University of Southern California Dissertations and Theses

The Effect Of Conditions Of Risk, Internal Versus External Control Of Reinforcement, And Sex On Binary Choice Probability Learning

(USC Thesis Other)

The Effect Of Conditions Of Risk, Internal Versus External Control Of Reinforcement, And Sex On Binary Choice Probability Learning

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content This dissertation has been microfilmed exactly aB received 70-339 ARONOWITZ, Marion, 1938- THE EFFECT OF CONDITIONS OF RISK, INTERNAL VERSUS EXTERNAL CONTROL OF REINFORCEMENT, AND SEX ON BINARY CHOICE PROBABILITY LEARNING. U niversity of Southern California, Ph.D., 1969 Psychology, clinical University Microfilms, Inc., Ann Arbor, Michigan THE EFFECT OF CONDITIONS OF RISK, INTERNAL VERSUS EXTERNAL CONTROL OF REINFORCEMENT, AND SEX ON BINARY CHOICE PROBABILITY LEARNING by Marion Aronowitz A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Psychology) June 1969 UNIVERSITY O F S O U T H E R N CALIFO RNIA TH E GRADUATE SC H O O L UNIVERSITY PARK LOS A N G ELES. CA LIFO RN IA 9 0 0 0 7 This dissertation, written by Marion Aronowitz under the direction of h&x.... Dissertation Com mittee, and approved by all its members, has been presented to and accepted by The Gradu ate School, in partial fulfillment of require ments for the degree of D O C T O R OF P H IL O S O P H Y 'Tn r Dean Date— JUNE 1969 DISSERTATION COMMITTEE .0. ___k . . s T . b . £. Chairman ACKNOWLEDGMENT S It is with genuine warmth and deep appreciation that I offer my thanks to Dr. Norman Tiber, chairman of the dis sertation committee. His guidance provided me with intel lectual stimulation and support while encouraging independ ent thinking and exploration of my own ideas. I also wish to express my appreciation to Dr. Henry Slucki and Dr. Murray Wexler for their suggestions and interest in the re search. Construction of the apparatus was funded by a grant from the Department of Psychiatry at the University of Southern California Medical Center. To Dr. Seymour Levitan I owe a special thanks. As supervisor, member of the committee, and friend, he has always known how to reach out and offer reassurance when I most needed it. Finally, I have a special sense of in debtedness to my good friend, Dr. Joseph White. Without his enthusiasm and encouragement, my graduate career would not have begun. TABLE OF CONTENTS Page ACKNOWLEDGMENTS................................... ii LIST OF T A B L E S ........................................ v Chapter I. INTRODUCTION .............. 1 II. REVIEW OF THE LITERATURE..................... 3 Non-Risk Taking Probability Learning Risk Taking Probability Learning Models of Choice Behavior Background Internal vs. External Control of Reinforcement Further Theoretical Considerations III. M E T H O D...................... 31 Design Hypotheses Subjects Apparatus Procedure Instructions Experimental Trials Post-Trial Interview IV. RESULTS........................................ 41 Predictive Behavior Estimating Behavior Chapter Page V. DISCUSSION.................................... 65 Predictive Behavior Estimating Behavior Future Research VI. SUMMARY...................................... 79 APPENDIX A -- INTERNAL-EXTERNAL QUESTIONNAIRE ..... 82 APPENDIX B--LETTER SENT TO SUBJECTS................. 88 APPENDIX C--RAW DATA.................................. 90 APPENDIX D--POST EXPERIMENTAL INTERVIEWS............. 96 LIST OF REFERENCES .'.....................................110 LIST OF TABLES Table Page 1. Summary of Analysis of Variance of Number of Predictions of the More Frequently Occurring Events .......................... 43 2. Means of the Experimental Risk Taking Conditions and Percentage of Total Predictions.............................. 44 3. Pairs of Means Which Differ Significantly Using Tukey’s Method Shown According to Condition of R i s k ........................ 45 4. Number of Subjects Who Tried to Predict Correctly Vs. Number Predicting Incor rectly Shown by Sex and Personality Type . 46 5. Exact Probabilities of Number of Subjects Who Tried to Predict Correctly Vs. Number Who Tried to Predict Incorrectly Shown by Pairs....................... 47 6. Rank Order of Prediction of the More Fre quently Occurring Event ................. 49 7. Means for the Internal-External and Sex Variables................................. 50 8. Means of the Predictions of the More Fre quently Occurring Event Shown According to Condition, Sex, and I-E................. 52 9. Pairs of Means Which Differ Significantly Using Tukey’s Method Shown by Condition, Sex, and I-E.............................. 53 10. All Pairwise Comparisons of Means for the the Reverse Condition Using Tukey's Method .............................. 54 Table Page 11. Summary of Analysis of Variance of Number of Predictions of the More Frequently Occur ring Event by I-E, Sex and High-Low Rein forcement Levels.......................... 56 12. Analysis of Variance of Estimates of the Percentage of Time The More Frequent Event Occurred............. 57 13. Means of Estimates of Percentage of Occur rence of the More Frequent Event Shown by Condition............................... 59 14. Rank Order of Predictions and Estimates Shown by Experimental Condition ......... 60 15. Means of Estimates of Percentage of Occur rence of the More Frequent Event Shown by Condition and Sex...................... 61 16. Significant Pairwise Comparisons of-Means of Estimates Using Tukey's Method Shown by Condition and Sex..................... 62 17. Analysis of Variance of Estimates of the Percentage of Time the More Frequent Event Occurred. Shown by I-E, Sex and High Versus Low Reinforcement Levels . . . 63 18. Raw Data Showing I-E Score, Number of Times Red Light Predicted, and Estimates of Percentage of Time Red Light Actually Occurred................................... 91 19. Post Experimental Interview Subjects' Comments on Their Behavior ............... 97 vi CHAPTER I INTRODUCTION Human decision making behavior is an area of inves tigation that has actively interested psychologists in the last fifteen years. One approach to this enormously complex subject has been to examine probability learning, which is considered to be a relatively simple case of dynamic decision making (Edwards, 196i). Typically, in probability learning situa tions, an individual is asked to repeatedly predict which of several possible events will occur next. Each event has a fixed probability of occurrence, appears randomly in the sequence, and is not contingent upon the individual's choice. Sometimes, the risk taking aspects of the situa tion are intensified by having the subject bet on his pre dictions . Unfortunately, despite the extensive experimental attention paid to this area, some of the most fundamental comparisons have not been made. The present investigation was designed to collect the kind of basic information 1 2 essential to the development of a systematic body of knowl edge about probability learning and risk taking behavior. It represents the first time that a study provided direct comparison of all the usual risk taking conditions using the same probability learning task and the same level of risk. The second area of concern of this investigation was the effect of personality variables upon probability learning behavior. This seems like promising ground and some have made a plea for explicitly incorporating person ality variables into theories of risk taking behavior (Scodel, I960). Thus far, however, relatively few studies have addressed themselves to the issue. The personality dimension chosen for the present investigation was Internal vs. External Control of Reinforcement (Rotter, 1966). This refers to an individual’s belief in the amount of personal control he exercises over the events in his environment. It would be expected that people who tend to perceive the locus of control over reinforcement as residing within themselves would behave differently in risk taking situa tions from people who feel that reinforcement is determined by forces external to themselves. Since sex differences relate to the Internal-External dimension, the relationship between sex of subjects and probability learning was also investigated. CHAPTER II REVIEW OF THE LITERATURE Non-Risk Taking Probability Learning In a standard non-risk (no payoff) probability learning situation, there are usually two possible events. The subject is asked to predict which of them will occur next, over a large number of trials. The events are pro grammed to occur randomly and they are not contingent upon the individual's responses. Instructions are usually such that if a given event were predicted and then occurred, the subject interprets that occurrence to mean his response was correct (Estes, 1957). The response itself may be verbal or a simple motor act such as pressing a button or marking an answer sheet. In this connection, Restle (1961) comments that no regular differences resulting from the characteris tics of the events predicted or the form of the response have been found, provided the events were highly discrimin- able. In the most commonly used task, the individual pre dicts which of two lights will light up next. When subjects simply are asked to predict events, 3 4 and no other variations are introduced into the situation, they tend to probability match. That is, they learn that one event occurs more frequently than the other and the rate at which they predict that event tends to approximate its objective frequency in the sequence -(Humphrey 1939). On each trial, subjects were asked to indicate whether or not a single bulb would light. They were tested where the proportion of "lights on" was 1.0, 0.5 and 0. The results showed that the proportion of guesses of "lights on" ap proached asymptotes of 1.0, 0.5 and 0, respectively. Hake and Hyman (1953) used random sequences of events where the probabilities were .50 and .75. Their groups tended to wards asymptotes of .50 and .75, respectively. Jarvik (1951) used lists made up of two works and had subjects predict which word would occur next. The most frequent event occurred 60%, 67% or 75% of the time. He found that at all three levels subjects tended to predict at rates that matched the objective probabilities. Similar results have been summarized by Grant, Hake and Hornseth (1951), Rapoport (1960), Edwards (1961b), Abelson (1964), Seigel and Goldstein (1964) and Green (1966). Experimenters have introduced many variations into these situations. Hake and Hyman (1953) used restricted probabilities. For example, the probability that B follows A is 0.5; the probability that B follows B is 0. As in the 5 standard non-risk probability learning situation, the groups tended towards asymptotes equal to the proportion of that event in the sequence. Wolin ert aJ. (1965) also used only partially random binary choice sequences. Probability matching occurred in the condition having their simplest contingency. On each even numbered trial the occurrence of event A was certain. On each odd numbered trial the proba bility of event A was .5. Thus, A had an overall probabil ity of .75 on any trial. As long as the contingencies were not very complex, probability matching occurred. Green (1966) used a three-choice situation in which subjects had to predict whether the next card would have a triangle, rectangle or dot drawn upon it. The events were in a 60:30:10 proportion. He again found that subjects tended to match the probabilities of each corresponding stimulus. Jones and Myers (1966) randomized the events in short blocks of 20 trials and long blocks of 300 trials. Using .60 and .75 as the probabilities of the more frequent events, they found a great tendency to probability match where the events had been randomized in short blocks. Under the long block condition, the trend was towards max imizing predictions of the more frequent event. Goodnow and Postman (1955) speculated that perhaps subjects in most: probability learning situations approach the task with a "probabilistic set.” That is, they believe they should 6 attempt to discriminate probabilities as an aid to making predictions. They questioned whether subjects would behave differently if they were told to seek a lawful solution. They found no difference in the probability matching of the two instruction groups. More recently, however, Erickson et al. (1968) found that subjects given problem solving in structions were closer to probability matching than sub jects who were told that the events were random. When subjects are asked to view a display of events and then estimate the proportion of one type of event, their estimates are within a few percentage points of the true values. This is true if the subjects see all the events at once, such as blue dots randomly scattered among green dots (Stevens and Galanter, 1957) or if they give an estimate after participating in a binary choice prediction situation (Tversky and Edwards, 1966). Thus, there seems to be no doubt that in a variety of non-risk probability learning situations subjects tend to predict the more frequent event at a rate close to its actual rate of occurrence. At first, regardless of the actual event structure, they tend to predict each with equal frequency. This changes gradually until something close to probability matching is reached at asymptote (Grant, Hake, Hornseth, 1951; Tune, 1961). Detailed de scriptions of typical probability learning curves have been 7 summarized by Edwards (1961b) and Tune (1961). Risk Taking Probability Learning In 1947 von Neumann and Morgenstern proposed their game theoretic model to predict human choice behavior. In essence, they posited a “rational man” who would act in a deterministic fashion. This meant that in a probability learning situation, an individual would learn to maximize the expected frequency of correct predictions by selecting the more frequent event on all trials. ' The results of the non-risk probability studies re viewed above do not support von Neumann and Morgenstern*s “maximizing” prediction. When, however, conditions of risk are introduced into the probability learning situation, then subjects do show a tendency towards maximizing. Compared to the examination of non-risk probability learning situations, relatively few investigators have studied probability learning under risk taking conditions. To date, three different risk taking conditions have been explored. These are: 1) Win-Lose--the subject wins some thing for a correct prediction and loses something for an incorrect prediction, 2) Win or break even--the subject wins something for each correct prediction and stands (does not lose anything) on incorrect predictions, 3) Lose or break even--the subject stands on a correct prediction and loses on incorrect predictions. A non-risk (no payoff) 8 condition is frequently employed as a base measure. Thus, four conditions have been explored by previous investiga tors: No payoff, Win-Lose, Win or Break Even, Lose or Break Even. The only study comparing as many as three condi tions in a probability learning situation is that of Seigel and Goldstein (1964). They studied the no payoff, win or •lose and win or break even conditions in a typical binary choice probability learning situation. Subjects had to predict which of two lights would flash on each trial. The lights were randomly programmed in a 75:25 proportion. In the payoff conditions subjects were given 75£ and stood to win or lose 5£ per trial. The following table shows the proportion of predictions of the more frequent event (des- igated as TT> under the three conditions. Trial number No Payoff Win-Lose Win or stand 0-100 -69 .95 .78 n = .75 101-200 .74 .95 .85 201-300 .75 .95 .86 Very clearly, probability matching occurs only in the no payoff condition. When subjects can win and lose, they tend to maximize their chances of being correct by predicting the more frequent event on most trials. Where they can either win or break even, they overshoot the 9 actual tt level but do not maximize quite as much as those in the win-lose condition. Tversky and Edwards (1966> told all of their sub jects that the sequence of events was.random and there was no pattern. They then told half of the subjects that the probabilities for each event would remain stationary throughout the experiment. The other half was not told the truth and was informed that the probabilities would be con tinuously changing. Subjects could win or lose 5£ per trial. They were tested where the more frequent event oc curred either 60% or 70% of the time. At both levels, sub jects overpredicted the frequent event when they believed the probabilities were stationary and predicted very slightly under the actual probabilities when they thought the probabilities would change. The overshooting repre sents the same tendency towards maximizing as found in the Seigel and Goldstein study, but to a lesser degree. Some contradictory evidence is found in the work of Jones and Myers (1966>. They obtained no significant dif ferences in the behavior of subjects under win-lose condi tions involving no money, five cents, and twenty-five cents. The authors make no attempt to explain their con tradictory results. To date, no investigator has studied the Lose or Break Even condition in a traditional probability learning 10 situation. The most related work is a gambling study by Edwards. He compared gambling decisions in a Win or Break Even condition. His results indicate that subjects tend to overestimate their chances of winning money but give fairly objective estimates of their chances of losing it. Obvi ously, this is not directly comparable to the simple two choice situation used by Seigel and Goldstein. A further complication is Edwards’ use of only five subjects. Keep ing these limitations in mind, the results are still sug gestive of a trend. The following table summarizes the re sults of both studies. No Payoff Win or Lose Win or Stand Lose or Stand Seigel & probabil- maximize overshoot Goldstein ity match. but less--------- ----- than maximize Edwards overshoot probability match or slight overshooting Clearly, the introduction of a risk taking compon ent in the form of chips or money alters the situation. Subjects then behave differently than when there is no pay off other than learning that one's prediction was correct. When there is the possibility of winning something, the tendency is toward the kind of strategy that von Neumann 11 and Morgenstern would consider optimal if one were behaving ’♦rationally.” In