperature match the
recommended numbers.

104

Table 6.1 - Comparison between values set by STANDARDS and measured vales (Based on human
factors)

Table 6.2 - Comparison between values set by STANDARDS and measured values (based on spatial
factors)

Hierarchy of factors: The factors were ranked for different occupant groups; all groups expressed
least satisfaction with acoustic quality and most satisfaction with the air quality. The remaining
three qualities took different ranks for different groups. Patterns in the dataset that affect occupant
work productivity and their satisfaction with the indoor environment were identified through
decision tree results. Lighting and air quality variables formed the main attributes. Thermal quality
variables formed the last nodes of the tree which means that thermal quality did not have a
noticeable impact on users’ work productivity levels.
Gender and age: By studying the results between women and men it was observed that women and
men reacted similarly to most questions, and statistical significance was only seen in four questions
105

(Q1. Light on the desk for paper-based tasks, Q2. Overall air quality, Q3. Temperature at your
workstation, and Q18 (overall lighting quality in your work area) and in all the questions the mean
desired values were lower for women than men. Studies have shown that women prefer warmer
temperatures and that they are more sensitive to temperature changes. One of the objectives was
to identify which gender group was more comfortable than the other, women and men reacted
similarly and had about the same average values except for one result which indicated that men
were more comfortable than women in the Swing season. Overall, the comfortable range was
between 22.75 and 24 degrees C. Out of the three age groups, the junior group, comprising 22%
of the population, was the most satisfied and the mid-age group making 48% of the population set
was the most dissatisfied with the indoor conditions followed by the senior group. Based on
average self-reported productivity levels, junior women (3.98) and senior women (4.09) were less
productive than junior men (4.51) and senior men (4.77) respectively, and mid-age women (4.33)
were more productive than mid-age men (4.11).
Occupation: Occupants with a professional background were the most in number, followed by
managerial, administrative, technical, and doctorate positions. Occupants that held doctorate or
managerial positions had higher satisfaction levels indicating that people at higher positions had a
better workstation or had the ability to modify their settings and preferences.
The first objective was to find out whether thermal quality has the most impact on occupants, and
it was found that it is actually acoustic quality that has the most impact. The second objective was
to find out about self-reported productivity levels in closed and open workstations. It was found
that occupants working in open plan and closed plan layouts have different productivity levels;
people in closed spaces had higher productive levels (1.82) than people in open workstations
(1.67). However, it is not very significant as the difference in the average self-reported productivity
106

levels is not large. Although occupants using perimeter workstations (located within 15’ away
from the external walls) reported higher productivity levels (average values of 1.72) than people
working in the interior zone (average value of 1.68), work location did not matter as again, the
difference is not large.

107

REFERENCES

Administration, U.S. General Service. 2019. 6.15 Lighting. 02 26. Accessed 02 09, 2021.
https://www.gsa.gov/node/82715.
Agency, United States Environmental Protection. 1991. "Indoor Air Facts No. 4 (revised) Sick Building
Syndrome."
al, V Loftness et. 2009. "The value of post occupancy evaluation for building occupants and facility
managers." INtell Build Int. Vol 1 249-268.
Anna Carolina, Andrew Cripps, Dino Bouchlaghem, Richard Buswell. 2011. Predicted vs. Actual Energy
Performance of Non-Domestic Buildings:Using Post Occupancy Evaluation Data to Reduce the
Performance Gap. Elsevier.
Bhaskar N. Patel, Satish G. Prajapati and Dr. Kamaljit I. Lakhtaria. 2012. "Efficient Classification of Data
Using Decision Tree." Bonfring International Journal of Data Mining, Vol. 2, No. 1, 6-12.
Blyussen, Philomena M. 2009. The Indoor Environment Handbook: How to Make Buildings Healthy and
Comfortable. London, New York: Earth Scan.
Cone, James E. 1998. "Indoor Environmental Quality." Western Journal of Medicine.
Council, Federal Facilities. 2001. LEARNING FROM OUR BUILDINGS - A State-of-the-Practice Summary of
POE. Washington DC: Natiional Academy Press.
Crozier, Karim Hadjri and Carl. 2008. "Post-occupancy evaluation: purpose, benefits and barriers."
Emerald 21-33.
Danielsson, Christina Bodin, Holendro Singh Chungkham, Wulff Cornelia, and Hugo Westerlund. 2014.
"Office design's impact on sick leave rates." Ergonomics; Taylor & Francis 139-147.
EPA, US. 1991. "Indoor Air Facts No.4 Sick Building Syndrome."
Federal, and Facilities Council. 2002. Learning from Our Buildings: A state-of-the-Practice Summary of
Post-Occupancy Evaluation. National Academic Press.
Feinn, Lily. 2016. This Chair Keeps You Warm In Your Freezing Office. April 20. Accessed February 25th,
2021. https://www.bustle.com/articles/155938-smart-chair-keeps-you-warm-at-your-office-
desk-so-you-can-leave-the-summer-sweater.
Finell, Eerika, and Jouko Nätti. 2019. "The Combined Effect of Poor Perceived Indoor Environmental
Quality and Psychosocial Stressors on Long-Term Sickness Absence in the Workplace:A Follow-
Up Study." International Journal of Environmental Research and Public Health 1-13.
Freudenrich, William Harris & Craig. 2000. How stuff works. July 10. Accessed October 10, 2020.
https://science.howstuffworks.com/light.htm> 9 October 2020.
Group, BRE. n.d. "Post Occupancy Evaluation (POE)."
Hadjri, Karim, and Carl Crozier. 2009. "Post-occupancy evaluation: purpose, benefits and barriers."
Facilities vol 27 21-33.
108

Hess, Aaron S. Hess and John R. 2017. "Clinical Research focus." Transfusion 2319-2320.
Hintze, Dr. Jerry L. 2007. Two-Sample T-Test. Kaysville, Utah : NCSS.
Hopkinson, R.G. 1972. "Glare from daytighting in buildings." Applied Ergonomics 206-215.
Huizenga, C, S Abbaszadeh, and Leah Zagreus . 2006. "Air quality and thermal comfort in office buildings:
Results of a large indoor environmental quality survey." Proceeding of Healthy Buildings 393-
397.
IBM. n.d. IBM SPSS Statistics. Accessed October 2nd , 2020. https://www.ibm.com/products/spss-
statistics?lnk=STW_US_HP_T1_BLK&psrc=LOT13&pexp=CDOTO&lnk2=trial_SPSSstat.
Iuliano, Monica Franzese and Antonella. 2019. "Correlation Analysis." Elsevier: Encyclopedia of
Bioinformatics and Computational Biology, Volume 1 709-721.
J.H. Choi, V. Loftness and A. Aziz. 2012. "Post-occupancy evaluation of 20 office buildings as basis for
future IEQ standards and guidelines." Energy Build, vol 46 167-175.
JoonHo Choi, Azizan Aziz, Vivian Loftness. 2010. "Investigation on the impacts of different genders and
ages on satisfaction with thermal environments in office buildings." Elsevier 1529-1535.
Kossowski, Jan Hauke and Tomasz. 2011. "COMPARISON OF VALUES OF PEARSON’S AND SPEARMAN’S
CORRELATION COEFFICIENTS ON THE SAME SETS OF DATA." Questiones Geographcae 87-93.
Kumar, Himani Sharma and Sunil. 2013. "A Survey on Decision Tree Algorithms of Classification in Data
Mining." International Journal of Science and Research (IJSR) 2093-2097.
Mallinger, Stephen. 1996. Occupational Safety and HEalth Administration . June 5.
https://www.osha.gov/dts/hib/hib_data/hib19960605.html.
Maohui Luo, Zhe Wang, Kevin Ke et al. 2018. "Human metabolic rate and thermal comfort in buildings:
The problem and challenge." Building and Environment 44-52.
Marmot, A.F. 1991. "The Good Office: POST-OCCUPANCY EVALUATION OF OFFICE BUILDINGS." Facilities,
Vol 9 10-13.
Miller, Jenesse. 2019. USC News. May 22. Accessed September 20, 2020.
https://news.usc.edu/157448/female-productivity-warmer-temperature/.
Minitab. 2020. What is ANOVA? Accessed September 19th , 2020. https://support.minitab.com/en-
us/minitab/19/help-and-how-to/statistical-modeling/anova/supporting-topics/basics/what-is-
anova/#:~:text=Learn%20more%20about%20Minitab%2019,at%20the%20different%20factor%2
0levels.
Moon, Jehyun. 2016. "Comprehensive POE Research on Office Environment of Southern California."
NASA Science. 2012. Measuring Light Levels | Sustainability Workshop. October 9th. Accessed October
22, 2020. https://science.nasa.gov/ems/09_visiblelight.
O.A. Sepp€anen, W. Fisk. 2006. "Some quantitative relations between indoor envi-ronmental quality and
work performance or health." Hvac& R Res. 12 957-973.
109

Özgür Göçer, Kenan Göçer, Ebru Ergöz Karahan & Işıl İlhan Oygür. 2017. "Exploring mobility & workplace
choice in a flexible office through post-occupancy evaluation." Taylor and Francis 226-242.
Park, Ji Hyun. 2015. Are Humans Good Sensors? Using Occupants as Sensors for IEQ assessment and for
developing thresholds that matter. Pittsburgh: CMU.
Persily, Andrew. 2020. "Quit Blaming ASHRAE Standard 62.1 for 1000 ppm CO2."
Preiser, Wolfgang FE. 1995. "Post-occupancy evaluation: How to make buildings work better." Facilitites
13. no 11 19-28.
Prevention, Center for Disease Control and. 2015. The National Institute for Occupational Safety and
Health (NIOSH). September 1. Accessed January 28, 2021.
https://www.cdc.gov/niosh/topics/indoorenv/temperature.html#:~:text=ASHRAE%20also%20re
commends%20that%20indoor,mold%20growth%20%5BEPA%202012%5D.
Rokach, Lior, and Oded Z Maimon. 2008. Data Mining with Decision Trees: Theory and Applications.
World Scientific Publishing Co Pte Ltd .
Schiavon, Stefano, and Clara Peretti. 2011. "Indoor environmental quality surveys. A brief literature
review." escholarship.
Shengxian Kang, Dayi Ou, Cheuk Ming Mak. 2017. "The impact of indoor environmental quality on work
productivity in university open-plan research offices." Building and Environment 78-89.
Smithgroup. n.d. "Post-Occupancy Evaluations ."
Statistics, Laerd. 2018. One-way ANOVA in SPSS Statistics.
Stehlik-Barry, Anthony J. Babinec and Kenneth. 2017. Data Analysis with IBM Statistics. Packt Publishing
.
Syed Ihtsham ul Haq Gilania, Muhammad Hammad Khanb, William Pao. 2015. "Thermal comfort analysis
of PMV model Prediction in Air conditioned and Naturally Ventilated Buildings ." Elsevier: Energy
Procedia 1373-1379.
Toyinbo, Oluyemi. 2019. "Indoor Environmental Quality." Sustainable Construction Technologies. 107-
122.
Waikato, University of. n.d. WEKA The workbench for machine learning. Accessed January 20th , 2021.
https://waikato.github.io/weka-wiki/development/.
Wargocki, P. 2011. "Productivity and Health Effects of High Indoor Air Quality." Elsevier 688-693.
Young, Derek Scott. 2017. "Basics of Regresion Models." In Handbook of Regression Methods, by Derek
Scott Young, 3-25. Milton: Chapman and Hall/CRC.
Yousef Al Horr, Mohammed Arif, Amit Kaushik, Ahmed Mazroei, Martha Katafygiotou and Esam Elsarrag.
2016. "Occupant productivity and office indoor environment quality: A review of the literature."
Elsevier: Building and Environmnet 369-389.
110

Yousef Al horr, Mohammed Arif, Martha Katafygiotou, Ahmed Mazroei, Amit Kaushik, , Esam Elsarrag.
2016. "Impact of indoor environmental quality on occupant well-being and comfort: A review of
the literature." International Journal of Sustainable Built Environment 1-11.
Zagreus, L, C Huizenga, and E Arens. 2004. Listening to the occupants: a web-based indoor
environmental quality survey. Survey report, Denmark: Blackwell Munksgaard.
Zdenka Budaiova, Silvia Vilcekova. 2015. "Assessing the effect of indoor environmental quality on
productivity at office work ." SSP - JOURNAL OF CIVIL ENGINEERING Vol. 10, Issue 1 37-46.

111

APPENDIX

The charts below are bar charts of survey responses to each of the scale based IEQ question.

47
105
129
298
255
691
443
0
100
200
300
400
500
600
700
800
1 2 3 4 5 6 7
Light for paper based work (Q1)
80
141
200
386
272
686
284
0
100
200
300
400
500
600
700
800
1 2 3 4 5 6 7
Air quality (Q2)
112

167
253
478
299
365
388
146
0
100
200
300
400
500
600
1 2 3 4 5 6 7
Temperature in your work area (Q3)
76
156
232
454
288
537
252
0
100
200
300
400
500
600
1 2 3 4 5 6 7
Aesthetics of your work area (Q4)
113

399
354
336
227
221
279
162
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7
`Privacy to have conversations (Q5)
251
288
276
312
283
418
220
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7
Level of visual privacy in your work area (Q6)
114

313
395
404
306
243
262
129
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7
Noise due to people's conversations (Q7)
113
144
270
246
292
607
384
0
100
200
300
400
500
600
700
1 2 3 4 5 6 7
Workspace size (Q8)
115

128
159
284
47
278
528
267
0
100
200
300
400
500
600
1 2 3 4 5 6 7
Background noise (Q9)
41
117
144
353
316
697
379
0
100
200
300
400
500
600
700
800
1 2 3 4 5 6 7
Light for computer work (Q10)
116

110
208
302
498
277
414
165
0
100
200
300
400
500
600
1 2 3 4 5 6 7
Air movement in your work area (Q14)
195
299
316
591
255
299
123
0
100
200
300
400
500
600
700
1 2 3 4 5 6 7
Your ability to alter physical conditions in your work area
(Q15)
117

334
215
180
214
176
396
538
0
100
200
300
400
500
600
1 2 3 4 5 6 7
Your access to a view of outside from where you sit
(Q16)
114 110
212
451
292
550
323
0
100
200
300
400
500
600
1 2 3 4 5 6 7
Distance between you and other people you work with
(Q17)
118

48
121
190
357
337
641
355
0
100
200
300
400
500
600
700
1 2 3 4 5 6 7
Overall quality of lighting in your work area (Q18)
214
226
396
440
256
363
154
0
50
100
150
200
250
300
350
400
450
500
1 2 3 4 5 6 7
Frequency of distraction from other people (Q19)
119

163
206
227
440
250
478
268
0
100
200
300
400
500
600
1 2 3 4 5 6 7
Degree of enclosure of your work area (Q20)

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Impacts of indoor environmental quality on occupants environmental comfort: a post occupancy evaluation study

PDF

Considering occupants: comprehensive POE research on office environment of Southern California

PDF

Nursing comfort study: are hospitals performing for nurses? IEQ analysis of working conditions inside hospitals

PDF

Enhanced post occupancy evaluation (POE) for office building: improvement of current methodology to identify impact of ambient environment

PDF

Impact of occupants in building performance: extracting information from building data

PDF

Impact of bamboo materials on indoor air quality

PDF

Natural ventilation in tall buildings: development of design guidelines based on climate and building height

PDF

IEQ, sleep quality, and IoT: meta-analysis on improving IEQ and sleep quality using IoT

PDF

Developing environmental controls using a data-driven approach for enhancing environmental comfort and energy performance

PDF

A proposal for building envelope retrofit on the Bonaventure Hotel: a case study examining energy and carbon

PDF

Post occupancy energy evaluation of Ronald Tutor Hall using eQUEST; computer based simulation of existing building and comparison of data

PDF

Human-building integration based on biometric signal analysis: investigation of the relationships between human comfort and IEQ in a multi-occupancy condition

PDF

Energy use intensity estimation method based on building façade features by using regression models

PDF

Building energy performance estimation approach: facade visual information-driven benchmark performance model

PDF

Leaf module panel system: novel vertical wall wind energy production

PDF

Human-building integration: Investigation of human eye pupil sizes as a measure of visual sensation in the workstation environment

PDF

Exploring participatory sensing and the Internet of things to evaluate temperature setpoint policy and potential of overheating/overcooling of spaces on the USC campus

PDF

Indoor air quality for human health in residential buildings

PDF

Bridging performance gaps by occupancy and weather data-driven energy prediction modeling using neural networks

PDF

Daylighting study of a LEED platinum laboratory building: a post-occupancy evaluation comparing performance in use to design intent

Asset Metadata

Creator Virupaksha Swamy, Aishwarya (author)

Core Title Impacts of building performance on occupants' work productivity: a post occupancy evaluation study

School School of Architecture

Degree Master of Building Science

Degree Program Building Science

Publication Date 04/15/2021

Defense Date 03/17/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag indoor environmental quality,OAI-PMH Harvest,occupant satisfaction,post occupancy evaluation,self-reported productivity

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Choi, Joon-Ho (committee chair), Noble, Douglas (committee member), Schiler, Marc (committee member)

Creator Email d.v.aishwarya@gmail.com,virupaks@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-443598

Unique identifier UC11668412

Identifier etd-Virupaksha-9468.pdf (filename),usctheses-c89-443598 (legacy record id)

Legacy Identifier etd-Virupaksha-9468.pdf

Dmrecord 443598

Document Type Thesis

Rights Virupaksha Swamy, Aishwarya

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA