University of Southern California Dissertations and Theses

Using rate-building to decrease latency and increase accuracy for task completion via telehealth

(USC Thesis Other)

Using rate-building to decrease latency and increase accuracy for task completion via telehealth

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
Copyright 2021 Andrew George Hall

Using Rate-Building to Decrease Latency and Increase Accuracy for Task Completion via
Telehealth
by

Andrew George Hall

A Thesis Presented to the
FACULTY OF THE USC DANA AND DAVID DORNSIFE COLLEGE OF LETTERS, ARTS,
AND SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(APPLIED BEHAVIOR ANALYSIS)
August 2021

TABLE OF CONTENTS
List of Figures……………………………………………………………………………………..ii
Abstract…………………………………………………………………………………………...iii
Introduction………………………………………………………………………………………..1
Methodology………………………………………………………………………………………5
Participants & Setting……………………………………………………………………..5
Materials………………………………………………………………………………..…5
Response Measurement and Inter-Observer Agreement (IOA)…………………………...7
Procedures…………………………………………………………………………………8
Preparation & Baseline……………………………………………………………………8
Intervention………………………………………………………………………………..9
Maintenance Phase………….……………………………………………………………11
Results...………………………………………………………………………………………….11
Discussion………………………………………………………………………………………..14
References………………………………………………………………………………………..21

ii
List of Figures
Figure 1- Multiple Baseline Graph Tracking Changes in Accuracy and Latency ………………23
Figure 2- Graphs Comparing Accuracy and Latency of Session 1 & 2 From Baseline to
Maintenance……………………………………………………………………………………...24

iii
To ensure the highest quality of life possible, performing skills with fluency is vital. Behavioral
fluency is a combination of both response accuracy and response rate, that when achieved results
in behavior occurring smoothly, rapidly, and with little apparent effort (Binder, 1996). Fluency is
important in the academic setting because students are expected to demonstrate the same, or
similar, skills across various assignments, especially since fluency assists both in generalization
and helps prevent relapse. This study focused on conducting rate-building and fluency training
via online sessions, using a changing criterion embedded within a multiple baseline. Specifically,
a fluency procedure was used to teach three participants to identify grammatical errors across
different writing samples. The data showed that across all participants, there was an increase in
the accuracy of errors corrected, and a decrease in the latency and time needed to complete the
task.
Keywords: Behavioral Fluency, Rate-Building, Tele-education, Education

1

A Revised Look at Using Differential Reinforcement to Decrease Latency and Increase
Accuracy for Task Completion During Skill Acquisition
While there currently exists a paucity of literature around the use of telehealth as a means of
delivering Applied Behavior Analysis (ABA) services, a handful of studies that have been
conducted before the COVID-19 pandemic. Tomlinson, Gore, and McGill (2018) stated training
individuals to implement ABA procedures using telehealth services (THS) looks “promising,”
and they suggest that telehealth may be an effective way to train people in ABA techniques. The
current pandemic and sudden grappling with the forced limitation notwithstanding, other
researchers from other professions within psychology have provided potential advantages in
developing and expanding current research, methodology, and technology. Myers and Comer
(2016) discussed the current difficulties in delivering mental health services that could be
remedied by implementing THS. They stated one current challenge in our current model for
delivering mental health services comes from the lack of child mental health specialists living in
rural areas. This leads to children living outside urban/suburban settings or outside of major
metropolitan areas underserved and in need of services (Meyers & Comer, 2016). Other
researchers such as McMahon and Thakore have begun tackling some of these shortcomings for
THS when applied to educational settings, emphasizing programming considerations such as the
amount information given at once and the kind of feedback given to learners when attempting to
teach via telehealth to endure lesson retention and not overwhelming the learners (2008).
Further research from other fields besides ABA also supports this idea. Pennefather,
Hieneman, Raulston & Caraway (2018) evaluated the use an online training program to teach
participants basic ABA implementation with their children. The three-week online training
program covered topics ranging from instruction in ABA principles and stress reduction

2

strategies, to meditation practice based on Acceptance and Commitment Therapy (ACT)
principles. The program also included weekly synchronous online meetings with other parents
and two parent educators. Researchers reported that when they asked parent participants what
effects the online training program had for them, the parents reported decreased amount of
parental stress, increased amounts of relevant knowledge, increased amounts of prosocial
behavior from their child, decreased amounts of hyperactive behaviors, and significantly high
levels of satisfaction with the results of the intervention (Pennefather et al., 2018).
Ingersoll et al. (2017) contributed to the argument of extending services through THS in that
their data suggested other strategies must be develop in order to support families using a THS
model for parent-mediated interventions, particularly in community settings (Ingersoll et al,
2017). In their research, they found that while THS can be used as a means to help teach and
implement interventions for children with ASD, there still were barriers for THS efficacy. Such
barriers included a lack of mobility/flexibility in the modality of the THS program being used, a
lack of consistent engagement from the parents using the THS, and the method by which a THS
modality attempted to impart information or techniques (i.e., the study primarily used narrated
slideshows which were 27.2 minutes long on average). The studies described above lend further
credence for the need to research and develop effective methodologies for delivering or
implementing ABA techniques through interfaces such as video chat.
Additionally, across multiple journals and continents, there exists a wealth of literature
regarding the use of positive reinforcement for behavior modification. A more recent example of
this was in Gill (2016) where the effects of reinforcement were evaluated on the motivation of
student learning. Based on both the literature review and the results of this research that positive
reinforcement was a powerful tool for both increasing student learning and for changing student

3

behavior . In addition, positive reinforcement can lead to long-lasting change within student
behavior (Gill, 2016). And while we do know a great deal about how to use reinforcement and
other ABA principles to increase the accuracy of an attempted task or skill, there still exists a
lack of methodologies we can use to teach skills to fluency while they are in the process of skill
acquisition. While there are methods such as Precision Teaching (PT) have begun to address this,
PT is not without its potential shortcomings, not to mention the lack of research supporting its
use on young children who qualify as Early Intensive Behavioral Intervention (EIBI), all of
whom are typically more severely impacted by developmental disorders like ASD (Heinicke et
al., 2010). Even then, PT still relies on our current paradigm that focuses on teaching a skill to a
high rate of accuracy first before trying to teach the skill to fluency, as opposed to teaching
accuracy and fluency at the same time for the skill being practiced. Given the lack of knowledge
in these areas, the importance in researching these areas, and the possible future of how ABA as
a field can grow, it would only benefit us to pursue research that adds to and expands upon our
knowledge in these areas.
If a person’s quality of life is determined in part by a person’s ability to perform daily
tasks both accurately and quickly, then it is not enough to simply implement a program that
teachers them a replacement behavior or skill with multiple steps. Rather, the intervention plan
needs to take into account how quickly someone can perform the task or behavior as demanded
by their daily lives as well (Kubina, 2019). A child may complete a math problem or go to the
bathroom by themselves, but if it takes that same child 10 minutes to solve a basic addition
problem or to pull their pants down to use the bathroom, even if they are accurately performing
each step, can we really say that they have the necessary skills in that domain to improve their
quality of life?

4

Questions like this is what influenced Ogden Lindsley to create PT, which puts an emphasis
on fluency, and what it can do for those with ASD (Lindsley, 1992). Lindsley emphasized the
importance of this kind of learning in his research; learning which also tends to be the most
efficient and least costly kind of learning, which can be applied to a variety of kinds of learning
and learning environments (Lindsley, 1992). These results and direction have been further
supported by other researchers who have systemically reviewed the effectiveness of PT for
individuals with ASD. Ramey et al. conducted such a review recently, and found that PT has had
a positive effect on a number of fields and kinds of skills, although more research is needed, and
it currently qualifies as an emerging treatment (2016). A similar evaluation was found by Wiess
et al, who also found that PT has a lot of promise in multiple educational settings (2010).
PT, as it is understood today, operates on paradigm that first focuses on skill accuracy before
implementing the speed of performance aspect that is measured by standard celeration charts
commonly implemented by PT practitioners. Even then, standard celeration charts require the
behavior to be performed accurately and only tracks the number of times a behavior occurs in a
given time frame in a given session, and does not pinpoint which step of the behavior being
performed was done incorrectly. As a result, a method that emphasizes fluency similar to how
PT emphasizes increases behavioral or skill fluency but allows for accuracy rates to be easily
tracked when attempting to implement ABA via THS would be useful for those attempting to use
THS in the future. This combined with a higher degree of methodological rigor and external
validity would contribute to the growing body of knowledge of PT, THS, and ABA overall.

5

Methodology
Participants and Setting
Due to safety precautions during the COVID-19 pandemic, all sessions were conducted
online to reduce face-to-face contact, with the frequency of sessions occurring no more than
twice a week. Additionally, all sessions were recorded to collect interobserver agreement by a
second researcher.
All the participants involved had and expressed an interest to improve upon the quality of
their writing. The three participants were between the ages of 25 and 28 years old, two female
and one male. Participant 1 identified herself as a 27-year-old female graduate student with no
diagnosis. Participant 2 identified himself as a 28-year-old male undergraduate student with a
physical disability from birth that limited his physical capabilities. While there is no precise
name for Participant 2’s physical disability, he was born with no feet, one finger on his right
hand, and a finger and thumb on his left hand. Participant 3 identified herself as a 25-year-old
undergraduate student diagnosed with Anxiety, Autism Spectrum Disorder (ASD), and Dyslexia.
All participants understood the concept of the passage of time and the use of a timer.
Materials
The experiments were conducted in a home setting with the researcher communicating
through telephone with the participant at a scheduled time. The researcher used a Google Doc
and projected it onto the screen through a screenshare feature. For participant selection,
individuals were required to have access to a laptop or computer with internet access, and either
a phone or webcam, so the researcher could talk with the participant. Google Docs was used so
that the researcher could work on the same worksheet as the participants at the same time,
without being in the same room. This also allowed the researcher to disable any autocorrect

6

functionality that would normally be present in a word processing document, and to observe the
participants work in real time.
Before testing, the first step involved developing both the materials needed for measuring
baseline and the intervention. Specifically, the text used to probe current skill levels in baseline
and intervention for each participant needed to be selected and prepared. For consistency and
validity, all the texts used came from the same author and from the same work. The texts chosen
for the worksheets were all from the same book and the same author, just different portions of
the overall text itself, so as to minimize the possibility of internal validity issues due to different
authors, writing styles, etc. In this case, the text chosen was “The Victorian City” by Judith
Flanders. This allowed the study and researcher to control for both genre and style of writing. All
the participants were presented the same prepared texts. This also made it easier to quantify how
many and the type of errors in each text, both in terms of accuracy and latency. This also
provided an effective way to monitor participant progress throughout the study.
There were 15 different sample texts, each between 1.5 and 2 pages long. Various errors
were added to the prepared texts. Additionally, the researcher also made an answer sheet for each
corresponding text that highlighted both where the error was, and the most direct way of
correcting the error. This allowed the researcher to efficiently compare the results turned in by
the participants to the answer sheet. This meant that every session throughout the study had two
copies of the text: one unmarked copy for the participant, and one marked copy used solely by
the researcher.
Participants were presented with worksheets on the screen with a predetermined number of
errors for the participants to identify. A total of 15 were made, each of them 1.5-2 pages double-
spaced in length, and contained anywhere between 30 and 50 errors. Errors included but were

7

not limited to errors in capitalization, punctuation, spelling, syntax, and grammar, all at a high-
school level.
Response Measurement & Inter-Observer Agreement (IOA)
In this study, accuracy was defined as the extent which the participants correctly identified
and corrected the errors in the body of text. For a correction made by the participant during the
trial to be counted as correct, the participant only needed to make a correction so that the end
result was complete and proper sentence in English. Accuracy was measured each session by
dividing the number of errors corrected in a body of text by the overall number of errors in the
body of text, calculating the percentage errors corrected after each trial.
Latency was defined as the amount of time it took to complete the entire task, from the time
correcting the text was started, to the time the participant said they were done, or when 30
minutes had elapsed, whichever happened first. Latency was measured by using a timer to time
how long it took for the participant to the entire task, and was measured in minutes, seconds, and
milliseconds. On the data collection sheet used by the researcher, there was a space to record
how long it took to complete the task presented.
IOA was calculated by a second observer who scored the data independent from the
primary investigator. The protocol for evaluating the permeant product recorded was trained by
the researcher and practiced by the outside observer until both the researcher and outside
observer had 100% agreement on the results for a couple sessions. The results of the second
observer’s data were compared to the evaluation results of the primary researcher, and IOA was
calculated using both the total count method and the interval-by-interval method. Total count
was calculated with the following formula: (
𝑠𝑚𝑎𝑙𝑙𝑒𝑟 𝑐𝑜𝑢𝑛𝑡 𝑙𝑎𝑟𝑔𝑒𝑟 𝑐𝑜𝑢𝑛𝑡 ) × 100. Interval-by-interval was
calculated with the following formula: (
# 𝑜𝑓 𝑖𝑛𝑡𝑒𝑟𝑣𝑎𝑙𝑠 𝑎𝑔𝑟𝑒𝑒𝑑 # 𝑜𝑓 𝑖𝑛𝑡𝑒𝑟𝑣𝑎𝑙𝑠 𝑎𝑔𝑟𝑒𝑒𝑑 +# 𝑜𝑓 𝑖𝑛𝑡𝑒𝑟𝑣𝑎𝑙𝑠 𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒𝑑 ) × 100. IOA

8

levels of 95% were calculated using the total count method, and 93% was calculated using the
interval-by-interval method.
Procedures
Given the current limitations of the pandemic and its impact upon health and service
delivery, all the procedures, materials, and methodologies implemented in this study were aimed
at keeping physical contact between the proctor and the participant to a minimum. The research
design was a changing criterion design embedded within a multiple baseline.
Preparation & Baseline
Participants were instructed to identify and correct errors in a body of text presented on a
shared Google Doc while timed. The participant did not have to vocally identify or correct any
errors found in the text. The participant was instructed to correct any errors and mark them on
the document on the screen itself. The participant only spoke up to let the administrator know
when they were finished. The researcher gave instructions to the participant on what to do before
beginning each assessment, in addition to informing the participant that they will be timed. When
the researcher told the participant to begin, the participant read the text presented, and corrected
any errors seen by editing the Google Doc as quickly as possible. Both the researcher and the
participant had access to the Google Doc simultaneously so that the behavior of the participant
was observed in real time. Additionally, given that the grammar and spelling features were
disabled on the document, the researcher could watch how the participant detected and corrected
errors without the aid of software. To ensure validity, no body of text or data was used more than
once in both the pre-intervention and intervention phase.
The participant for each baseline session had a maximum of 30 minutes from the time the
researcher told the participant to begin to find and correct as many errors as possible. The

9

researcher then recorded the amount of time it took to complete the task and compared the
amount of errors fixed to the number errors there were overall by comparing it to the answer
sheet. If the participant finished early, the participant informed the researcher that they were
finished. The time for the participant to make as many corrections as the participant detected
before saying that he or she is finished was recorded. If the time limit elapsed before the
participant was finished, then the session would have been halted, and the participant would have
been evaluated on what they completed. If they were not finished, edits or parts of the worksheet
that they were not able to get to were not counted against them.
Baseline data consisted of three data points, or until there was a consistent trend in the data.
In baseline, a timer was used to measure how long it took participants to complete the task, the
results of which were recorded on the data sheet.
Intervention
Much of the protocol used in baseline to measure participant performance remained the
same in intervention. Based upon the evaluation of the participant’s performance in the baseline
phase, a tutoring session was first conducted between the researcher and the participant. The
purpose of this was to use a combination of error correction and skill shaping as necessary to
correct patterns of errors not caught or made by the participants in the pre-intervention phase.
Such activities included reviewing rules for punctuation, spelling, or grammar (depending upon
each participant’s needs), opportunities to practice reading or error correction strategies, and ask
questions. Each tutoring session was between 15 and 20 minutes long and was conducted 15 to
20 minutes before attempting the next task presented as was done in the baseline phase.
Before the participant began the assigned task, they were informed that they had a shorter
amount of time to complete the task than before. The amount time each participant had to

10

complete the task in the intervention phase was equal to the average of their time to complete the
task in the pre-intervention phase, rounding up to the closest whole minute. Specifically,
Participant 1 initially had 16 minutes to complete the task, whereas Participant 2 and Participant
3 had 18 minutes initially. As before, each participant would either tell the researcher they were
finished or would be stopped once the given time had elapsed and would be graded based on
what they had. After performing, the researcher evaluated their performance by comparing the
edits the participant made with the answer sheet prepared by the researcher beforehand. As long
as the edits made by the participant fixed the error in question, it was counted as correct. This is
especially true for some kinds of errors more than others, such as grammar errors, where there
are often multiple ways to fix a grammar error in any given sentence. Capitalization and spelling
errors on the other hand only have one correct answer and were easier to judge as to whether the
error was fixed or not. After evaluation, the participant was told how many mistakes they were
able to edit correctly, and their time from start to finish.
If at the end of a session they achieved 70% accuracy or above in finding and correcting
errors and completed the task within the time given, the criterion for latency was reduced by 1 or
2 minutes for the next session, depending on the difference between the recorded latency and the
time given for the task presented. If during the session the participant scored less than 70%
accuracy or reached the time limit and had not finished, the criterion for latency would not
change and be the same for the next session. Once a steady trend had been established in the
intervention, the next session’s latency was increased by 30 seconds as a form of mini-reversal,
to see if the participant’s latency would increase with the increase in time. Participant 1 and
Participant 3 both had mini-reversal sessions in the intervention, but Participant 2 did not. This
decision was made due to the degree of variability in his accuracy, and it was decided that it

11

would be clinically in the best interest of the participant to focus on skill fluency, the main focus
of the research.
Each intervention session was no more than 1 hour in total length and was conducted no
more than twice a week. Intervention sessions were conducted until each participant showed a
trend in performance improvement, and after a minimum of 5 sessions in the intervention phase.
Once the participant was ready based on these criteria to move on to the next phase, there was a
two-week delay between the last intervention session and the first post-intervention session.
Maintenance Phase
Maintenance probes were conducted two weeks after the intervention phase was completed.
The probes involved using the same evaluation procedures used during the baseline and
intervention phases and used two of the same worksheets from the pre-intervention phase. Both
accuracy and latency were recorded and calculated in the same way that it was during baseline,
with the results of the post-intervention probe being compared to the pre-intervention results.
There were two maintenance probes conducted over the course of a week, using the same
scheduling guidelines that was used in both of the previous phases. The maintenance probes
were designed to measure both the accuracy and latency of each participant in a similar manner
as was conducted in the previous phases. To compare the results and improvements in accuracy
and latency as a result of the intervention, the first two sheets that were used in the baseline
phase were also used in the two maintenance probes, each of them presented in the same order
and with the same protocol as they were in the baseline phase.
Results
The results of the intervention and post-intervention probes have shown two key changes in
behavior. The graph presented in Figure 1 presents the data of all 3 participants. The accuracy

12

and latency were recorded for each session and recorded on the graph. Figure 1 is a double y-
axis graph, with the percentage of accuracy on the left y-axis with data shown in solid lines, and
the latency in minutes labeled on the right y-axis with data shown in dashed lines. The triangles
placed on the graph in the intervention phase show the amount of time available to them to
complete the task for that session, also in relation to the right hand y-axis tracking time in
minutes. As demonstrated by the graph, when compared to baseline, all participants
demonstrated significantly greater accuracy by the end of the intervention phase and in the
maintenance phase.
In addition, all participants also demonstrated a significantly decreased latency by the end of
the intervention phase and the maintenance phase compared to the baseline phase. These trends
in accuracy and latency across all 3 participants persisted with high levels of consistency from
the intervention phase to the maintenance phase. Comparing accuracy bar graphs side-by-side, in
all but one occurrence, there was an increase in accuracy in the post-intervention probe
compared to baseline. In the one occurrence where this did not happen, Participant 3 had exactly
the same accuracy as they did when they attempted the same task during baseline.
For Participant 1, accuracy increased from a baseline average of 69.74% to an average
accuracy by the end of the intervention phase of 85.19%, a trend which persisted into the
maintenance phase with an average accuracy of 88.22%. Latency also showed significant
improvement, with an average latency of 15 minutes 28.5 seconds (928.5 seconds) in baseline,
which was reduced to an average of 12 minutes 49 seconds (769 seconds) by the end of the
intervention phase, a reduction of 17%. The trend continued in the maintenance phase, with an
average latency of 11 minutes 31.9 seconds (691.9 seconds).

13

For Participant 2 accuracy increased from a baseline average of 64.55% to an average
accuracy by the end of the intervention phase of 73.44%, a trend which persisted into the
maintenance phase with an average accuracy of 85.24%. Latency also showed significant
improvement, with an average latency of 18 minutes 53.4 seconds (1133.4 seconds) in baseline,
which was reduced to an average of 13 minutes 37.2 seconds (817.2 seconds) by the end of the
intervention phase, a significant reduction in latency by 27.92%. The latency stayed steady in the
maintenance phase with an average latency of 13 minutes 58.9 seconds (691.9 seconds).
For Participant 3 accuracy increased from a baseline average of 70.52% to an average
accuracy by the end of the intervention phase of 79%, a trend which persisted into the
maintenance phase with an average accuracy of 82.37%. Latency showed significant
improvement as well, with an average latency of 18 minutes 53 seconds (1133 seconds) in
baseline, which was reduced to an average of 14 minutes 35.5 seconds (875.5 seconds) by the
end of the intervention phase, another significant reduction in latency by 22.72%.
The results of the accuracy between baseline and maintenance need to be understood in the
context of comparing baseline to maintenance. Figure 2 compares both the accuracy and the
latency from the first two sessions of the baseline phase with the two sessions of the maintenance
phase, given that the worksheet and task presented in session 1 and two of the baseline phase
were the exact same ones presented to them in session 1 and 2 of the maintenance phase. The
black bars represent either accuracy or latency of the task attempted in baseline, with the grey
bars representing the accuracy or latency of the exact same task attempted during the
maintenance phase. Comparing the latency graphs, each time the participants attempted the same
task presented in baseline, participants completed the task with a significantly shorter latency.
And when comparing accuracy, when participants attempted the same task after the intervention

14

phase increased in all but one instance, where the accuracy for Participant 3 in both baseline
session 1 and maintenance session 1 remained the same.
The results also indicate that changing criterion also had a positive effect upon the
behavior of all participants as well. The gradual change in the allowed latency to finish the task
presented seems to result in a gradual decrease in the amount of time needed to complete the task
presented accurately. The mini-reversals of latency for Participant 1 and Participant 3 yielded
mixed results, with Participant 1 responding to the increased time allowed with an increase in
latency, and Participant 3’s latency actually decreasing in the session with the increased time
allowed.
Furthermore, the results suggest that the intervention as applied has some degree of
maintenance, given that the latency of all but one of the post-intervention trials for all
participants in the post-intervention probe stayed within the range during the intervention phase,
as well as, significantly below the latency range demonstrated by participants in baseline. The
accuracy of all participants in the maintenance phase not only either stayed in the range of the
intervention phase or continued the trend of the intervention phase, they were also consistently
higher than the accuracy presented in the baseline phase, with the only exception being
Participant 3’s first maintenance probe. Her accuracy was exactly the same as before, but the
latency was nearly 4 minutes shorter compared to the first time.
Discussion
Given our current lack of research on how to develop a methodology for teaching skill
fluency and accuracy simultaneously, research like this can serve as a first step to developing and
discovering new knowledge and developing new technologies in an area of our discipline that
has the capacity of be widely practiced in applied situations, and has the capacity to push the

15

general applicability of our field to situations previously thought to be less applicable.
Developing a simple and applicable methodology to increase fluency and accuracy
simultaneously can be applied to a wide variety of settings. Such settings include more
traditional ABA clinics and educational settings, but also includes less traditional settings such as
Organizational Behavior Management (OBM) settings, occupational training settings,
competitive sports and athletics, and other environments where it would be desirable to teach
both accuracy and fluency simultaneously, particularly in high-stress situations.
Given the current worldwide changes that have been forced as a result of the COVID-19
pandemic, the applicability of the results here are even more apparent. A large majority of the
education for K-12 students has been forced online given the need for social distancing during
this pandemic, which has brought up questions about the efficacy of learning online compared to
learning in person. Methodologies like this one show that online learning does not have to be a
barrier to learning and achieving skill fluency. While many more research studies would need to
be conducted and evaluated before the we can say that we have an effective methodology for
addressing some of the concerns regarding the quality of learning online compared to in person,
this is a first step in addressing these gaps in knowledge that need now more than before to have
answers.
The results of the current study suggest there was a demonstrable and significant
improvement in the performance from all participants. The results also show that a relationship
between performance accuracy and latency was significantly weakened as a result of the
intervention. Specifically, all of the participants’ performances drifted further apart during the
intervention when compared to baseline, and stayed there during the post-intervention probes as
well. In Precision Teaching, the demonstration of accuracy increasing as latency decreased is

16

known as the “jaws of effect,” because the data points being tracked open up and form what
looks like a “jaw.” Importantly this demonstrates that the correlation between the participant’s
accuracy and the participant’s latency has been weakened. This is because as a result of the
intervention, they no longer need as much time or needed to put in as much effort as they did
before the intervention to perform as accurately.
There were also some other advantages to being able to run sessions via telehealth. The
use of an internet connection and telephone mean that the researcher and participant did not have
to be in the same room, or even in the same city, to work together and affect behavior change.
This highlights one of the main advantages of THS, that distance and geography are no longer
barriers to treatment or participation. This makes it much easier to plan for and provide sessions
and allows for more flexibility as needed when scheduling. THS also made it easier to record and
collect not just data, but a form of permanent product (by means of the worksheets worked on by
the participants) which allows for more reliable data collection and calculation, increased
accountability on the part of the researcher, and a more flexible means of having the data needed
for conducting and calculating IOA data. Finally, the use of THS means that even natural
disasters like a pandemic are less of a barrier to access to services, given that the protocol
allowed for sessions to be conducted without the need to be in the same room, there was a large
reduction in the possibility of becoming exposed to COVID-19.
In addition, practicing in the naturalistic environment would allow for greater ease in
generalizing the skill being taught, by having the environment the participants learn the skill be
the same as where they would be implementing it. As children get older and progress to middle
and high school, they will be working on papers and other longer writing assignments on
computers and laptops, which provides perfect opportunities to implement protocols similar to

17

the one described above to improve their writing abilities. This would allow for students to not
only be more comfortable writing on a computer, it would also facilitate becoming more skillful
with their preferred writing processor (such as Microsoft Word, Google Docs, etc). It also could
be a unique and potentially very effective form of intervention to use the writing provided by
children in their assignments as a means of improving skill fluency, which could be a future
direction for research. Other future directions for research include taking this methodology and
applying it to different contexts, or using the protocol above to target different skills, academic
or otherwise.
This study and the results demonstrated some limitations. Part of the reason for the use of
THS was because of the COVID-19 pandemic, which did allow for sessions to be run without
needing to be in the same room with the participant. However, it remains to be seen if this
methodology implemented would be more effective when done in person compared to being
done over THS, which is a common reservation and concern for using THS as opposed to in-
person services. While the relatively small sample size means that replicability is much easier, its
general applicability remains somewhat limited without other studies that can replicate the
results. Future replications or expansions on this study would want a much larger sample size to
increase the external validity of the results. Also, the general similarity in ages between the
participants leaves the efficacy of this methodology with other age groups an open question. It
also remains to be seen if the procedures could be effectively replicated in person. Additionally,
the efficacy of this methodology remains to be seen for participants that are still in skill
acquisition or learning the basics of a new skill, instead of taking a skill already at a base level of
mastery and bringing it to better levels of fluency.

18

A potential limitation to note in this study is the use of materials more than once with the
first two sessions of baseline and the two post-intervention probes. Given that the exact same
sheet was used across these two phases for the first two sessions could give rise to a concern over
the practice effect being the cause in the fluency increase. The protocol for this research however
makes this unlikely, due to the length of time of the intervention phase, and the 2 week gap
between the end of the intervention phase and the beginning of the post-intervention probes. This
means that the shortest time possible for seeing the same worksheet would be a minimum of 9
weeks, a time gap of about two months, as was experienced by Participant 1. For Participant 2
and 3, because their baseline and intervention phase was longer than Participant 1’s, this gap in
exposure was extended out to 12 weeks and 14 weeks respectively, and still performed as
described above in the results section. This gap in time from the protocol makes the practice
effect in this case highly unlikely to explain the consistent results gathered, but the potential
limitation does need to be noted.
Another weakness of the study described was the small number of sessions run in the
maintenance phase. To gain a better understanding of how well this methodology maintains after
a period of no intervention, extending the number of sessions for the maintenance phase to 3 or 4
total sessions would give a better demonstration of the efficacy of the intervention after its
application. Periodic maintenance probes or extending the amount of time from the end of the
intervention to maintenance would also contribute to evaluating the efficacy of this intervention.
The attempts at using mini-reversals of latency also showed mixed results, so no clear
conclusions can be drawn from them. One participant responded to the reversal, while the other
did not, meaning just how much of the behavioral control can be established through latency
remains to be seen. Further research in the future would likely be able to answer this question.

19

Furthermore, it remains to be seen if this application of ABA is only effective for those
who already have some level of skill mastery, or if it would also be effect for learning a novel
skill without any prior learning history. All of the participants in this study went through and
graduated from the American education system with a high school degree, so the amount of prior
learning history is substantial. Because of this, future research in teaching skills with fluency in
mind while the learner is still involved in skill acquisition using this methodology would be a
logical next step in expanding the external validity and applicability of the methodology
described.
Future directions for furthering this research and evaluating the usefulness of this
hypothetical methodology could go in a couple different directions. The first proposed direction
would be to attempt to replicate the results in person, instead of applying them over the internet.
Expanding the number of participants involved would increase the external validity of the
methodology implemented here, as well as demonstrating replicability. Along with this,
attempting the same methodology a different age range of participants, or with participants who
are much more moderately or severely impacted by certain disabilities such as ASD or IDD than
was attempted here would also increase the external validity as well.
The second proposed direction would be choosing another skill or kind of task to teach
(such as a self-care task). Being able to apply this methodology to another kind of task instead of
an academic one would also greatly improve upon this design as well, showing that this approach
not only has external validity, but also flexibility in the kinds of skills that can be taught or
addressed.
Future research can also include a more systematic approach to the use and
implementation of mini-reversals used in this research. Given that Participant 1 did respond to

20

the mini-reversal, Participant 3 did not, and that Participant 2 did not undergo a mini-reversal at
all, no results or conclusion can be drawn from this portion of the data. A more systematic
approach would likely generate clearer results, and by extension a clearer conclusion on the
matter.
One last suggestion for future directions for research revolves around implementing
differential reinforcement using this methodology while the participant is still in skill acquisition.
Given that skill fluency combines high degrees of accuracy with low rates of latency,
differentially reinforcing lower latency rates over higher latency rates across multiple
participants all working on the same skill or task chain could show promise if successfully
implemented. All these modifications and expansions would not only strengthen and clarify the
external validity of the intervention, but also build up a body of knowledge that can be applied
by others in various kinds of situations and practices and offer a small improvement to how we
practice ABA as a field overall.

21

References
Binder C. (1996). Behavioral fluency: Evolution of a new paradigm. The Behavior Analyst,
19(2), 163–197. https://doi.org/10.1007/BF03393163
Gill, F. (2016). The Role of Positive and Negative Reinforcement in Motivating Student
Learning. ASIAN JOURNAL OF MULTIDISCIPLINARY STUDIES, 4(8). Retrieved from
http://www.ajms.co.in/sites/ajms2015/index.php/ajms/article/view/2001, 4(8).
Heinicke, M. R., Carr, J. E., Leblanc, L. A., & Severtson, J. M. (2010). On the use of fluency
training in the behavioral treatment of autism: a commentary. The Behavior analyst,
33(2), 223–229. https://doi.org/10.1007/bf03392221
Ingersoll, B., Shannon, K., Berger, N., Pickard, K., & Holtz, B. (2017). Self-Directed Telehealth
Parent-Mediated Intervention for Children With Autism Spectrum Disorder: Examination
of the Potential Reach and Utilization in Community Settings. Journal of Medical
Internet Research, 19(7). doi: 10.2196/jmir.7484
Kubina, R. M. (2019). The precision teaching implementation manual. Pittsburgh, PA: Greatness
Achieved Publishing Company.
Kubina, R. M., Morrison, R., & Lee, D. L. (2002). Benefits of adding precision teaching to
behavioral interventions for students with autism. Behavioral Interventions, 17(4), 233-
246. doi:10.1002/bin.122
Lindsley, O. R. (1992). Precision teaching: Discoveries and effects. Journal of Applied Behavior
Analysis, 25(1), 51-57. doi:10.1901/jaba.1992.25-51
McMahon, T., & Thakore, H. (2008, April). TEACHING ONLINE - A RESEARCH-BASED
GUIDE FOR ACADEMICS. Retrieved from https://files.eric.ed.gov/fulltext/EJ870183.pdf

22

Myers K, Comer JS. The case for telemental health for improving the accessibility and quality of
children’s mental health services. J Child Adolesc Psychopharmacol. 2016;26(3):186–
191. doi: 10.1089/cap.2015.0055.
Pennefather, J., Hieneman, M., Raulston, T. J., & Caraway, N. (2018). Evaluation of an online
training program to improve family routines, parental well-being, and the behavior of
children with autism. Research in Autism Spectrum Disorders, 54, 21–26. doi:
10.1016/j.rasd.2018.06.006
Ramey, D., Lydon, S., Healy, O., McCoy, A., Holloway, J., & Mulhern, T. (2016). A systematic
review of the effectiveness of precision teaching for individuals with developmental
disabilities. Review Journal of Autism and Developmental Disorders, 3(3), 179-195.
doi:10.1007/s40489-016-0075-z
Tomlinson, S. R. L., Gore, N., & Mcgill, P. (2018). Training Individuals to Implement Applied
Behavior Analytic Procedures via Telehealth: A Systematic Review of the Literature.
Journal of Behavioral Education, 27(2), 172–222. doi: 10.1007/s10864-018-9292-0
Weiss, M. J., Pearson, N., Foley, K., & Pahl, S. (2010). The importance of fluency outcomes in
learners with autism. The Behavior Analyst Today, 11(4), 245-252. doi:10.1037/h0100704

23

Figure 1

Multiple Baseline Graph Tracking Changes in Accuracy and Latency

24

Figure 2
Graphs Comparing Accuracy and Latency of Session 1 & 2 From Baseline to Maintenance

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A preliminary evaluation of a telehealth approach to acceptance and commitment training (ACT) for enhancing behavioral parent training (BPT) for Chinese parents

PDF

Behavioral coaching for athlete health behaviors

PDF

Use of a clinical decision model to support the transition of services for individuals with autism spectrum disorder

PDF

Using Acceptance and Commitment Training and a virtual patient model for supporting families with children with autism spectrum disorder

PDF

Using acceptance and commitment training to enhance the effectiveness of behavioral skills training

PDF

Let’s do something different: evaluation of lag schedules of reinforcement to increase flexibility and promote generalization in children with autism

PDF

What “works” and why it still isn't enough: sexual harassment training measures of effectiveness

PDF

The effects of dual coding theory on memory retention

PDF

Training staff to implement the interview-informed synthesized contingency analysis (IISCA)

PDF

The effects of acceptance and commitment therapy-based exercises on eating behaviors in a laboratory setting

PDF

The use and perceptions of experimental analysis in assessing problem behavior of public education students: the interview - informed synthesized contingency analysis (IISCA)

PDF

Effects of preferred physical activity on stereotypical behaviors in children with autism spectrum disorder: adapting from in-person to telehealth

PDF

Evaluating the effects of virtual reality and acceptance and commitment therapy on music performance anxiety

PDF

Evaluating the effectiveness of interview-informed synthesized contingency analysis for survivors of traumatic brain injury

PDF

The effects of bilingual acceptance and commitment training (ACT) on exercise in bilingual international university students

PDF

Values and persistence

PDF

Metacognitive experiences in judgments of truth and risk

PDF

Clinical decision analysis for effective supervision: a review of the supervisor-supervisee relationship

PDF

Meta-accuracy as a moderator of the association between social experience and emotional adjustment during childhood

PDF

Behavior analysis in medicine: a clinical decision model to support patient care

Asset Metadata

Creator Hall, Andrew George (author)

Core Title Using rate-building to decrease latency and increase accuracy for task completion via telehealth

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Applied Behavior Analysis

Degree Conferral Date 2021-08

Publication Date 07/18/2021

Defense Date 07/16/2021

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

Tag accuracy,behavioral fluency,education,latency,OAI-PMH Harvest,rate-building,Reading,telehealth,Writing

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Tarbox, Jonathan (committee chair), Aclan, Megan (committee member), Manis, Frank (committee member)

Creator Email aghall@usc.edu,andyghall92@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15610514

Unique identifier UC15610514

Legacy Identifier etd-HallAndrew-9787

Document Type Thesis

Format application/pdf (imt)

Rights Hall, Andrew George

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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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

Repository Email cisadmin@lib.usc.edu