addition to the four conditions discussed above, there is a fifth condition which has not previously been investigated. This is a Reverse condition in which sub jects lose a chip or money for correct responses. Unfortunately, risk taking behavior in probability learning situations has never been systematically studied. Instead, many investigators have chosen to study risk taking in a wide variety of non-probability learning situ ations such as. judging the length of a line or engaging in competitive games (Julian and Katz, 1968). Since the ex perimental tasks have been limited only by the imagination of the researcher, what has resulted is a mass of experi mental data, much of it not tied to theory and most of it not directly comparable to other data accumulated in risk taking experiments. To date no one has investigated proba bility learning under all five conditions (No payoff, Win- Lose, Win or Break Even, Lose or Break Even and Reverse) with the same population. This represented the first pur pose of the present investigation. Models of Choice Behavior Background Many theorists have devoted their efforts to 12 building models of choice behavior (see Becker and McClintock, 1967; Edwards, 1961a; Estes, 1957; Seigel, 1964; and, von Neuman and Morgenstern, 1947). One such model was the game theoretic model of von Neuman and Morgenstern discussed earlier. In essence, they posited a "rational man" who would act in a deterministic fashion. This meant that in a probability learning situa tion, an individual would learn to maximize the expected frequency of correct predictions by selecting the more fre quent event on all trials. Implied in such an approach is the notion of transivity in choice situations. That is, if a person prefers A over B and B over C, then he prefers A over C. Von Neuman and Morgenstern are credited with stim ulating a great deal of theoretical and experimental work on behavior under conditions of uncertainty (Chipman, 1960). However, their game theoretic model has generally been re jected by psychologists. One reason is its failure to pre dict probability matching behavior under non-risk condi tions. Psychologists have turned instead to probabilistic models which retain the essence of transivity by requiring stochastic transivity. That is, if the probability of choosing A over B is greater than .5, and the probability of choosing B over C is greater than .5, then the probabil ity of choosing A over C must be greater than some minimum 13 value (Abelson, 1964). Using this probabilistic framework, theorists focus on two aspects of the situation. One is the person’s expectations about being correct. This in volves all his beliefs about the structure of the situation and the probability of being paid off. Obviously, such probabilities are subjective and may or may not approximate the objective probabilities. The other factor is the util ity or value of the payoff to the individual. Payoff, as used here, refers to any and all feedback which tells the subject whether or not his prediction was correct. Among such models, the most popular is the Subjec tively Expected Utility Model (SEU) (see Becker and McClintock, 1967; Edwards, 1961a; and Seigel, 1964). It predicts that individuals, using their own subjective prob abilities, will try to maximize the occurrence of events that have subjective value (utility) for them. Thus, al though it might seem more "rational” to predict the more frequent event in a binary choice situation, the person may find greater personal satisfaction in trying to predict the infrequent event, or may suffer from kinesthetic and/or cognitive boredom if he predicts the same event all the time, or may view this as a problem solving situation in which he is seeking a lawful solution (Seigel, 1964a; Wolin jet al., 1965). Becker and McClintock (1967) reject the idea that 14 the parameters of the SEU model can be satisfactorily meas ured. They argue that the most meaningful SEU model is one which treats utility as a random variable. It assumes that the utility associated with each outcome is itself a random variable. Fluctuations in this lead to fluctuations in the person's choices. This means the individual's perception of the outcome is very complex and varies with factors such as mood and set. Thus, the random utility model meets the challenge (see Abelson, 1964; Chipman, 1960; Rapoport, 1960) of allowing for the wide variety of psychological variables that affect human behavior, but it is too complex a model to be practical at this time. Becker and McClintock conclude that the principle contribution of these mathematical approaches . . . probably rests more in having established a common conceptual orientation and a common language for studying choice behavior than in providing fundamental laws of human behavior (1967). Another important model of choice behavior is Rotter’s Social Learning Theory (1954). He states that the potential for a specific behavior directed toward a rein forcement to occur is a function of the individual's expectancy of that reinforcement following the behavior in that situation and the value of the reinforcement in that situation. It should be noted that this notion is very similar to the SEU mathematical model of choice behavior discussed above, which postulates that a person will 15 maximize his subjectively expected utilities. The main difference between the two is that SEU theorists are seek ing precise numerical ways of measuring a person's subjec tive probabilities (expectancies) and his utilities. Rotter (1967) on the other hand, points out that a particu lar behavioral choice may involve several reinforcements which all effect the person's expectancies. It is imprac tical, he feels, to attempt to measure each one exactly and combine it with measurements of all the other reinforce ments to form a numerical statement of a person's subjec tive probability in a given situation. Reinforcement, according to Rotter, acts to strengthen an expectancy that a particular behavior or event will be followed by that reinforcement in the future. In a typical probability learning situation, the subject makes a prediction and then an event occurs. This occur rence of the event provides the subject with 1) information about the nature of the sequence, and 2) information about whether or not his prediction was correct. In Social Learning Theory terms, the event is a reinforcement which should strengthen his expectancies. It should be noted that reinforcement as used by Rotter includes everything that acts to strengthen an expectancy regardless of whether it is “informational” or "rewarding” in nature. Some the orists have made a distinction between the information 16 value and the reinforcement value of feedback (Tversky and Edwards, 1966>. However, it is just as feasible to con ceive of these two operations as two types of generalized reinforcers (Marston, 1967). Since they have not been suc cessfully experimentally separated, from a practical point of view it seems more useful to view all of the feedback as reinforcement. More importantly, as pointed out by Deacon (1968>, if one assumes that information reduces uncertainty and that this is a valued outcome, then information is re inforcing. Or, in Social Learning Theory language, it. is assumed that information strengthens expectancies and is therefore reinforcing. As previously discussed, trying to measure and mathematically combine an individual's specific expectan cies in a given situation is an enormously complex task. In Rotter's opinion it has shown itself to be an unprofit able approach to the problem. Instead, he urges the in vestigation of generalized beliefs or expectancies involv ing broad classes of situations. For example, Rotter the orized that if an individual believed a given situation or task to be mainly a matter of skill, his expectancies for reinforcement would differ markedly from his expectancies if he felt that same task was a matter of chance. He and his colleagues have done considerable research on this di mension of learning under skill vs. chance conditions (see 17 Lefcourt, 1966; Rotter, 1966). Although it is beyond the purpose of the present investigation to review all of this literature, some of the findings are of interest. Phares (1957), using an ambiguous color matching task, gave half the subjects instructions which led them to believe match ing was a matter of skill, and half instructions that it was a matter of chance. Both groups received a fixed order of partial reinforcement (right or wrong) with the measure of expectancy being the number of chips a subject would bet on his probability of being correct on that trial. He found that reinforcements under skill conditions had a greater effect on raising or lowering expectancies for future reinforcements. He also found that subjects shifted their expectancies more often under skill conditions. An other group of studies used two different tasks where the instructions and number of reinforcements were the same but the tasks were such that one task was likely to be seen as 3 chance determined and the other skill determined (see Rotter, Liverant and Crowne, 1961; Bennion, 1961; Blackman, 1962). The results indicate that subjects are less likely to see reinforcements as chance controlled when the per centage of reinforcement significantly deviates from 50-50 in a right-wrong situation, when the sequence of reinforce ments appears to have a pattern, and when unusually long sequences of one of two alternative, events occur (Rotter, 18 1966). It seems well established that task structure func tions as an important determinant of one's expectancies for reinforcement. In addition, there is now considerable evidence that the individual's own generalized beliefs or expectancies about control of reinforcement are very impor tant determinants of how he will behave. Given the same situation, individuals differ in the degree to which they would tend to attribute personal control to reward. Those who see a causal relationship between what they do and the reinforcements they receive believe in internal control of reinforcement. External control refers to the belief that events are frequently unrelated to one's own behaviors and therefore beyond personal control. It would be expected that people who have generalized expectancies for internal control of reinforcement would react to the reinforcements of some situations very differently than those who believe control is external. It follows as a general hypothesis that when the reinforcement is seen as not contingent upon the subject's own behavior that its occurrence will not increase an expectancy as much as when it is seen as contingent. Conversely, it's non-occurrence will not reduce an expectancy so much as when it is seen as contingent. (Rotter, 1966) Sufficient data has now been accumulated on the Internal vs. External Control of Reinforcement dimension to illustrate the value of this kind of variable in studying 19 the relationship between beliefs and observable behavior (Rotter, 1966; Lefcourt, 1966}. Internal vs. External Control of Reinforcement The first attempt to measure this variable was by Phares (1957}. Using twenty-six items selected on a priori grounds, he found that test items worded in an external di rection gave predictions approaching statistical signifi cance that people who had external attitudes would behave similarly to all subjects when placed in chance vs. skill situation. James (1957} revised the Phares scale and ob tained similar, but significant results. J. B. Rotter, M. Seeman and S. Liverant undertook to broaden the test and control for social desirability (Rotter, 1966}. The final version of the I-E Scale is a twenty-nine item, forced choice test with low scores representing the internal end of the dimension. Six filler items are included to make the purpose of the test more ambiguous. Test-retest reli ability is consistent and acceptable, varying between .49 and .83 for varying samples and time periods (Hersch and Scheibe, 1967}. Correlations with intelligence were negligible us ing a sample of college females (Strickland, 1962) and a sample of male prisoners (Eadwig, 1963}. Cardi (1962} ob tained similar results for male college students but found 20 a low correlation for females. The mean I-E scores for various college populations reported by Rotter (1966) are between 7.7 and 9.7, with fe males slightly, but consistently higher than males. Several samples of college students have been ex amined for the relationship between the I-E Scale and the Rotter Incomplete Sentences Blank (see Rotter, 1966). Lin ear correlations were not significant. Hersch and Scheibe (1967), in examining personality correlates in a college sample found many significant correlations between I-E scores and the scales of the California Psychological In ventory (CPI; Gough, 1964) and the Adjective Check List (ACL; Gough and Heilbrun, 1965). On the ACL, internal scorers were high on measures of Defensiveness, Achievement, Dominance, Endurance, and Order and lower on the scales re flecting Succorance and Abasement. On the CPI they were high on scales of Dominance, Tolerance, Good Impression, Sociability, Intellectual Efficiency, Achievement vs. Con formance and Well Being. The converse may be said to hold for externals. The authors note: Twenty-three adjectives were checked more often by the internal individual (p<.05) and present a fair ly coherent description of him, at least as he sees himself. The adjectives more frequently checked by internals were: clever, efficient, egotistical, enthusiastic, independent, self-confident, ambi tious, assertive, boastful, conceited, conscien tious, deliberate, persevering, clear-thinking, dependable, determined, hard-headed, industrious, ingenious, insightful, organized, reasonable and 21 stubborn. On the other hand, only one adjective was checked significantly more often by the exter nals --self -pi tying. The authors also comment that internals seem to be a more homogeneous group than externals. A detailed review of the construct validity of the scale is offered by Rotter (1966>. The notion that internals feel less powerless than externals in controlling themselves and their environment is supported by several studies. Seeman and Evans (1962) found that among patients in a tuberculosis hospital, in ternals knew more about their own condition, questioned doctors and nurses more about their illness and expressed less satisfaction at the amount of information they were getting about their condition from hospital personnel. Seeman (1963) also studied reformatory inmates to investi gate memory for various kinds of information to which they were exposed in an incidental fashion. As hypothesized, there was a significant correlation between internality- externality and the amount of information remembered about how the reformatory was run, parole, and economic facts relevant to the person's situation after release from the institution. Gore and Rotter (1963) studied southern Negro college students regarding civil rights activities. Those who were willing to participate in a march on the State Capitol or join a freedom rider's group were significantly 22 more internal than those who were only willing to attend a rally or those who were not interested in participating at all. Phares (1965) found that internal college students were more successful than externals in changing the atti tudes of fellow students regarding an important campus issue. The above studies are primarily concerned with controlling the environment. There also seems to be a re lationship between internality and the feeling that one can control one's self. Straits and Sechrest (1963) found that non-smokers were significantly more internal than smokers. This finding was replicated by James, Woodruff and Werner (1965). They also found that following the Surgeon General's report on cigarette smoking, those who quit and did not return to smoking were more internal than those who believed the report but did not quit smoking. The differ ence was not significant for females. Lefcourt (1966) points out that many of the I-E studies have used only males. Where females have been used, the dimension has not seemed as effective in predicting behavior (see Crandall et al., 1962). He suggests that more investigations using sex as a variable are necessary. Of more direct interest is a study by Marston (1964) in which subjects were raeas~ ured on I-E and then learned a set of verbal discrimina tions to criteria while being reinforced with a green light: 23 for correct responses. Following this, the subject was given control of the reinforcing light and told to press the switch whenever he was confident he had answered cor rectly. It was found that internals tended to increase in frequency of correct responses over trials. Thus, inter nals not only maintained, but actually improved their per formance when experimenter administered reinforcement was removed. Externals, in sharp contrast, showed a deteriora tion in performance when external reinforcement was reduced and they had only themselves to rely upon for reinforce ment. To date, there have been no investigations which have evaluated the variable of I-E in a probability learn ing situation. The most closely related studies concern I-E and risk taking. James (1957) found that externals tend to make more unusual shifts in expectancies. That is, they increase their expectancies following failure and de crease their expectancies of being correct following suc cess (the gambler's fallacy). Rotter (1966) comments that several investigations have found this difference to be significant or near significant. The trend is always in the same direction with externals tending to produce more unusual shifts. Eiverant and Scodel (1960) hypothesized that internals would believe they could exert some control in a chance situation while externals would see the out- comes as occurring randomly. Subjects had to choose the 24 amounts to bet as well as one of seven possible outcomes in thirty rounds of dice throwing. They predicted that inter nals would pick more high probability, low payoff bets than externals. The findings showed that internals chose sig nificantly more intermediate probability and significantly fewer low probability bets than did the externals. Also, more internals never picked an extreme high or low proba bility bet and they wagered more money on cautious than risky bets. Thus, internals showed greater self-regulatory tendencies with regard to the objective probabilities. Baron (1968) found a low but significant correlation be tween I-E and conservatism in a risk taking situation, with internals being less conservative. To the degree that they are comparable, these findings do not agree with those of Liverant and Scodel. Julian and Katz (1968) placed sub jects in a competitive word game where the individual could earn points by relying on his own judgment, or relying on his opponent’s judgment at no cost to himself. Feedback was controlled so that the opponent appeared to be doing better. It was found that internals had a significantly greater preference for relying upon themselves rather than their more competent opponent. The same tendency was found when subjects were given a number task and instructions that the task was basically a matter of chance. The primary purpose of the present study was the 25 investigation of the effect of various conditions of risk on probability learning. The second purpose was to examine the effect of personality on probability learning. Since Internal vs. External Control of Reinforcement has been ex tensively studied and has shown itself to be a useful di mension, it was chosen as the personality variable for this investigation. Finally, because experimental evidence seems to suggest that sex differences may relate to I-E, that variable was included. Thus, the third major purpose was to study the relationship of sex of the subject to proba bility learning. The third variable, sex, has never before been di rectly examined in a probability learning study. One would expect differences, especially where risk taking is in volved, since typically, females seem to have less exposure to, and less interest in, both mathematics and gambling. Kogan and Wallach (1964) did a massive correlational, study on risk taking in which subjects were measured on five personality variables and tested in seven different, newly created, risk taking situations. The study is fraught with logical errors which do not permit valid inferences to be drawn (Becker and McClintock, 1967). It is still, however, of interest because there were sex differences in risk taking behavior throughout the study. No general statement can be made about these differences except that they were not consistently in any one direction. Nonetheless, the results provide some support for a hypothesis that men and women will perform differently in probability learning sit uations. Further Theoretical Considerations The present study then was directed at investigat ing binary choice probability learning under the five ex perimental conditions of: No Payoff, Win or Lose, Win or Stand, Lose or Stand and Reverse (lose for a correct guess and stand on an incorrect guess). In addition, the effects of sex differences and Internal vs. External Control of Reinforcement on probability learning were examined. An extension of Rotter’s Social Learning Theory (1954> provided a useful conceptual system. It accounted for the experimental evidence presented earlier on risk taking in probability learning situations (Seigel and Goldstein, 1964; Edwards, 1955>, and allowed for some logi cal predictions in the present investigation. Social Learning Theory regards a reinforcement as acting to strengthen an expectancy that some behavior or event will be followed by that reinforcement in the future. In the present experiment, subjects predicted which of two lights would come on next. Part of the reinforcement then, was the actual lighting of one of the two bulbs on each trial. This reinforcement was the same in all of the experimental conditions. However, in all but one of the experimental conditions the subjects also bet chips. The gain or loss of a chip can be thought of as additional reinforcement. All experimental evidence indicates that subjects apparent ly get to a stage of probability matching at some point. The issue is why some continue past this level and maximize their predictions of the more frequent event while others stay at a rate that closely approximates the objective fre quency of occurrence. It is hypothesized that the more re inforcement available in an experimental condition, the more effectively subjects will learn. Behaviorally, this means they will approach the "optimal strategy" which is maximizing one’s predictions of the more frequent event. The following diagram (p. 28} should help to clarify this point. Let us assume, in this preliminary theorizing, that winning and losing chips both serve to alter expectancies, and, like Rotter, anything that alters expectancies will be considered a reinforcement. Then it can be seen from the diagram that the no payoff condition has the least rein forcement. The lose or stand condition has next to the least reinforcement because the individual only receives reinforcement when he predicts incorrectly and that is not what occurs most often. Individuals in the win or stand condition receive next to the highest amount of reinforce- : ment. It^ is given when the more frequently occurring event ‘ CONDITION: No Payoff IPkiiDiu'i'Jib tfhbAVlok: fcrob. Match Win or Lose Win or Stand Lose or Stand Between ProbV Match and Overshooting Maximize Overshoot |S predicts more ITrequent event IS predicts less frequent event 0 0 0 0 0 0 0 0 ;C = correct prediction 11 = incorrect prediction |Shaded area = most frequently occurring situation (i.e., S predicts the more frequent event and is correct) "~ i0 = no reinforcement beyond the basic situation of learning if the prediction was j correct } i j+ = reinforcement through winning a chip i |- = reinforcement through losing a chip to oo 29 is correctly predicted. That is what all subjects do most often, so subjects in this condition are receiving rein forcement most of the time. Finally, subjects in the win,/ lose condition receive the most additional reinforcement because it occurs on every trial. It is predicted that as the amount of reinforcement increases from one condition to another, the predictions of the more frequently occurring event will increase. This hypothesis also follows the ex perimental evidence discussed earlier (Seigel and Goldstein, 1964; Edwards, 1955}. There, the trend seemed to be to wards probability matching in the no-payoff condition, pre- dieting slightly above the objective frequency in the lose- or-stand condition and maximizing in the win-or-lose condi tion. The present study also included a fifth, unique con dition. Called the Reverse condition, subjects lost a chip if they predicted correctly and broke even if they pre dicted incorrectly. On logical grounds, it was expected that subjects would "reverse probability match.” That is, they would predict each event at approximately the frequen cy that the other event occurs in the sequence. Another general hypothesis of this study concerned the Internal vs. External Control of Reinforcement dimen sion. From the evidence previously cited it would be ex pected that internals would tend to approach the situation with a great belief that 1) they can learn useful 30 information about the nature of the event sequence, and 2) based on that information they can either alter the event sequence or alter their own behavior in such a manner as to increase the amount of external reinforcement they receive. It was therefore predicted that in the probability learning task used in this study, internals would predict the more frequent event (i.e., maximize) more than externals, in all experimental conditions. Lastly, the study predicted that males and females would behave differently in this probability learning situ ation. Because of the paucity of literature in this area, there was no attempt to specify the direction of the dif ference. CHAPTER III METHOD Design The primary purpose of the present investigation was to evaluate the prediction that behavior in a binary choice probability learning situation will vary with the type of risk condition involved. The experimental task required the subject to predict which of two lights, a red one or a green one, would light up on each trial. Two hundred trials were administered to each subject under one of the following five experimental risk conditions: No Payoff— the subject received no payoff other than learning if his prediction was correct or incorrect by seeing which light came on. Win Or Lose--the subject bet one chip per trial. In addition to seeing the light come on, he won an additional chip for a correct prediction and lost a chip for an incorrect prediction. Win Or Stand--the subject bet one chip per trial. In addition to seeing the light come on, he won an additional chip for a correct prediction and stood __________ 31______ ________________ _ __ 32 (did not lose anything) on an incorrect prediction. Lose Or Stand--the subject bet one chip per trial. In addition to seeing the light come on, he kept his chip if he predicted correctly and lost it if he predicted incorrectly. Reverse--the subject bet one chip per trial. In addition to seeing the light come on, he lost a chip if he predicted correctly and stood if he predicted incorrectly. Following completion of the experimental task, each subject was asked to 1) estimate what percentage of the total time the more frequent event (red light) occurred, and 2) describe how he decided on his predictions at vari ous points during the experiment. This was done on an ex ploratory basis to allow for more detailed examination of some of the factors involved in probability learning. A second purpose of the study was to examine sex differences in probability learning. Accordingly, an equal number of males and females were tested under each condi tion. Within each experimental condition, half of each sex group were Internals and half were Externals on the Control of Reinforcement dimension. The final concern of this study was an evaluation of the prediction that Internals and Externals would per form differently in a probability learning situation. To 33 allow for appropriate statistical evaluation of this hypoth esis, an equal number of Internals and Externals were as signed to each experimental condition. Hypotheses 1. The frequency of prediction of the more fre quently occurring event will differ according to the experimental condition. It will be low est in the Reverse condition and will increase by condition in the following order: No Pay off, Lose or Stand, Win or Stand, Win or Lose. This is the effect' of increasing amounts of reinforcement. 2. Females will differ from males in the frequency with which they predict the more frequently occurring event. 3. Internals will predict the more frequently oc curring event more often than Externals. This is due to their more effective use of available reinforcements.. 4. Subjects in high additional reinforcement con ditions (i.e., Win-Lose and Win-Stand) will predict the more frequently occurring event significantly more often than subjects in the low additional reinforcement conditions (i.e., No Payoff and Lose-Stand>. All of the above hypotheses refer to behavior at asymptote. Based on previous experimental evidence, this is defined as the last fifty trials in a series of two hundred trials. Subjects The experimental subjects consisted of seventy males and seventy females. They were recruited from in troductory psychology classes at the University of Southern I California. Their participation gave the students credit towards a course requirement that they serve in a "subject pool.” The procedure for recruiting subjects was as fol lows. The author visited seven classes and had students complete the Internal-External Scale (Appendix A). She was first introduced by name by the instructor who also told the class that she was a graduate student doing research. She then gave the students the following information: I would like your cooperation in completing this questionnaire, which is being used for experi mental purposes. Anybody who does not wish to fill it out may leave the form blank. It is strictly voluntary. Individuals who do complete the form may be asked to participate in further experimental tasks at a later date. That will count as required serv ice in the subject pool. It takes about ten minutes to complete the questionnaire. Work quickly and do not dwell on any one question. At the end of the questionnaire, subjects indicated YES or NO to: Would you be willing to participate in an experi- c. . ment on how people behave in gaming situations? It would require about one hour of your time. Data from students who indicated NO were not in cluded in any subsequent calculations. There were three such individuals. Separate distributions of scores were then made for each sex for each class. Within each distribution, 35 Externals were designated as those falling in the upper third of scorers and Internals were the lower third of scorers. The data was then pooled to form four groups: Male-High, Male-Low, Female-High and Female-Low. Within each group subjects were then randomly assigned to one of the five experimental conditions. Each of these students was then telephoned by the author. Basically they were told: My name is Marion Aronowitz. I'm running a psychology experiment. You filled out a question naire for me in Dr. __________ ’ s class and you in dicated that you were willing to serve as a subject in the experiment. Are you still interested in participating? All but three students indicated that they were interested. An appointment time was then arranged. Several days prior to the scheduled appointment each subject was sent a re minder letter (Appendix B>. There were ten subjects who failed to keep the appointment. These drop-outs came from all of the sex-personality categories. A total of one hundred and forty-four subjects were tested in the proba bility learning situation.. Data collected from four of them were not used because of equipment failure. Apparatus The apparatus consisted of two separate pieces of equipment connected by a cable. One was a small flat metal box which was placed near the subject. On the top surface 36 of the box were three lights. The middle light was white and lit up to indicate the start of a trial. The light on the subject’s left was red and the one on his right was green. Under each of these two lights was a small button which the subject pressed to indicate his prediction of that light. Following each prediction, the red or green light lit up. This piece of equipment was connected by a long cable to a wooden box which ran a piece of video tape. Punched onto this tape was the program for the sequence in which the lights occurred. The tape moved continuously and as it passed through photo electric cells, the appropriate lights lit up on the metal box in front of the subject. The program was arranged for a 75:25 proportion. Within each block of twenty trials the red light appeared fifteen times and the green light five times. Randomizing within blocks of twenty trials is typical in probability learning studies. The order of presentation was random. The green light never occurred consecutively more than two times. The white warning light remained on for three sec onds. After a one-second pause, the red or green light came on for three seconds. The intertrial interval lasted four seconds during which bets were settled by the experi menter, using white plastic poker chips. Procedure 37 Instructions The experimenter greeted each subject individually in the reception area of the Psychological Research and Service Center at The University of Southern California and brought him into the experimental room. The subject was seated at a large wooden desk. The metal box with the lights was on the desk about four ipches from the subject. The connecting cable ran from the box across the top of the desk and was plugged into the programmer which was placed on a low chair on the far side of the desk. It was previ ously determined that the noise from the apparatus provided no cues as to which light would occur next. In all but the No Payoff condition, there were 200 white chips on the subject’s right. The experimenter was seated directly across from the subject. The chair with equipment was on her immediate right. In the Payoff conditions, on the desk to her left were an additional 200 white chips. The subject was then given the following instruc tions: You probably remember that you signed up to be a subject in a study on gaming behavior. You will be given two hours of credit towards the four-hour subject pool requirement for your Psychology 200 class. I would like to make it clear before you begin that nothing embarrassing, unpleasant or painful will happen to you. There are no tricks in this experiment and your task will be exactly what I 38 describe to you. However, I can't tell you any more about the study at this time. As you probably know, frequently in psychology experiments subjects cannot be given complete information until after it is all over, because advance knowledge may affect their behavior. This experiment will be concluded next spring, and at that time I will be happy to answer any questions you may have. You have my home number and can call there. In the meantime, it is essen tial that you do not discuss anything that we do today with any other students. Your task in this study will be to predict whether the red light or the green light will light up next. Each trial will start when the yellow warning light comes on. You are to immediately-- place a bet by taking one chip from the stack of 2QQ chips on your right and place it here. Never place more than one chip. Then^---decide whicK light you think will light up and press the button under that light. Remember, your task is to pre dict which light will come on. Your responses will be recorded here (Point to equipment). During the experiment, I will be moni toring the equipment to insure that it functions properly. I am not altering it in any way--just making sure it does what it should do automatical ly . For subjects in the Win Or Lose Experimental Group, the instructions continued as follows: Each time your prediction is correct, you get to keep the chip you bet and you win a chip. As soon as you have been paid off you should take both of the chips and put them back with your stack of chips. If your prediction is incorrect, you lose the chip you just bet. For subjects in the Win Or Stand Experimental Groups, the instructions continued as follows: Each time your prediction is correct you get to ^The underscored instructions were not given to subjects in the No Payoff condition. 39 keep the chip you bet and you win a chip. If your prediction is incorrect, you still keep your origi nal chip. You merely do not win an additional chip. As soon as you either have been paid off or know that no payoff is due, remove all the chips from this trial and put them back with your stack of chips. For subjects in the Lose Or Stand Experimental Group, the instructions continued as follows: Each time your prediction is correct you get to keep the chip you bet. As soon as you know that you were correct, remove the chip and put it back with your stack of chips. If your prediction is incorrect, you will lose the chip you just bet. For the Reverse Experimental Group, the instruc tions continued as follows: Each time your prediction is correct, you lose the chip you just bet. If your prediction is in correct, you get to keep the chip you just bet. As soon as you know that you are to keep the chip, remove it and put it back with your stack of chips. For all subjects the instructions continued as fol lows* This procedure will be repeated 200 times. Do you have any questions? To make sure that the in structions were completely clear, I would like you to describe to me what you are to do after the warning light goes on. (If not covered by the sub jects, ask: What will happen if you predict cor rectly? What will happen if you predict incorrect ly?) Experimental Trials Each subject received 200 trials. The lights were programmed to occur randomly in a 3-1 ratio with the red light as the more frequently occurring eyent. The payoff 40 transactions were handled by the experimenter. During the trials the experimenter looked at the piece of recording equipment. There was no eye contact with the subject, even when giving or taking a chip. The experimenter did not talk to the subject and conversation on the part of the subject was discouraged tactfully. Post-Trial Interview Following the trials, each subject was asked: *What would be your estimate of the percentage of time the red light came on?” Subjects were then asked to think back to the beginning of the trials and retrace aloud the various strategies they used in making their predictions. CHAPTER IV RESUETS The results can be grouped under two main headings j The first relates to predictions of events in a binary choice probability learning situation. This was the main area of interest in this investigation and all hypotheses concern predictive behavior. The second area concerns estimating the frequency of a.complete set of probability learning events after all the events have occurred. This was examined bn an explora tory basis. Predictive Behavior The three variables under investigation were con ditions of risk, I-E, and sex of the subject. The first hypothesis to be tested was concerned 3 with conditions of risk. It predicted 1) a significant difference among the five experimental conditions in pre dicting the more frequent event and 2) the order of occur rence of the means. A three-way analysis of variance was calculated. 42 Since the interest was in behavior at asymptote, only the last block of fifty trials was used. This was done follow ing a preliminary analysis which showed that there were no significant differences in the data using the last 100 trials and the last fifty trials. It was thus determined that asymptote had been reached by the last fifty trials. The results are summarized in Table 1. The resulting F of 11.3 for Conditions was significant at the .001 level. In order to determine the specific sources of significant mean differences among conditions, pairwise multiple comparisons were made using Tukey's method (Myers, 1966). Table 2 shows the means of the five conditions and the percentage of predictions of red for the last fifty trials. Table 3 lists those pairwise Tukey comparisons which were signifi cant. Although the main effect was significant, supporting the first part of hypothesis 1, the comparison of means shows that the effect was mainly due to the Reverse condi tion. This condition was of particular interest because subjects interpreted the instructions in two different ways. Some took the instructions literally and attempted to predict correctly, even though they lost chips for do ing so. Others decided to try to keep as many chips as possible. To do this, they tried to predict incorrectly. Table 4 shows the number of subjects taking each approach. Table 5 shows the probabilities for each pair using 43 TABLE 1 SUMMARY OF ANALYSIS OF VARIANCE OF NUMBER OF PREDICTIONS OF THE MORE FREQUENTLY OCCURRING EVENTS Source df Mean Squares F Sex (S> 1 65.8282 1.22 Internal-External (I-E) 1 71.4289 1.31 Conditions (C) 4 611.3461 11.3** S x I-E 1 358.3996 6.6* S x G 4 46.0251 0.86 C x I-E 4 26.8391 0.48 S x I-E x C 4 376.3195 6.9** Within replicates 120 54.2515 Total 139 * p<.02 ** pc.001 44 TABLE 2 MEANS OF THE EXPERIMENTAL RISK TAKING CONDITIONS AND PERCENTAGE OF TOTAL PREDICTIONS Condition Mean Percentage No Payoff Win-Lose Win-Stand Lose-Stand Reverse 39.2499 41.6071 41.4999 40.5714 30.4999 78.4% 83.2% 82.8% 81.0% 60.8% 45 TABEE 3 PAIRS OF MEANS WHICH DIFFER SIGNIFICANTLY USING TUKEY'S METHOD SHOWN ACCORDING TO CONDITION OF RISK No Payoff -- Reverse Win-Lose — Reverse Win-Stand — Reverse Lose-Stand -- Reverse 46 TABLE 4 NUMBER OF SUBJECTS WHO TRIED TO PREDICT CORRECTLY VS. NUMBER PREDICTING INCORRECTLY SHOWN BY SEX AND PERSONALITY TYPE* Subject Type Correctly Incorrectly Male External 3 4 Male Internal 7 0 Female External 6 1 Female Internal 4 3 Total 20 8 ^Correctly is picking the more frequent event more than half the time. Incorrectly is picking the less fre quent event more than half the time. 47 TABLE 5 EXACT PROBABILITIES OF NUMBER OF SUBJECTS WHO TRIED TO PREDICT CORRECTLY VS. NUMBER WHO TRIED TO PREDICT INCORRECTLY SHOWN BY PAIRS* Female-Externals Female-Internals: .01 Male-Internal .09 Male-External -3, Male-Internals Male-Externals Female-Externals: .5 .12 Male-Externals Male-Internals: .001 ^Computed using Fisher's Exact Test (Hays, 1963). 48 Fisher's exact test (Hays, 1963). The Internal and Exter nal males differed significantly, as did the females. The second part of hypothesis 1 predicted the order in which the means of the conditions would occur. Table 6 compares the predicted and actual orders, where 1 indicates the greatest frequency of prediction of the red light and 5 the least frequency. A rank order correlation was com puted. The Kendall Tau Coefficient was 1.0 (Hays, 1963). Thus, although the mean differences were not great enough to be significantly different (except for the Reverse con dition), they were in the predicted direction and in the exact order. The second hypothesis to be tested concerned sex differences in probability learning. As can be seen from Table 1, there was no significant main effect due to the sex variable. Similar results were obtained regarding the I-E variable, which was the variable involved in hypothesis 3. That is, there was no main effect. However, both hypothe sis 2 and 3 are partially supported by the significant two-way interaction between sex and I-E. The F of 6.6 was significant at the .02 level. Table 7 shows the means for these two variables. Using the Tukey multiple comparison method, the only significant difference was between Male Externals and Male Internals, where the Internals tended to 49 TABLE 6 RANK ORDER OF PREDICTION OF THE MORE FREQUENTLY OCCURRING EVENT Experimental Condition Hypothesized Rank Actual Rank No Payoff 4 4 Win-Lose 1 1 Win-Stand 2 2 Lose-Stand 3 3 Reverse 5 5 50 TABLE 7 MEANS FOR THE INTERNAL-EXTERNAL AND SEX VARIABLES External Internal Male 37.0571 41.6857 Female 38.8857 37.1142 51 maximize as predicted. The difference between Male and Female Internals was almost significant. Table 1 also shows that there was a significant 3- way interaction among the Sex, I-E and Condition variables. The F of 6.9 was significant at the .001 level. The means are shown in Table 8. Table 9 lists the significant pair wise comparisons of the means from Table 8 using Tukey’s method of comparison. Since all significant pairs involve the Reverse condition, it was decided to examine this con dition separately. Table 10 shows those comparisons. Only the Male Internals differ significantly from other groups. This is the result of two factors. First, they interpreted the instructions literally and tried to predict correctly, despite losing chips (see Table 4). However, Female Exter nals tended to do the same thing. The difference is that Male Internals also tended to maximize their prediction of the red light (see Table 8>. Among the Female Externals, although six out of seven predicted the red light more of ten than the green light, five out of these six predicted the green light at a rate above its actual occurrence. That is, they were undershooting the probability matching level while Male Internals were maximizing. It can be seen from Table 8 that Male Internals in the Reverse condition maximized more than any other group in the study. The fourth hypothesis predicted that subjects in I | TABLE 8 i MEANS OF THE PREDICTIONS OF THE MORE FREQUENTLY OCCURRING EVENT SHOWN ACCORDING TO CONDITION, SEX AND I-E No Payoff Win-Lose Win-Stand Lose-Stand Reverse I E I E I E I E I E Male 41.42 37.85 41.57 42.42 40.42 42.28 41.42 39.42 43.57 23.28 Female 38.57 39.14 40.57 41.85 43.14 40.14 41.57 39.85 21.71 33.42 Cl CO TABLE 9 53 PAIRS OF MEANS WHICH DIFFER SIGNIFICANTLY USING TUKEY’S METHOD SHOWN BY CONDITION, SEX AND I-E* ME: No Payoff ME: Reverse ME: No Payoff - FI: Reverse ME: Win-Lose - ME: Reverse ME: Win-Lose - FI: Reverse ME: Lose-Stand — ME: Reverse ME: Lose-Stand — FI: Reverse ME: Reverse _ MI: No Payoff ME: Reverse - MI: Win-Lose ME: Reverse - MI: Win-Stand ME: Reverse - MI: Lose-Stand ME: Reverse — MI: Reverse ME: Reverse - FE: No Payoff ME: Reverse — FE: Win-Lose ME: Reverse - FE: Win-Stand ME: Reverse - FE: Lose-Stand ME: Reverse - FI: No Payoff ME: Reverse - FI: Win-Lose ME: Reverse - FI: Win-Stand ME: Reverse - FI: Lose-Stand FE: No Payoff - FI: Reverse FE: Win-Lose - FI: Reverse FE: Win-Stand — FI: Reverse FE: Lose-Stand — FI: Reverse FI: No Payoff - FI: Reverse FI: Win-Lose - FI: Reverse FI: Win-Stand - FI: Reverse FI: Lose-Stand - FI: Reverse MI: No Payoff FI: Reverse MI: Win-Lose - FI: Reverse MI: Win-Stand — FI: Reverse MI: Lose-Stand — FI: Reverse MI: Reverse “ FI: Reverse *M - Male F - Female I - Internal E - External 54 TABLE 10 ALL PAIRWISE COMPARISON OF MEANS FOR THE REVERSE CONDITION USING TUKEYT S METHOD Female- Internal : Female-External Not Significant Male-Internal Male External Significant Not Significant Female- Male-Internal Male-External External: Not Significant Not Significant Male- Male-External Internal: Significant 55 the Win-Lose and Win-Stand conditions would predict the more frequently occurring event at a higher rate than sub jects in the Lose-Stand and No Payoff conditions. To test this hypothesis, a three-way analysis of variance was calculated. Win-Lose and Win-Stand were com bined to form one experimental condition and No Payoff and Lose-Stand were combined to form the other. The results are summarized in Table 11. The resulting F of 3.59 for conditions was significant at level between .05 and .10. The mean for the low reinforcement group was 39.91. The mean of the high reinforcement group was 41.55. Estimating Behavior Upon completion of the experimental task, each sub ject was asked, ”What would be your estimate of the per centage of time the red light came on?” (The actual per centage was 75.> In previous studies subjects have been quite accurate in their estimates. Since the present study involved experimental conditions which have not been examined before, estimates were obtained on an exploratory basis to see if accuracy of estimate would persist. A three-way analysis of variance was calculated to determine if there were any significant differences in the estimates of different groups. The results are summarized in Table 12. The F of 3.36 for experimental conditions was significant at the .02 level. To determine the specific 56 TABLE 11 SUMMARY OF ANALYSIS OF VARIANCE OF NUMBER OF PREDICTIONS OF THE MORE FREQUENTLY OCCURRING EVENT BY I-E, SEX, AND HIGH-LOW REINFORCEMENT LEVELS Source df Mean Squares F Sex (S) 1 1.7499 <1 Internal-External (I-E) 1 14.2856 <1 High-low Conditions (HLC) 1 75.5713 3.59* S x I-E 1 0.00001 <1 S x HLC 1 0.00001 <1 I-E x HLC 1 26.0355 1.2 S x I-E x HLC 1 34.3202 1.6 Within Replicates 104 21.0960 Total 111 *p<.10 57 TABLE 12 ANALYSIS OF VARIANCE OF ESTIMATES OF THE PERCENTAGE OF TIME THE MORE FREQUENT EVENT OCCURRED Source___________________df Mean Squares_____F Sex (S) 1 70.0070 1.32 Internal-External (I-E) 1 49.2069 <1 Condition (C) 4 179.6783 3.36* S x I-E 1 34.00 <1 S x C 4 165.7927 3.09* i I-E x C 4 97.3498 1.81 S x I-E x C 4 64.9931 1. 21 Within Replicates 120 53.6349 Total 139 *p<.02 58 source of significant differences, multiple pairwise com parisons of means were made using the Tukey method. The means are shown in Table 13. The only significant differ ence using Tukey's method of comparison was between the means of the No Payoff and the Win-Lose conditions. Table 14 compares the rank order of estimates to the rank order of predictions. The Spearman Rank Order Correlation Coefficient yields a value of .7 (Hays, 1963). If, however, the Reverse condition were not included, it can be seen that there would be a perfect rank order corre lation between predictions and estimates. Table 12 also shows that there was a significant interaction effect between sex and conditions. The F of 3.09 was significant at the .02 level. Table 15 shows the sex by condition means. Table 16 indicates which pairwise comparisons of the means in Table 15 were significant using Tukey's method of comparison. These results will be dis cussed in a later section. To further explore the effect of high additional reinforcement (i.e., Win-Lose and Win-Stand conditions) vs. low additional reinforcement (i.e., Lose-Stand and No Payoff conditions) on subject's estimates, an analysis of variance of estimates was calculated with the Win-Lose and Win-Stand groups combined and the No Payoff and Lose-Stand groups combined. The results are shown in Table 17. The 59 TABLE 13 MEANS OF ESTIMATES OF PERCENTAGE OF OCCURRENCE OF THE MORE FREQUENT EVENT SHOWN BY CONDITION Condition_______________Mean % Estimated No Payoff 75.49 Win-Lose 81.07 Win-Stand 80.28 Lose-Stand 75.89 Reverse 77.42 60 TABLE 14 RANK ORDER OF PREDICTIONS AND ESTIMATES SHOWN BY EXPERIMENTAL CONDITION Predictions Estimates No Payoff 4 (78.4) 5 (75.49) Win-Lose 1 (83.2) 1 (81.07) Win-Stand 2 (82.8) 2 (80.28) Lose-Stand 3 (81.0) 4 (75.89) Reverse 5 (60.8) 3 (77.42) *Number in parentheses is the mean. 61 TABLE 15 MEANS OF ESTIMATES OF PERCENTAGE OF OCCURRENCE OF THE MORE FREQUENT EVENT SHOWN BY CONDITION AND SEX Condition Male Female No Payoff 77.07 73.92 Win-Lose 80.71 81.42 Win-Stand 83.07 77.49 Lose-Stand 72.85 78.92 Reverse 79.99 74.85 62 TABLE 16 SIGNIFICANT PAIRWISE COMPARISONS OF MEANS OF ESTIMATES USING TUKEYT S METHOD SHOWN BY CONDITION AND SEX 1. Male: Win-Lose Male: Lose-Stand 2. Male: Win-Lose Female: No Payoff 3. Male: Win-Stand Male: Lose-Stand 4. Male: Win-Stand Female: No Payoff 5. Female: Win-Lose Male: No Payoff 6. Female: Win-Lose Female: No Payoff 7. Male: Win-Stand Female: Reverse 8. Male: Lose-Stand Male: Reverse 9. Female: Win-Lose Female: Reverse 63 TABLE 17 ANALYSIS OF VARIANCE OF ESTIMATES OF THE PERCENTAGE OF TIME THE MORE FREQUENT EVENT OCCURRED. SHOWN BY I-E, SEX AND HIGH VERSUS LOW REINFORCEMENT LEVELS. Source df Mean Squares F Sex (S) 1 6.5089 <1 Internal-External (I-E) 1 27.0085 <1 High-Low Conditions (HLC) 1 695.0087 11.7* S x I-E 1 37.7233 <1 S x HLC 1 106.0805 1.7 HLC x I-E 1 16.5089 <1 S x HLC x I-E 1 0.5009 <1 Within Replicates 104 59.1511 Total 111 *p<.001 64 F of 11.7 for conditions was significant at the .001 level. That is, people who received more reinforcement had higher estimates of how often the red light occurred than people who received less reinforcements. The means were 75.69 for the low reinforcement group and 80.67 for the high rein forcement group. The low reinforcement group was therefore much closer to the actual frequency of 75% than the high reinforcement group. CHAPTER V DISCUSSION Predictive Behavior The primary purpose of the present investigation was to examine probability learning under varying condi tions of risk. It was hypothesized that as the amount of available reinforcement increased from condition to condi tion, the frequency of prediction of the red light would also increase. Specifically, it was expected that the num ber of predictions of red would increase by condition as follows: No Payoff, Lose or Stand, Win or Stand, Win or Lose. This order represents increasing total amounts of reinforcement received by the subject using the conceptual scheme outline in Chapter II. The Reverse condiction does not easily fit this scheme because the subject receives negative reinforcement (i.e., loses a chip) for selecting the positively reinforcing light (i.e., the correct light). It was expected, however, that subjects would try to keep chips by predicting the incorrect light and on that basis would have the fewest predictions of red. 65 66 Although there was a highly significant main effect for experimental conditions (Table 1), it was caused solely by the subjects in the Reverse condition. Because of the unusual nature of this condition, and the subject's reac tion to it, it will be reviewed separately later in the discussion. The other four conditions were in the exact order predicted but the effect was not strong enough to be significant. It is possible that a stronger effect might have been obtained using money instead of chips. However, from experimental evidence, it is not likely that this was the crucial factor. Seigel and Goldstein (1964) used money and obtained significant differences in a binary choice probability learning study, while Jones and Myers (1966), using the same arrangement, did not. Also, significant re sults have been obtained in risk taking studies using chips (Phares, 1957). It may be that some other type of rein forcement, or larger amounts of reinforcement, would pro duce a main effect. It also seems particularly likely that a reinforcement could be chosen which would have different appeal to males vs. females or to internals vs. externals. If that were true, it would certainly add to our under standing of the role of reinforcement in probability learn ing. The fact that the four conditions were in the exact order predicted points out the potential usefulness of 67 conceptualizing the conditions in terms of how much rein forcement is available in each. When the conditions were grouped according to high and low reinforcement, the effect approached significance (p between .05 and .10) with the low reinforcement group closer to probability matching and the high reinforcement group tending to maximize (Table 11). It could be argued that the high reinforcement group is also a positive reinforcement group while the low rein forcement group is not. This, however, does not help ex plain the order of conditions in previous studies (Seigel and Goldstein, 1964) or in the present investigation. It is also possible to view correct answers as receiving posi tive reinforcement and incorrect ones as receiving negative reinforcement. However, efforts at developing a probabil istic mathematical model which would predict the empirical results have not proven fruitful (Becker and McClintock, 1967). The simpler and more direct explanation for the usual risk taking conditions used in binary choice proba bility learning is the concept of amount of reinforcement referred to above. All subjects are exposed to exactly the same sequence of reinforcing lights. The difference in conditions depends upon whether or not additional rein forcements such as chips are used, and if so, the rules for dispensing and/or removing them. Where the rule provides additional reinforcement on every trial* as in the Win-Lose 68 condition, the subject learns best. Obviously, when the subject predicts incorrectly, the additional reinforcement involved is negative, in that he loses a chip. The impor tant point, however, is that he is always reinforced. In the Win-Stand condition he is reinforced on most, but not all trials. In the Lose-Stand condition he receives the least additional reinforcement and it is all negative. In the No Payoff condition he receives no additional rein forcement. Thus, the amount of reinforcement available in the experimental condition for predicting red, which is what all subjects do most often, seems to be the crucial factor. The more difficult problem is why this is so. One possibility is that additional reinforcement serves to underscore the occurrence of the event. That is, it may aid the subject’s memory of the event and therefore give him a more accurate picture of the total event structure. The results also showed a significant interaction between Sex and Internal vs. External, but no main effect for either. Rotter (1966) points out that although I-E has proven itself to be a very useful variable, the spread of scores in a relatively homogeneous population such as col lege students is not as great as one would like. The ques tion then becomes whether a more refined measure of this broad characteristic could be developed. A review of the questions used in the I-E Scale (Appendix A) indicates that 69 many of them center around school grades and politics and are somewhat repetitious. It is possible that a broader sampling of topics would be more effective. More import antly, the I-E Scale does not tap beliefs in control over feelings and impulses coming from within the person. It is concerned solely with beliefs about external reinforcement. However, it seems to this author that what a person thinks he can do about his internal reactions is probably related to his beliefs about controlling external events. A scale ■ / which tapped both might prove more powerful in discrimin ating individuals and groups. Turning to the sex-I-E interaction, male-internals differed significantly from male-externals. The difference between male-internals and female-internals approached sig nificance. A few previous studies have provided evidence that in some situations, the I-E dimension may operate dif ferently for males than females (see Lefcourt, 1966; Cardi, 1962). Cardi (1962) found a low correlation between intel ligence and I-E for a sample of college females but no cor relation for the males. Crandall (1962) found a relation ship between I-E and achievement behaviors for boys but not for girls, and James, Woodruff and Werner (1965) reported that among males who quit smoking and did not return, most were internals. The I-E dimension was not significant in predicting the behavior of the females. The findings of 70 the present study provide further support for the inclusion of sex as a variable in future studies on I-E. From the evidence cited above, it appears that I-E is more usually useful in distinguishing the behavior of males than females. It thus seems that many of the previous studies involving I-E should be done again using sex as a variable. The most interesting finding of the present inves tigation was the behavior of subjects in the Reverse condi tion, where they lost a chip for predicting correctly and broke even on incorrect predictions. It had been expected that subjects would predict the opposite light in order not to lose chips. In fact, some subjects did exactly that while others interpreted the instructions literally, tried to predict correctly and lost chips. Table 4 shows the number of subjects taking each approach. Table 5 shows the exact probabilities of obtaining these results. It can be seen that some groups clearly differ in what they chose to do. In particular, internality vs. externality very clear ly distinguishes the pairs within each sex grouping. Among the male internals, every subject chose to predict correct ly and lose chips. This fits the conception of the inter nal as a person who feels he can control or redefine the situation to make it satisfying to himself, rather than be bound by what is usually socially expected. In a gambling situation such as this, the usual thing is to try to win. 71 Typically, winning involves accumulating money or chips. Forced with this unusual situation of having to choose be tween not following the task instructions or losing chips, the male-internal does the latter. That is, he is not as dependent upon the external reinforcement of the chip, or at least the chip as it is usually used. From their re marks afterwards (Appendix D), it seemed that male- internals were either disinterested in the chips or were able to accept a new way of handling them. Some of their comments follow. MI wanted to guess right. The chips had no meaning. It was being correct that counted.” ”1 did what you told me. The chips didn't matter." "Predicting correctly seemed right. I figured the object was to get rid of the chips like in certain card games." Some male-externals did the same thing, but most took a more traditional approach. That is, they actually did the opposite of what they were instructed to do, just to keep chips. '*To me, being wrong was being right." "You said this was a gambling experiment, so it seemed more log ical not to lose." It can be seen from Table 10 that not only were the number of subjects predicting correctly dif ferent for male internals and externals, but the mean num ber of predictions of the more frequent event significantly differed for the two groups. The female internals and externals, who also 72 differed significantly in the number predicting correctly and incorrectly in the Reverse condition, present a very different picture than the males. First, the order is ex actly opposite from the males. That is, there are more ex ternals who predict correctly among the women. However, as can be seen from Table 10, the mean number of predictions of the more frequent event did not differ significantly. Closer study of the means for the Reverse group, as shown in Table 8, gives a clue as to what occurred. The female externals had only one person try to predict incorrectly, yet the mean for that group is quite low (33.42). One might hypothesize they were more ambivalent about the chips and could not easily decide whether to keep them or lose them. Instead they tried to do some of both, as indicated by some of their comments. "At first it bothered me. Then I got into a thing of if I moved a chip (i.e., across the desk to E) I won.11 "It really bothered me in the begin ning. When you win you want to keep something. If I lost a chip to you (i.e., E) I was still thinking about it by the next trial. I thought about predicting the other light --then decided not to. I don't know Why. Then I got used to giving them to you. Then I wanted them on your side.1' Interestingly, some of the female-internals seemed to have had the same problem. "I wanted to pick the right one but I wanted to have the most chips. I kind of tried 73 picking the right one more. It got so I’d rather not have the chips after a while. The main thing became to pick correctly.'* "At first I didn't know. Then I decided that when you gamble you usually want to keep them (i.e., chips). That means winning." "At first it was the more chips you (E) had was winning. Then it was keeping chips." Thus, the differences among the males were strong, clear-cut, and in line with the general concept of internal vs. external control of reinforcement. The differences among the females were in the opposite direction, the means did not differ significantly, and women seemed to have much more difficulty in selecting a definite approach and stick ing with it. This further supports the possibility sug gested earlier that the I-E variable is less useful in pre dicting behavior among females. It is clearly quite useful with males, particularly in situations like the Reverse condition where it is not simply a matter of predicting more or less red lights (i.e., probability matching vs. maximizing). Rather, it is a matter of choosing between two distinct courses of action (i.e., predicting green vs. predicting red). Estimating Behavior When subjects are asked to view an entire display of events and then estimate the frequency of occurrence of one type of event in the display, it has been reported that 74 their estimates are quite accurate. Generally, they are within a few points of the true value (Edwards, 1961a). For example, following a binary choice probability learning task, Tversky and Edwards (1966) had their subjects esti mate how often the more frequent light had occurred. Their estimates were close to the actual frequency with an abso lute deviation of 3.8%. However, it appears that none of these investigators have analyzed the estimates according to experimental conditions. Since it seemed possible that estimates might vary with the type of subject and especial ly with the type of risk condition, it was decided to in clude this area in the present investigation' on an explora tory basis. Following the probability learning task, all subjects were asked, "What would be your estimate of the percentage of time the red light occurred?” Both the main effect by experimental condition and the sex by condition interaction were significant at the .02 level (Table 12). The No Payoff and Win-Lose conditions were the only pair that differed significantly. Using the system outlined earlier, it can be seen that the No Payoff condition is the one with the least reinforcement and the Win-Eose condition is the one with the most reinforcement. Thus, at least in the most extreme conditions, subjects differ significantly in their perception of what occurred. When all of the data (except the Reverse condition) is viewed in terras of 75 conditions of high vs. conditions of low reinforcement (Table 17) the effect is even more dramatic and is signifi cant at the .001 level. The sex by condition interaction (Table 12) is somewhat harder to interpret. From Table 16, it can be seen that for the first six pairs, significant differences are always between a high reinforcement and a low rein forcement pair. The reinforcement value of the Reverse condition is harder to determine since it partly depends upon the subject’s interpretation of the instructions. If he elected to try to not lose chips, it seems to be similar to the Lose-Stand condition and therefore of low reinforce ment value. It is less clear-cut how to categorize the reinforcement value for those who elected to lose chips and predict correctly. From the comments of several subjects who said that they just reversed the usual idea and made losing a chip the goal, an argument could be made for call ing this a high reinforcement situation. The latter would apply primarily to males, since of those who tried to pre dict correctly, they showed a much greater tendency to max imize. Using this formulation, the five experimental con ditions can be categorized according to their additional reinforcement value as follows: 76 No Payoff - low Win-Lose - high Win-Stand - high Lose-Stand - low Reverse where predicting correctly - high where predicting incorrectly - low Returning to Table 16, it can be seen that for the first six pairs significant differences in estimates are always between a high and a low reinforcement condition. This trend seems to continue in pairs seven through nine. Table 8 shows that males as a group tended to predict cor rectly (i.e., predict red) more than females. Thus, if male-reverse is considered to be a high reinforcement situ ation, all nine comparisons are between a high-low rein forcement pair. There does not seem to be any clear-cut way to describe the role of the sex variable in this inter action. Finally, and perhaps most importantly, there is a .7 rank order correlation between predictions and esti mates. Eliminating the Reverse condition, the correlation is perfect. One possibility is that the reinforcement com bination for a particular condition affected the individual’s predicting behavior and the predicting behav ior in turn affected the estimating behavior. That is, those who predicted red more often remembered red as occur ring more often. It seems apparent that estimating behavior is as 77 complex a phenomenon as predicting behavior. Pointing out the fact that people estimate what has gone on fairly ac curately seems to have prevented investigators from notic ing that people also estimate with very different degrees of accuracy depending upon the reinforcements they have re- ceived. Interestingly, those who receive the most rein forcement are the least accurate in their estimates. Future Research The most interesting results for both predictive and estimating behavior come from the Reverse condition. Although this study seemed to indicate that male-internals maximized in the Reverse condition because they Were less bound by the external control of the chips, other explana tions exist. Future research might be directed toward consideration of the possibility that this group is actual ly less flexible and maximizing represents rigid or compul sive behavior. There are also two other possible Reverse conditions which should be studied. One is where the sub ject would lose,for being correct and win for being incor rect. The other is where he would stand on a correct pre diction and win on an incorrect prediction. An extremely interesting possibility is the use of probability learning tasks to study choice behavior among various psychiatric populations. The issue is whether dif ferent diagnostic groups would function differently and the 78 effect of type and amount of reinforcement on their choice behavior. Finally, the effect of the sex of the experimenter on probability learning should be studied. In the present investigation all subjects were tested by a female experi menter. The strongest experimental effects were obtained among the males. This suggests a possible interaction ef fect between the sex of the subject and the sex of the ex perimenter on probability learning. CHAPTER VI SUMMARY The main purpose of the study was to examine binary choice probability learning under five conditions of risk, with the level of risk held constant. All subjects were given 200 trials in which they predicted which of two lights, a red or a green one, would light up next. The lights were pre-programmed with the red one occurring three times more often than the green one. The sequence of oc currence was random. The conditions of risk were varied so that some subjects received no payoff other than learning whether or not their prediction was correct, some won a chip for being correct and lost for being incorrect (Win- Lose), some won for being correct and broke even for being incorrect (Win-Stand) and some broke even for being correct and lost for being incorrect (Lose-Stand). Finally, a unique Reverse condition was introduced where subjects lost for being correct and broke even for being incorrect. From previous evidence it was predicted that the frequency of predicting the red light would be least in the Reverse 79 80 condition and would increase by condition in the following order: No Payoff, Lose-Stand, Win-Stand, Win-Lose. This order represents the increasing amounts of reinforcement (positive and/or negative) available in each condition. The study was also concerned with whether the sex of the subject or his belief in Internal vs. External Control of Reinforcement would affect performance in the experimental situation. Following the experimental task, subjects were asked to estimate the frequency with which the red light had occurred. This was included on an exploratory basis. The results indicated that the frequency of predic tion of the red light increased from condition to condition in the exact order predicted. However, except for the Reverse condition, the differences were not great enough to be significant. It was also found that sex and internality- externality interacted significantly. In particular, male- internals tended more toward the optimal strategy of maxi mizing predictions of the red light than any other group. The most interesting finding occurred as a result of the Reverse condition. Some subjects chose to predict correct ly and lose chips while others chose to do exactly the op posite of what they had been instructed to do, in order to not lose chips. There were differences among the male in ternals and externals with all of the internals choosing to predict correctly. There were also differences among the 81 females but they were less clear-cut. The.results support the belief that internality-externality functions differ ently among men than women. Finally, the estimating behavior of subjects was analyzed. It was demonstrated that estimates vary with the experimental condition and tend to increase as the amount of reinforcement in the experimental condition increases. APPENDIX A INTERNAL-EXTERNAL QUESTIONNAIRE 83 This is a questionnaire to find out the way in which certain important events in our society affect different people. Each item consists of a pair of alternatives let tered "a** or ”b.” Please select the one statement of each pair (and only one) which you more strongly believe to be the case as tar as you're concerned. Be sure to select the one you actually believe to be more true rather than the one you think you should choose or the one you would like to be true. This is a measure of personal belief: obvi ously there are no right or wrong answers. Please print your name and the other information re quested. Then finish reading these directions and circle your answers to the items. NAME________________________________________ MALE___FEMALE___ MAILING ADDRESS_______________________________________________ LOCAL TELEPHONE NUMBER AGE OCCUPATION ARE YOU CURRENTLY A COLLEGE STUDENT? IF YES, NAME OF SCHOOL MAJOR FIELD FRESHMAN SOPHOMORE JUNIOR SENIOR GRADUATE STUDENT Please answer these items carefully but do not spend too much time on any one item. Be sure to find an answer for every choice. Circle item Ma” or "b"--whichever you choose as the more true statement. In some instances you may discover that you believe both statements or neither one. In such cases, be sure to select the one you more strongly believe to be the case as far as you're concerned. Also try to respond to each item independently when making your choice; do not be influenced by your previous choices. Children get into trouble because their parents punish them too much. The trouble with most children nowadays is that their parents are too easy with them. (Filler) Many of the unhappy things in-peoplefs lives are partly due to bad luck. People's misfortunes result from the mistakes they make. One of the major reasons why we have wars is be cause people don't take enough interest in politics. There will always be wars, no matter how hard peo ple try to prevent them. In the long run people get the respect they deserve in this world. Unfortunately, an individual's worth often passes unrecognized no matter how hard he tries. The idea that teachers are unfair to students is nonsense. Most students don't realize the extent to which their grades are influenced by accidental happen ings. Without the right breaks one cannot be an effective leader. Capable people who fail to become leaders have not taken advantage of their opportunities. No matter how hard you try some people just don't like you. People who can't get others to like them don't understand how to get along with others. Heredity plays the major role in determining one's personality. It is one's experiences in life which determine what they're like. (Filler) I have often found that what is going to happen will happen. Trusting to fate has never turned out as well for me as making a decision to take a definite course of action. 85 10. 11. 12. 13. 14. 15. 16. 17. 18. a. In the case of the well prepared student there is rarely if ever such a thing as an unfair test. b. Many times exam questions tend to be so unrelated to course work that studying is really useless. a. Becoming a success is a matter of hard work, luck has little or nothing to do with it. b. Getting a good job depends mainly on being in the right place at the right time. a. The average citizen can have an influence in gov ernment decisions. b. This world is run by the few people in power, and there is not much the little guy can do about it. a. When I make plans, I am almost certain that I can make them work. b. It is not always wise to plan too far aheadbe- cause many things turn out to be a matter of good or bad fortune anyhow. a. There are certain people who are just no good. b. There is some good in everybody. (Filler) a. In my case getting what I want has little or noth ing to do with luck. b. Many times we might just as well decide what to do by flipping a coin. a. Who gets to be the boss often depends on who was lucky enough to be in the right place first. b. Getting people to do the right thing depends upon ability, luck has little or nothing to do with it. a. As far as world affairs are concerned most of us are the victims of forces we can neither understand nor control. b. By taking an active part in political and social affairs the people can control world events. a. Most people don’t realize the extent to which their lives are controlled by accidental happenings. b. There really is no such thing as “luck.11 19. a. One should always be willing to admit mistakes, b. It is usually best to cover up one’s mistakes. (Filler) 86 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. a. It is hard to know whether or not a person really likes you. b. How many friends you have depends upon how nice a person you are. a. In the long run the bad things that happen to us are balanced by the good ones. b. Most misfortunes are the result of lack of ability, ignorance, laziness, or all three. a. With enough effort we can wipe out political cor ruption. b. It is difficult for people to have much control over the things politicians do in office. a. Sometimes I can't understand how teachers arrive at the grades they give. b. There is a direct connection between how hard I study and the grades I get. a. A good leader expects people to decide for them selves what they should do. b. A good leader makes it clear to everybody what their jobs are. (Filler) a. Many times I feel that I have little influence over the things that happen to me. b. It is impossible for me to believe that chance or luck plays an important role in my life. a. People are lonely because they don't try to be friendly. b. There's not much use in trying too hard to please people, if they like you, they like you. a. There's too much emphasis on athletics in high school. b. Team sports are an excellent way to build charac ter. (Filler) a. What happens to me is my own doing. b. Sometimes I feel that I don't have enough control over the direction my life is taking. a. Most of the time I can't understand why politicians behave the way they do. . In the long run the people are responsible for bad government on a national as well as on a local 87 level. WOULD YOU BE WILLING TO PARTICIPATE IN AN EXPERIMENT ON HOW PEOPLE BEHAVE IN GAMING SITUATIONS? IT WOULD REQUIRE ABOUT ONE HOUR OF YOUR TIME. YES NO APPENDIX B LETTER SENT TO SUBJECTS 88 89 Dear Thank you for volunteering to serve as a subject in my psychology experiment. This is just a reminder that your appointment is for DAY: DATE: TIME: Please come to: PSYCHOLOGICAL RESEARCH & SERVICE CENTER (Second Floor) (Kerkchoff Hall) 734 West Adams Boulevard, Los Angeles Tel: 746-2287 Should you be unable to keep this appointment, I would greatly appreciate being notified at 624-4616 or at the above phone number. Thanks again for your assistance. Yours truly, Marion Aronowitz APPENDIX C RAW DATA 90 91 TABLE 18 RAW DATA SHOWING'I-E SCORE, NUMBER OF TIMES RED LIGHT PREDICTED, AND ESTIMATES OF PERCENTAGE OF TIME RED LIGHT ACTUALLY OCCURRED Group I-E Predictions Estimated Percent 3lock 1 Block 2 Block 3 Block 4 Male-External 14 37 44 33 31 75 Condition 1 11 39 32 37 36 85 11 36 39 44 44 70 15 27 35 40 44 70 14 32 31 38 36 85 12 29 35 37 37 75 12 30 34 38 37 80 Male-External 14 33 40 42 44 85 Condition 2 17 25 36 33 34 75 13 40 48 50 45 80 11 31 40 46 47 75 15 22 33 36 41 85 14 27 43 42 44 75 13 33 40 41 42 75 Male-External 15 41 45 48 49 85 Condition 3 10 34 39 41 42 75 13 35 40 48 44 94 13 26 25 33 40 75 19 39 44 42 48 90 12 20 33 44 37 75 14 31 41 39 36 92 Male-External 13 29 47 43 46 65 Condition 4 15 39 45 47 42 70 11 26 34 32 29 65 11 35 39 45 44 90 16 25 36 40 30 70 13 29 37 42 42 70 15 35 35 41 43 60 92 TABLE 18 (Continued) Group I-E Predictions Estimated Percent Block Block Block Block 1 2 3 4 Male-External 14 22 12 9 10 75 Condition 5 11 22 19 12 9 80 12 12 6 2 5 67 15 23 14 14 12 80 19 30 38 39 43 90 16 26 39 39 46 80 15 38 38 39 38 80 Male-Internal 4 40 45 45 46 83 Condition 1 5 23 21 39 37 75 5 36 47 44 45 90 7 31 36 35 41 75 6 27 30 39 39 76 1 31 42 48 46 70 7 35 36 39 46 70 Male-Internal 4 37 44 48 47 90 Condition 2 3 41 39 44 42 90 3 31 37 44 44 80 5 35 41 41 40 90 4 23 39 39 39 70 6 25 34 33 39 75 6 27 41 39 40 85 Male-Internal 4 25 31 30 37 90 Condition 3 1 36 41 43 44 92 5 29 39 42 42 85 5 30 41 40 42 80 1 29 33 33 34 70 1 25 39 45 47 80 5 23 30 39 37 80 Male-Internal 2 26 34 40 38 75 Condition 4 4 22 32 35 37 70 4 20 28 32 38 75 5 26 37 38 41 70 5 35 40 46 42 75 3 34 44 41 44 90 4 29 47 48 50 75 93 TABLE 18 (Continued) Group I -E Predictions Estimated Percent Block Block Block Block 1 2 3 4 Male-Internal 4 34 37 38 41 75 Condition 5 2 30 41 45 44 70 2 38 44 50 50 90 6 33 41 45 43 75 3 38 39 44 42 85 3 37 41 44 40 80 5 37 43 49 45 93 Female-External 14 30 32 38 36 70 Condition 1 17 27 30 38 37 75 12 25 29 41 42 75 14 33 39 42 42 85 15 39 43 48 46 80 19 23 29 35 34 80 15 27 36 30 37 75 Female-External 15 26 34 40 43 80 Condition 2 14 24 40 43 45 80 14 29 37 43 44 95 16 33 36 40 38 85 18 17 22 37 43 75 15 32 36 36 . 36 85 17 25 41 39 44 90 Female-External 23 34 38 44 40 80 Condition 3 18 33 41 45 46 85 17 30 37 35 36 75 15 23 28 32 30 75 15 32 48 49 49 60 15 29 35 41 39 75 20 37 41 45 41 80 Female-External 12 35 39 39 36 75 Condition 4 15 32 36 40 32 70 21 22 30 35 34 80 15 30 40 43 40 75 14 27 39 46 45 75 18 38 43 40 45 80 16 34 41 47 47 70 94 TABLE 18 (Continued) Group I -E Predictions Estimated Percent Block Block Block Block 1 2 3 4 Female-External 14 15 11 7 9 75 Condition 5 13 38 35 36 32 75 17 20 28 42 38 70 19 24 23 30 31 60 17. 28 31 39 39 75 19 29 28 26 38 75 18 42 41 47 47 88 Female-Internal 5 36 41 44 42 60 Condition 1 4 33 42 42 44 80 4 35 37 38 36 65 7 30 33 43 37 85 6 28 34 46 38 80 3 29 37 36 31 65 2 32 37 42 42 60 Female-Internal 7 26 41 39 43 85 Condition 2 5 31 39 42 47 75 8 37 41 42 43 90 4 22 32 33 35 70 9 30 36 43 42 85 6 29 35 39 39 70 7 34 37 38 35 75 Female-Internal 6 41 46 49 43 90 Condition 3 6 32 39 44 44 80 8 35 43 48 40 80 5 30 36 36 43 80 3 36 47 49 44 70 3 28 42 48 48 75 4 35 42 41 40 80 Female—Internal 6 25 34 36 39 85 Condition 4 8 35 35 38 42 80 3 22 34 38 38 80 6 33 35 36 36 75 8 20 30 41 42 80 4 39 40 50 47 95 0 35 42 49 47 85 TABLE 18 (Continued) 95 Group I -E Predictions Estimated Percent Block Block Block Block 1 2 3 4 Female-Internal 7 38 41 43 43 75 Condition 5 6 25 9 9 11 75 8 5 5 2 4 75 7 23 36 29 27 80 3 29 28 48 19 65 3 22 22 16 7 75 7 34 39 44 41 85 APPENDIX D POST EXPERIMENTAL INTERVIEWS TABLE 19 97 POST EXPERIMENTAL INTERVIEW SUBJECTST COMMENTS ON THEIR BEHAVIOR No Payoff--Male Externals_____________________________________ 1. Kept working systems. Always failed. Near the end, stayed with red. Seemed better than half the time— decided to try it. 2. At first random guessing. Then looked for pattern-- thought I had one. Towards end, seemed like a fixed pattern. 3. A first, tried various procedures. Later, picked ac cording to feeling for the percentages. 4. Noticed right away that red occurred more. Tried to guess mostly red and estimate when green seemed due. Not counting reds and greens. “Waited for sensations.** 5. Trying to find system, but didn’t work. By the end it seemed random. Picked green by sense. Hated being wrong about green. 6. Beginning--picked randomly. Noticed sequences but did not count. Selected by feel. Looking for a solution. 7. Tried different patterns. Never found one that worked. No Payoff--Male Internals_____________________________________ 1. Was trying to solve the problem. 2. Tried different plans. Got frustrated. Decided there either was no plan or it was too complicated to figure out. 3. Started looking for a pattern. Counted reds and greens at first. Toward the end, still seeking pattern, but just guessed when green due. 4. “Tried to figure out what the machine was thinking. How it responded to my pushing buttons.” TABLE 19 (Continued) 98 No Payoff--Male Internals (continued)_______________________ 5. Started counting about half way through. Seemed to be a pattern, but it seldom repeated. 6. Beginning, thought there must be a pattern--didn't have a chance if it was random. Decided that he was best off if he got red most of the time. Trying to predict green was bad. When felt green was due, just hit it. 7. At beginning, switched back and forth. Noticed red predominated. Basically went to red. Stopped counting. Had feeling about how long the sequence would run. No Payoff--Female Externals_________________________________ 1. Decided right away that it was just a matter of guess ing. 2. Pressed red because green didn't come up often. 3. Thought there was a pattern. That didn't work. Went on intuition. 4. Decided there must be a pattern but could not figure it out. Stopped counting. Switched back and forth on guesses. ’’ Maybe ESP.” 5. Realized reds came up more often after a while. Felt there was a reason. Thought she found pattern, but then it changed. Just picked green by feel after a while. 6. Hit green according to number of red that had appeared, but had no specific plan. 7. Tried for a pattern but gave that up. That it might be contingent upon her own responses. Gave that up too. By the end, just hit green when it seemed due. No Payoff— Female Internals____________________________ 1. Could not find a pattern so picked red most of the time. TABLE 19 (Continued) 99- No Payoff--Female Internals (continued)_____________________ 2. Tried many hypotheses, e.g., maybe white light had something to do with it, maybe sound of instruments had something to do with it. Finally decided to stick mostly with red. 3. Tried all the way through to figure a pattern. Thought it was fun but sometimes frustrating. 4. At first thought it was random. Then looked for a pat tern. Did not count. “Guess I need a green now.” 5. Looked for a pattern. Sometimes had it, sometimes did not. 6. Looked for a pattern after a while. Decided that if she picked incorrectly, had to go to other light to correct it. Finally, decided there was no pattern. ’They wouldn't make a test for us where there was a pattern you could adapt to.” 7. Obviously more red, so often picked red. Looked for patterns but did not really find one. Seemed harder to figure out at end than beginning. Win-Lose--Male Externals 1. First figured it was red light on a majority basis. Then tested to see if machine contingent upon his re sponses. Then ”1 was trying to guess what they would guess I would guess.” 2. Looked for a pattern. Counted at first. Then felt “innately” when green would come. Stopped counting after a while. 3. Tried to find a relationship between the two lights. Tried many different strategies right until the end. 4. Assumed red occurred 75 to 80 per cent of the time so he hit green every fifth time. 5. At first just used trial and error. Then used “intui tion as to when green due.” Did not count. 6. Looked for a sequence. Could not find one. TABEE 19 (Continued) 100 Win-Lose--Male Externals (continued) 7. Hit red more than green. Counted for a while, then stopped. Win-Los: e--Male Internals 1. Tried to find a pattern. Counted. Then decided that there was a pattern but it was erratic. Kept trying patterns until the end. 2. Seemed like a pattern after a while. Started counting about one third of the way through. Mainly stayed with red toward end. If it seemed like red was going too long, hit a green. 3. Tried to stick with red and develop a pattern. Usually did not count. 4. Tried to look for some point about the task-considered probabilities. Eventually I decided to go to mainly reds. 5. Started by looking for a pattern. Could not find one so mostly pressed reds. 6. Saw a pattern. Was never positive about it but had a general feeling for it. 7. Tested out different patterns. Win-Lose--Female Externals 1. Red came up the most, so pressed it the most. Usually did not count. 2. Haphazard in beginning. Noticed red came up more. Then started counting. 3. Became obvious that it was better to stay with red— would have more chips in the end. Went through a peri od of staying with red. Then tried to calculate when green would show up. Did not count. 4. Looked for a pattern. Did not count. TABLE 19 (Continued) 101 Win-Lose--Female Externals (continued) 5. Realized red occurred most of time so hit it because it was less risky. Then started to hit more greens for a while because she felt red would give out on her. 6. Hit mostly reds. Every sixth to eighth one, hit a green. 7. No particular plan. Mostly hit reds. After a long string of reds, hit a green. Win-Lose--Female Internals 1. Counting did not work. Just arbitrarily chose green periodically. 2. f,Tried to count reds and greens. Gave up and followed whatever I felt like.” 3. Tried out various series. Did this all the way through. 4. Tried to find a pattern until the very end. More sat isfaction in seeing the solution than in just being correct every time. 5. Hit reds more and when she felt green was due, she hit that button. 6. Tried all the way to figure out if there was a se quence. Did not find any one pattern. 7. Did not count. Just hit green after long string of reds. Win-Stand--Male Externals 1. Tried different patterns. Stuck with red most of time after realizing it came up most. Starting counting about midway. 2. More reds than green. Seemed like there would be a pattern. Looked for one until end. TABLE 19 (Continued) 102 Win-Stand— Male Externals (continued) 3. Tried many different plans, e.g., tried to find a pat tern. Then decided that if there was one he was too stupid to find it. Tried to see if timing of lights had anything to do with it, tested the machine sound to see if related to lights, tried to use ESP. 4. Picked almost all reds. “Stand with a winner. You can’t beat the machine.” 5. Tried to seek a system at first. Then decided to hit almost all reds. 6. Switched to green about every fourth light. 7. Counted. Tried different systems. Then gave up and hit mostly reds. Win-Stand--Male Internals 1. Picked long runs of reds. No particular plan. 2. Looked for patterns, but they seemed to vary. Enjoyed being correct on greens. 3. Tried various systems all the way to the end. 4. Tried to find a system. Tested different ones through out . 5. Looked for patterns. Also tested to see if lights were contingent on his responses. 6. Looked for patterns and mathematical sequences. Started counting later on. Did better after started counting. 7. Started with idea that there was a pattern. Did not count. Decided by the end that it really was random. Win-Stand--Female Externals 1. Tried different patterns. By the end she felt that the only gamble was after a green— whether it would return j to red or be green two times in a row. TABLE 19 (Continued) 103 Win-Stand--Female Externals (continued) 2. Tried to solve the problem. 3. Tried various mathematical series. Then tried sub tracting. Finally ended up with mostly reds and never more than two greens in a row. 4. Seemed like a pattern for a while--tried to follow it. Did not count--just approximated. Near the end, just hit red and occasionally hit green because it seemed like a pattern. 5. At first just did what she felt. Then looked for pat terns. Kept winning on reds so decided to stick with it. 6. Red went for long periods. Did not really count. Just estimated. 7. Intuition, guessing. Sometimes pressed green and did not know why. Tried plans in the beginning. Did not work so just pressed red more. Win-Stand--Female Internals 1. Red occurred about 80 per cent so picked it most. Every once in a while pushed green--after about 7 or 8 reds. nUsually I was too early on green and lost two chips.” 2. Started to count. No pattern, so stopped. Just hit green by feel. 3. Tried a system. Then stayed with red. 4. Started looking for patterns. They would seem to work only once or twice. Sometimes seemed to be a progres sion. Tried to see if sound of buzzer was related. Towards end, chose by what seemed logical, or by in stinct. 5. When red came up so much, just stayed with it because was gaining so much. Losses were not significant. 6. Could not find a series so stopped looking for one. Just hit red most of the time. TABLE 19 (Continued) 104 Win-Stand--Female Internals (continued) 7. Once realized red came up more, she just hit that. Then it got boring so she tried to outguess the se quence. Lose-Stand— Male Externals 1. At first looked for a sequence. Then just picked red mostly. 2. At first assumed a pattern that had nothing to do with his response. Then it became obvious red occurred more often so he mostly picked red. Lost when he tried to guess green. 3. Tried to detect a pattern. At first seemed like he had one. Then gave up and just guessed. 4. Looked for a pattern that would repeat. Could not find one. Knew that there was least error in hitting reds but wanted to catch the green light after a long series of reds. 5. Tried to decide if it was a matter of probability or it was programmed. Finally decided to stay with a winner and hit reds. 6. Usually hit green every third or fourth trial. Decided at the end that there was no set sequence. 7. At first tried a scheme but it broke down. Then went basically to reds with a single or doublt green occa sionally. Lose-Stand--Male Internals 1. Counted and looked for a pattern at first. Felt good when he could outguess the green. The red was easy, 2. Looked for a ratio. Did not find one. 3. Kept looking for patterns until the very end. None worked. TABLE 19 (Continued) 105 Lose-Stand--Male Internals(continued) 4. Positive there was a pattern. At first just wanted to find it. Then got enthused about trying to get more chips than the experimenter. Kept watching both our stacks. Out to beat the machine "even though I know it is programmed and has no intelligence." Also thought about others who had done this task. Tried to compete with them and get more chips than they might have got ten. 5. Assumed there was a pattern and kept looking for it. 6. Tried different combinations. Tested to see if buzz of the machine was related. 7. Seemed like it was in multiples of three. Though the buzz might have something to do with it. Basically stayed with every third time making a change. Lose-Stand--Female Externals 1. Tried to establish a set pattern but could not. Did not count. 2. Tried counting at first. Then gave it up but still looked for a pattern. 3. Counted for a while. Really could not figure it out. 4. Looked for a series. Did not find one. Just guessed on the greens after a while. 5. Seemed there was a certain time span between greens but couldn’t figure it out. Didn’t always count. 6. Mostly just hit reds. Did not count. Did not seem to see a pattern. 7. Became obvious red came on more often. Then looked for a pattern. Did not find one. Then went back to mostly pushing reds. Eose-Stand--Female Internals 1. Mainly hit red. Tried counting each color but did not have any plan. __ . • . TABLE 19 (Continued) 106 Lose-Stand--Female Internals (continued) 2. Tried to solve the problem. Got very frustrated when wrong. 3. Tried to find a pattern but never did. 4. Was sure that there was a pattern in the middle. Pressed red sometimes when she was sure it was green. Doesn't know why she did not press green. 5. Tried to look for a pattern. Felt the odds were on the red. Eventually, only hit green when it followed an other green. 6. Realized red came on most and just hit it most after that. Felt she won more doing that than trying to out guess green. 7. Tried to figure out the pattern. Thought that it would be better to hit red all the time but felt it was more of a challenge to try to figure it out. Reverse--Male Externals 1. Tried to find a pattern. Saw his chips depleting, so bet green. 2. Did not want to lose chips, so after a while he hit green mostly. 3. Looked for a pattern but never found one. “Chips had no significance to me. Only the lights were of inter est. But if it were really my own money, I would try to predict incorrectly.” 4. ”To me, being wrong was being right.” Seemed awkward at first. Just pressed light and hoped he was wrong. Hit the reds on impulse. 5. Thinks there was a pattern and tried to figure it out all the way to the end but did not have much luck. It never occurred to him to keep the chips. Tried to lose: as many as he could. Interpreted task as a game--see how many chips he could lose. TABLE 19 (Continued} 107 Reverse--Male Externals (continued} 6. “You said this was a gambling experiment so it seemed more logical not to lose.” Stayed with green most of the time. 7. Tried to match the lights as often as he could. Tried to pile up as many chips as he could on ET s side. ”It was like I was gaining them instead of losing them. Counted at first but it did not work. Then just mostly hit red. Reverse--Male Internals 1. Tried to find patterns. The chips did not really mat ter. They weren't really his. Only part of an experi ment so they didn't matter. 2. Tried to predict the correct light. Figured out the percentage of greens and used number lost as an indi cator of how well he was doing. 3. Kept pressing red mostly after he realized it occurred more often. "Predicting correctly seemed right. I figured the object was to get rid of the chips like in certain card games. 4. At first it seemed like I wouldn't want to give away chips--but then he got to waiting for E to take them. Saw basic task as getting rid of chips. Mostly pressed red. If buzzer seemed louder on machine, pressed green. 5. ”1 did what you told me. The chips didn't matter.” Goal was to find the pattern. 6. ”1 wanted to guess right. The chips had no meaning. It was being correct that counted.” 7. Tried to out-think the machine at first. Then just started pushing red. Made immediate decision to give up chips. TABLE 19 (Continued) 108 Reverse--Female Externals 1. ”At first it bothered me. Then I got into a thing of it--I moved a chip (i.e., across to E) and I won." Did not count. Felt there was a rhyThm and picked that way. 2. ”It really bothered me in the beginning. When you win you want to keep something. If I lost a chip to you (E), I was still thinking about it by the next trial. I thought about predicting the other light--then de cided not to. I don’t know why. Then I got used to giving them to you. Then I wanted them on your side.” Counted all the way through. Looked for patterns. 3. Decide to get rid of chips. That would be winning. Mostly hit reds and hit a green when she got bored. 4. ”1 have a natural bias toward red. I have a flashy red car. But in gambling, green is money. I pictured it that the less chips I have, the better it is. I twisted it in my mind. I really didn’t use a strategy. The first color that comes into my brain--that ’ s it. Just watch the white light and whatever hits my mind.” 5. Looked for patterns and counted all the way. Wanted to get the right light. Chips did not matter. 6. Did it mostly by intuition. Only counted for a short time. Winning was predicting correctly and losing chips equalled winning. 7. ”1 thought the idea was to see how many chips I could keep. If a green light came on she hit red once and then went back to green.” Reverse--Female Internals 1. ”1 wanted to pick the right one but I wanted to have the most chips. I kind of tried picking the right one more. It got so I’d rather not have the chips after a while. The main thing became to pick correctly.” 2. ”At first I didn’t know. Then I decided that when you gamble you usually want to keep them (i.e., chips). That means winning.” Counted for a while and then stopped. TABLE 19 (Continued) 109 Reverse--Female Internals (continued) 3. "At first it was the more chips you (E) had was win ning. Then it was keeping the chips.” If a light came up she stayed with it for a while. Did not count. Had no idea if there was a pattern or not. 4. Wanted to be right as often as possible. Pressed red mostly. 5. Tried to find a pattern. Counted all the way. Felt there was a pattern but she just could not figure it out. 6. ”1 played it that if I kept the chip I won, because I like the chips.” 7. Tried to keep chips. Thinks there was a sequence. Did not count. LIST OF REFERENCES 110 REFERENCES Abelson, R. P. , f The Choice of Choice Theories,” in S. Messick and A. Brayfield (eds.), Decision and Choice. New York: McGraw-Hill, I9'64J Baron, R. A. “Authoritarianism, Locus of Control and Risk Taking,” Journal of Psychology, 1968, 68, 141-143. Becker, G. M. and McClintock, C. G. “Value: Behavioral Decision Theory,” Annual Review of Psychology, Vol. 18. Palo Alto: Annual Review, Inc., 1967. Bennion, R. C. ”Task, Trial by Trial Score Variability of Internal vs. External Control of Reinforcement.” Unpublished doctoral dissertation, Ohio State University, 1961. Blackman, S. “Some Factors Affecting the Perception of Events as Chance Determined," Journal of Psychology, 1962, 54, 197-202. Bush, R. R. and Mosteller, F. Stochastic Models for Learn ing. New York: Wiley and Sons, 1955. Butterfield, E. C. "Locus of control, test anxiety, reac tions to frustration and achievement attitudes,” Journal of Personality, 1964, 32, 298-311. Cardi, M. “An examination of internal vs. external control in relation to academic failures." Unpublished master*s thesis, Ohio State University, 1962. Chipman, J. S. "Stochastic Choice and Subjective Probabil ity," in D. Willner (ed.), Decisions, Values and Groups, Vol. 1. New York: Pergamon Press, 1960. Cohen, J. Behavior in Uncertainty: Its Social Implica- tiotH New York: Basic Books, 1964. Ill 112 Deacon, S. "The--effect of feedback on communication of medical prescription to diabetic patients," Unpublished doctoral dissertation, University of Southern California, 1968. Edwards, W. "Behavioral Decision Theory,” Annual Review of Psychology, Vol. 12. - Palo Alto: Annual Review, Inc., 1961 (a). _. f*The Prediction of Decisions Among Bets," Journal of Experimental Psychology, 1955, 50, 201- ' T D T .----------- “ _. "Probability Learning in 1000 Trials,” Journal of Experimental Psychology, 1961, 62, 385-394. Erickson, J.R., O’Hara, J. W., and Jones, M. R. "Effects of a Memory Aid in Probability Learning," Psycho- nomic Science, 1968, 31), 137-138. Estes, W. K. "Theory of Learning with Constant, Variable or Contingent Probability of Reinforcement," Psychometrika, 1957, 22, 113-132. Goodnow, J. J. and Postman, L. "Learning in a Two Choice Probability Situation with a Problem Solving Set ting," Journal of Experimental Psychology, 1955, 49, 16-22. Gore, P. M. and Rotter, J. B. "A Personality Correlate of Social Action," Journal of Personality, 1963, 31, 58-64. Grant, D. A., Hake, H. W., and Hornseth, J. P. "Acquisi tion and Extinction of a Verbal Conditioning Re sponse with Differing Percentages of Reinforce ment," Journal of Experimental Psychology, 1951, 42, 1-5. Green, P. C. "Three Choice Probability Learning: Effect of Varying Information Level,” Psychonomic Science, 1966, 6, 267-268. Hake, H. W. and Hyman, R. "Perception of the Statistical Structure of a Random Series of Binary Symbols," Journal of Experimental Psychology, 1953, 45, 64- T%~. Hays, W. E. Statistics for Psychologists. New York: Holt, Rinehart, and Winston, 1^63. 113 Hersch, P. D. and Scheibe, K. E. "Reliability and Validity of Internal-External Control of Reinforcement as a Personality Dimension," Journal of Consulting Psychology, 1967, 3_1, 609™6T3"1 Holden, K. B. and Rotter, J. B. "A Non-verbal Measure of Extinction in Skill and Chance Situations," Journal of Experimental Psychology, 1962, ji3, 519-52(5"! Humphrey, L. G. "Acquisition and Extinction of Verbal Expectations in a Situation Analagous to Condition ing," Journal of Experimental Psychology, 1939, 25, 294-301; James, W. H. "Internal vs. External Control of Reinforce ment as a Basic Variable in Learning Theory," Unpublished doctoral dissertation, Ohio State University, 1957. James, W. H. and Rotter, J. B. "Partial and 100% Rein forcement Under Chance and Skill Conditions," Journal of Experimental Psychology, 1958, 55, 397- WT. James, W. H., Woodruff, A. B. and Werner, W. "Effect of Internal and External Control Upon Changes in Smoking Behavior," Journal of Consulting Psychology, 1965, 29, 184-186. Jarvik, M. G. "Probability Learning and Negative Recency Effect in Serial Anticipation of Alternative Symbols," Journal of Experimental Psychology, 1951, 41, 291-297. Jones, M. R. and Myers, J. L. ”A Comparison of Two Methods of Event Randomization,” Journal of Experimental Psychology, 1966, 72, 909-911. Julian, J. W. and Katz, S. B. "Internal vs. External Control and the Value of Reinforcement," Journal of Personality and Social Psychology, 1968, W, 89-94. Kogan, N. and Wallach, M. Risk Taking: A Study in Cogni tion and Personality! New York: Holt, Rinehart, and Winston, 1964. Ladwig, G. W. "Personal, Situational and Social Determi nants of Preference for Delayed Reinforcement." Unpublished doctoral dissertation, Ohio State University, 1963. 114 Lefcourt, H. M. "Internal vs. External Control of Rein forcement: A Review,” Psychological Bulletin, 1966, 65, 206-220. Liverant, S. and Scodel, A. "Internal and External Control as Determinants of Decision Making Under Conditions of Risk," Psychological Reports, 1960, 7, 59-67. Marston, A. R. "Personality Variables Related to Self reinforcement,” Journal of Psychology, 1964, 58, 169-175. Myers, J. L. Fundamentals of Experimental Design. Boston: Allyn and Bacon, 1966. Phares, E. J. "Expectancy Changes in Skill and Chance Situations," Journal of Abnormal and Social i Psychology, 1957, &T, 339-342. "Internal-External Control as a Determinant of Amount of Social Influence Exerted," Journal of Personality and Social Psychology, 1965, 642- T W . Rapoport, A. "Introduction," in D. Willner (ed.), Decision, Values and Groups. New York: Pergamon Press, 1960. Restle, F. Psychology of Judgment and Choice: A Theoretical Essay. New York: Wiley and Sons, v z z r . ----------------------: --------- Rotter, J. B. "Beliefs, Social Attitudes and Behavior: A Social Learning Analysis," in R. Jessor and S. Feshbach (eds.), Cognition, Personality and Clinical Psychology. San Francisco: Jossey Bass, i w : --------------- _________. "Generalized Expectancies for Internal vs. External Control of Reinforcement," Psychological Monograph, 1966, jSO (1, whole no. 609). Social Learning and Clinical Psychology. New York! Prentice-Hall, 1954. Rotter, J. B., Liverant, S., and Crowne, D. P. "The Growth and Extinction of Expectancies in Chance Controlled and Skilled Tests," Journal of Psychology, 1961, 52, 161-177. 115 Rotter, Schnorr Scodel, Seeman, Seeman, Seigel, Seigel, Shuford Stevens, J. B. and Mulry, R. C. nInternal vs. External Control of Reinforcement and Decision Time,” Journal of Personality and Social Psychology, 1965, 2, 598-604. , J. A., Lipkin, S. G., and Myers, J. L. "Level of Risk in Probability Learning: Within and Between Subjects Designs," Journal of Experimental Psychol ogy 1966, 72, 497-500. A., Ratoosh, P., and Menos, J. A. "Some Personali ty Correlates of Decision Making Under Conditions of Risk," in D. Willner (ed.), Decisions, Values and Groups, Vol. 1. New York: Pergamon Press, 1960. M. "Alienation and Social Learning in a Reforma tory," American Journal of Sociology, 1963, 69, 270-284. M. and Evans, J. W. "Alienation and Learning in a Hospital Setting," American Sociological Review, 1962, 27, 772-783. S. "Decision Making and Learning Under Varying Conditions of Reinforcement,1* in S. Messick and A. Brayfield (eds.), Decision and Choice. New York: McGraw-Hill, 1964 T^T, "Theoretical Models of Choice and Strategy Be havior," in S. Messick and A. Brayfield (eds.), Decision and Choice. New York: McGraw-Hill, T 9 W <by. ----- S. and Goldstein, D. A. "Decision Making Behavior in a Two Choice Uncertain Outcome Situation,** Journal of Experimental Psychology, 1959, 57, 37- 4Tm , E. H. "A Comparison of Subjective Probabilities for Elementary and Complex Events.** Report #20, The Psychometric Laboratory, University of North Carolina, Chapel Hill, North Carolina, 1959. S. S. and Galanter, E. H. "Ratio Scales and Cate gory Scales for a Dozen Perceptual Continua," Journal of Experimental Psychology, 1957, 54, 377- W T . 116 Straits, B. C. and Sechrest, L. "Further Support of Some Findings About Characteristics of Smokers and Non- smokers," Journal of Consulting Psychology, 1963, 27, 282. Strickland, B. R. trThe Relationship of Awareness to Verbal Conditioning and Extinction." Unpublished doctoral dissertation, Ohio State University, 1962. Tune, G. G. "Response Preferences: A Review of Some Rele vant Literature," Psychological Bulletin, 1964, 61, 286-302. Tversky* A. and Edwards, W. "Information vs. Reward in Binary Choices," Journal of Experimental Psychology, 1966, 71, 680-683“ von Neumann, J. and Morgenstern, 0. Theory of Games and Economic Behavior. Princeton: Princeton Universi ty Press, 1947. Wolin, B. R., Weichel, R., Terebinski, S. J., and Hansford, E. A. "Performing on Complexly Patterned Binary Event Sequences,1* Psychological Monographs, 1965, 79 (7, whole no. 6001.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Experimenter Expectancy Effect Examined As A Function Of Task Ambiguity And Internal Versus External Control Of Reinforcement

PDF

A Cognitive Dissonance Analysis Of Conformity Behavior, As Applied To The full Denture Patient

PDF

The Effect Of Discriminability On The Partial Reinforcement Effect In Human Gsr Conditioning

PDF

The Effects Of Nonreward Dissonance And Secondary Reward On Extinction And Attractiveness

PDF

Effects Of Group Behavior Therapy Imagery On Basketball Performance

PDF

An Experimental Analogue Comparing Two Models Of Psychotherapy: Reciprocal Inhibition And Implosive Therapy

PDF

Imagery And Response Styles In Desensitization

PDF

Effects Of Different Treatment Procedures On Reading Ability And Anxiety Level In Children With Learning Difficulties

PDF

The Effects Of Two Types Of Experimenter Intervention And Schedules Of Reinforcement On Verbal Operant Conditioning Of Affective Self-References

PDF

The Effect Of Anxiety And Frustration On Muscular Tension Related To The Temporomandibular-Joint Syndrome

PDF

Weight Reduction As A Function Of The Timing Of Reinforcement In A Covertaversive Conditioning Paradigm

PDF

Stimulus and response generalization of classes of imitative and nonimitative behavior as a function of reinforcement, task, cues, and number of therapists

PDF

The Effects Of A Self Shock Procedure On Hallucinatory Activity In Hospitalized Schizophrenics

PDF

The Effects Of Anxiety And Threat On Self-Disclosure

PDF

The Differential Effectiveness Of External Versus Self-Reinforcement On The Acquisition And Performance Of Assertive Responses

PDF

The Effects Of Cognitive Motivation To Avoid Awareness On Measures Of Performance And Awareness In Verbal Conditioning

PDF

The Enhancement Of Eeg - Alpha Production And Its Effects On Hypnotic Susceptibility

PDF

Role Playing Ability And Social Adjustment In Children

PDF

The Influence Of The Achievement Motive, The Affiliation Motive, And Incentive Conditions On Roleplaying Ability In Children

PDF

Conflict And Generalized Conflict In Verbal And Motor Responses

Asset Metadata

Creator Aronowitz, Marion Kramer (author)

Core Title The Effect Of Conditions Of Risk, Internal Versus External Control Of Reinforcement, And Sex On Binary Choice Probability Learning

Degree Doctor of Philosophy

Degree Program Psychology

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

Tag OAI-PMH Harvest,Psychology, clinical

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Tiber, Norman (committee chair), Levitan, Seymour (committee member), Slucki, Henry (committee member), Wexler, Murray (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-375422

Unique identifier UC11362104

Identifier 7000339.pdf (filename),usctheses-c18-375422 (legacy record id)

Legacy Identifier 7000339.pdf

Dmrecord 375422

Document Type Dissertation

Rights Aronowitz, Marion Kramer

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 au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA