14
SOME CORRELATES OF RISK TAKING LAWRENCE K. WILLIAMS CorneU University Introduction RECENTLY, there has been considerable interest in distin- guishing among individuals and their propensity to take risks. An increased concern with decision-making processes has cre- ated a need to further identify individual orientations toward decision making including the assumption of risk. Much of the research on individual risk taking has its origins in at- tempts to elaborate formal decision-making models. This has involved the examination of subjective utility and probability which in turn has led to an interest in such problems as the assessment of individual differences with regard to probability preference (e. g., Edwards, 1954a and 1954b). Another line of research has its origin in motivation theory. The research of Atkinson and associates, which is focused on need achievement and fear of failure as correlates of risk taking, is an important example of this second approach (Atkinson, 1957, 1958; Atkin- son & Litwin, 1960; Atkinson, et al., 1960). These and other lines of research have all resulted in specific tests which have successfully discriminated among individuals with regard to predicting at least one cognitive or behavioral variable associ- ated with risk taking. Slovic (1962), however, has noted the lack of convergent validation of these tests, including the one to be reported here, when the same criteria are used across all tests. An implicit assumption in much of the contemporary re- search is that one’s orientation toward risk is pervasive and affects all of his perceptions and behavior. This is strictly an assumption, and the lack of convergent validation may result from the tests’ appropriateness for only specific areas of be- havior. A second and related problem rests with the criteria that are used to validate such tests. Much of the research has 297

Some Correlates Of Risk Taking

Embed Size (px)

DESCRIPTION

Some Correlates Of Risk Taking

Citation preview

Page 1: Some Correlates Of Risk Taking

SOME CORRELATES OF RISK TAKING

LAWRENCE K. WILLIAMS

CorneU University

Introduction

RECENTLY, there has been considerable interest in distin- guishing among individuals and their propensity to take risks. An increased concern with decision-making processes has cre- ated a need to further identify individual orientations toward decision making including the assumption of risk. Much of the research on individual risk taking has its origins in at- tempts to elaborate formal decision-making models. This has involved the examination of subjective utility and probability which in turn has led to an interest in such problems as the assessment of individual differences with regard to probability preference (e. g., Edwards, 1954a and 1954b). Another line of research has its origin in motivation theory. The research of Atkinson and associates, which is focused on need achievement and fear of failure as correlates of risk taking, is an important example of this second approach (Atkinson, 1957, 1958; Atkin- son & Litwin, 1960; Atkinson, et al., 1960). These and other lines of research have all resulted in specific tests which have successfully discriminated among individuals with regard to predicting at least one cognitive or behavioral variable associ- ated with risk taking. Slovic (1962), however, has noted the lack of convergent validation of these tests, including the one to be reported here, when the same criteria are used across all tests.

An implicit assumption in much of the contemporary re- search is that one’s orientation toward risk is pervasive and affects all of his perceptions and behavior. This is strictly an assumption, and the lack of convergent validation may result from the tests’ appropriateness for only specific areas of be- havior. A second and related problem rests with the criteria that are used to validate such tests. Much of the research has

297

Page 2: Some Correlates Of Risk Taking

298 PERSONNEL PSYCHOLOGY

been based on gambling problems, often conducted in a labora- tory setting. Participation in an activity where the outcome is strictly a matter of chance, a.nd where the individual has no control, appears to be psychologically quite different from participation where the individual takes a “calculated risk” and where he seemingly exercises some control over the out- come. Poker players, for example, are often disinterested in a process such as roulette or the lottery where the outcome is apparently capricious or beyond the individual’s control. It is possible that the laboratory setting may enhance the percep- tion of capriciousness in some of the experiments.

The studies to be reported here comprise a series of attempts to relate a measure of risk-taking propensity, the Job Prefer- ence Inventory (JPI) , to various attitudes and behaviors within an industrial envir0nment.l With regard to the second issue raised, i. e., gambling versus risk taking, the items were designed to present a continuum ranging from minimum pro- pensity to take risk to a propensity to take the “good” calcu- lated risk. The exclusion of pure gambling choices from the continuum is consistent with the risk-taking model suggested by Atkinson and Litwin (1960) and Atkinson, Bastian, Earl, and Litwin (1960), where the findings included the observation that high need achievers indicate a preference for moderate or “calculated risk” while low need achievers with high fear of failure indicate a preference for either very conservative or very risky bets. As such, endorsement of gambling or very risky bets cannot be construed as opposite in their predictive consequence from endorsement of security seeking or non- risk-taking choices.

Finally, it should be noted that the form of the present test was dictated by the need to develop a test which was brief and which could be included in a pencil-and-paper questionnaire to be used in industrial organizations. The five studies re- ported here summarize a series of research projects where both behavioral and attitudinal correlates of the inventory were examined and where maximally different populations of work- ers were included.

‘The original scaling work for this project was done as part of the au- thor’s Ph.D. thesis (Williams, 1960).

Page 3: Some Correlates Of Risk Taking

LAWRENCE K. WILLIAMS 299

The Job Preference Inventory Measure

Following pretesting and editing of items, eight pairs con- cerning alternatives of job preference were selected (See Fig- ure 1). Guttman scaling operations were performed on succes- sive samples of one hundred: two homogeneous and two heterogeneous (job level and sex) samples (Guttman, 1947a, 1947b; Stouffer, et al., 1950). A quasi scale was obtained with reproducibility coefficients ranging between .81 and .85. While the removal of two items raised the reproducibility coefficients

FIGURE 1. JOB PREFERENCE INVENTORY All of us have different requirements for the job that we would find most

attractive. The following are a number of alternatives that you might be faced with in considering job opportunities. Please check one alternative in each of the following pairs.

The kind of job that I would most prefer would be: 1. Check one:

-(1) A job where I am almost always on my own

-(2) A job where there is nearly always someone available t o help me on problems I don’t know how t o handle

-(1) A job where I have t o 2. Check one:

make many decisions by myself

-(2) A job where I have t o make few decisions by myself

3. Check one: -(1) A job where my instruc-

tions are quite detailed and specific

tions are very general -(2) A job where my instruc-

4. Check one: -(1) A job where I am almost

always certain of my abil- i ty to perform well

-J2) A job where I am usually pressed t o the limit of my abilities

5. Check one: -(1) A job where I am the

final authority on my work -(2) A job where there is nearly

always a person or a pro- cedure that will catch my mistakes

6 . Check one: -(1) A job where I could be

either highly successful or a complete failure

-(2) A job where I could never be too successful b u t neither could I be a com- plete failure

7. Check one: -(1) A job that is changing

-(2) A job tha t is constantly very little

changing 8. Check one:

-(1) An exciting job but one which might be done away with in a short time

-(2) A less exciting job but one which would undoubtedly exist in the Company for a long time

Page 4: Some Correlates Of Risk Taking

300 PERSONNEL PSYCHOLOGY

to a range from .83 to .90, the eight items have been utilized in all subsequent research efforts?

Scoring involves the assignment of a weight of 1 for each risk alternative endorsed and a 0 for non-risk alternatives en- dorsed, with scale scores from 0 to 8 possible. An analysis of large samples has indicated that these scores form an approxi- mately normal distribution. Analysis of nonresponses has shown that the failure to respond to any one item is usually associated with an item which is close to but beyond (an item which is usually endorsed only by higher risk takers) an in- dividual’s provisional scale score. As such, it is safe to score a nonresponse as a non-risk endorsement if at least half of the items are attempted.

Study 1 The original study involved approximately one thousand

workers and their supervisors in a midwestern utility company. JPI was embedded in an attitude and opinion questionnaire. The following are some of the findings from this study:3

Resistance to Change. As new job situations present many unknowns, it seems reasonable to expect the low risk takers to be much less prone than high risk takers to desire any sig- nificant modification in their jobs. All of the respondents, at the time of this study, had just experienced a period of extreme organizational and job change as a result of a large scale com- puter installation. They were asked their feelings about en- gaging in another organizational change if one should become necessary and, secondly, their feelings about job rotation, a procedure which had been used with considerable success be- fore the change. Among approximately 562 non-supervisory

* Final reproducibility coefficients have been determined by using the methods suggested by Goodenough (1944). Use of the Cornell technique, i.e., utilizing cutoff points, resulted in somewhat higher reproducibility coefficients. It should also be noted that nearly half of the errors are accounted for by respondents with a scale score of 1 or 2. These respondents tend to endorse somewhat extreme items, i.e., items usually endorsed only by quite high risk takers. This suggests that subjects do become aware of the fact that they are highly security conscious in responding to the questionnaire, and that, be- cause this has negative social desirability implications, they therefore endorse some item that has high social desirability.

The research cited in Study 1 is more fully reported in Williams (1960).

Page 5: Some Correlates Of Risk Taking

LAWRENCE K. WILLIAMS 301

employees, there was a clear trend for an increasingly larger proportion of individuals to indicate a desire for change with increases in JPI scores. A relationship between JPI score and desire to experience job rotation was found for 601 non-super- visory employees. Approximately, the greater the JPI score (the greater the propensity to take risk) the greater the desire to experience job rotation. Both studies revealed an almost continuous relationship between risk-taking propensity and favorable perceptions of change in their job activities. In the former study, chi-square was 32.9, P < .01; in the latter x2 = 21.9, P < .02. The rank correlations were .96 and 3 5 , respec- tively. Of course, the questions used to measure “resistance to change” and those used in the JPI itself were quite similar, which would account in part for the high relationship obtained.

Job Satisfaction. A measure of the amount of risk in the job was derived from an independent rating of these jobs? De- partment heads and supervisors in the company rated 50 jobs, and a risk score was determined for each job. The original hypothesis was that the greater the congruity between pro- pensity for risk and the amount of risk in the job, the greater would be the job satisfaction as measured by a five-point job satisfaction item. Analysis indicated that, in fact, high-risk jobs were more satisfying to both high risk takers (JPI of five or more) and to low risk takers, indicating that there are fac- tors in the high-risk jobs other than the risk itself that cause them to be satisfying. The hypothesis w&s confirmed only for high risk takers. The mean job satisfaction index for high risk takers on high-risk jobs was 3.89 (5.0 would be complete satis- faction) and 3.32 on low risk-taking jobs. A t test applied to the difference between means was significant beyond the .01 level. Low risk takers were also more satisfied on high risk tak- ing jobs than on low risk taking jobs (3.84 versus 3.55, respec- tively) although the difference in this case is not significant. These fhdings suggest that while high risk takers are not nec- essarily more satisfied than low risk takers on high risk de- mand jobs, they are definitely more dissatisfied than low risk takers on low demand jobs. Stated in another way, the amount

’ This is more fully reported in Mann and Williams (1962).

Page 6: Some Correlates Of Risk Taking

302 PERSONNEL PSYCHOLOGY

Mean rank assigned to importance

of promotions

4.00 3.36 3.20 3.08 2.89 2.83 2.89 2.28 2.24

-

of risk on the job is of importance and affects the satisfaction of high risk takers but is not an explanatory factor regarding job satisfaction for low risk takers.

Job Incentives. The JPI score was related to various job incentives for non-supervisory employees. Starting with the proposition that low risk takers would be more attracted to the security offered by a company rather than to the job itself, it was hypothesized that high risk takers would be more at- tracted to the intrinsic aspects of the job and to promotion or mobility opportunities than would low risk takers. The re- spondents were asked to rank order six incentives, and a mean rank for each incentive as a function of the JPI score was computed. The results are given in Table 1. A low number cor- responds to a high ranking.

The second column in Table 1 is concerned with the mean rank assigned to the importance of promotions. There is a sig- nificant trend in this column indicating that, the lower the risk taking, the less the value attached to promotions. A one way analysis of variance applied to these data yields an F of 2.55 which indicates a significant group effect (.05 level).

The third column of Table 1 is concerned with the mean rank assigned to steady employment. With the exception of the zero score, there is a reasonable progression of mean ranks

-

TABLE 1 Comparison of Risk-Taking Score and Attraction of Various Incentives

( N = 567)

JPI Score N

20 52 86

101 120 78 63 29 17

Mean rank assigned to

steady employment

3.09 2.47 2.73 2.86 3.10 2.81 2.95 4.28 4.06

Mean rank assigned to doing thing

a t which best

3.00 3.60 3.16 2.99 3.03 3.01 2.68 2.38 2.18

Page 7: Some Correlates Of Risk Taking

LAWRENCE K. WILLIAMS 303

such that, the higher the risk-taking index, the less the im- portance attached to steady employment. An analysis of vari- ance applied to these data yielded an F of 3.76, significant be- yond the .01 level.

The fourth column involves the mean ranks assigned to doing things a t which one is best. This item was seen as a measure of interest in the job itself (intrinsic job involve- ment). In general, it was found that increasing importance is attached to doing things best with successively higher levels of risk taking. Analysis of variance applied to these data resulted in an F of 2.25, significant beyond the .05 level.

The demand of low risk takers for security was also reflected in several items concerned with mobility and promotions. Of particular interest were the responses to the item which asked whether they preferred promotion on the basis of seniority or ability. The relationship between the JPI score and response to this item was significant at the .02 level (chi-square test). Similarly, for an item which asked if the company should take care of only those who do a good job, or whether the company had a responsibility for all of its employees regardless of the kind of work they did, the rank correlation between JPI score and attitude toward the company was .99 (significant a t the -01 level), with between 80 and 100 percent of the high risk takers feeling that the company was obligated to employees only if they did a good job.

Another analysis, made on 315 non-supervisory employees having three or more years of service, presented a comparison of those who thought it would be “very difficult” or “quite difficult’’ to find a job elsewhere, and those who thought it would be “moderately difficult” or “not difficult” at all to find a job elsewhere as a function of the JPI score. Among high risk takers, 70 to 80 percent indicated that it would be only “mod- erately difficult” or “not difficult” to find a job elsewhere. Greater difficulty was perceived with successively lower JPI scores, and less than 35 percent of the very low risk takers en- dorsed these responses. The rank correlation with the JPI score was .95. This relationship, which was significant a t the .01 level of confidence, yielded a chi-square of 26.5, P < .01.

Page 8: Some Correlates Of Risk Taking

304 PERSONNEL PSYCHOLOGY

Study 2 Study 2, which involved a replication of some of the items

included in Study 1, was particularly designed to look at be- havior as contrasted with attitudinal correlates of the JPI measure. This research was conducted in a group of power plants. Two of the plants had been established in recent years and in each case had been staffed by volunteers from the al- ready existing plants. Study l had established that high risk takers are more likely to desire job change than low risk takers. Study 2 provided an opportunity to examine actual behavior by examining the risk-taking scores of those who volunteered to move with those who stayed in the older plants. Although further shifting of personnel had taken place by the time this study was conducted, there remained a significant difference in the risk-taking score between those in the two new plants where employees were faced with quite undefined new jobs in contrast to the older plants. The mean JPI scores for non- supervisory workers in the new plants (N’s = 94 and 123) were 3.78 and 3.67, while the mean risk-taking score of those in the older plants ( N = 583) was 3.39. There is a significant difference between the mean score of the new plants (3.72) and the mean score of the old plant (3.39) at the .01 level. Similar observations were made for personnel who had some managerial responsibility in the plants. In this case, the mean JPI scores for managers in the new plants (N’s = 49 and 38) were 4.8 and 4.6, while the mean risk-taking score for such personnel in the old plants ( N = 152) was 4.0. Again, there is a significant difference between means for the two groups at the .05 level. One might assume that the difference might have been greater had all who had volunteered been accepted.

Study 3 This was a comparison of thirty individuals who had left

semi-skilled and unskilled work in an assembly operation to take work elsewhere in the community and a matched sample who remained on the job.6 Here, a behavioral correlate of risk-taking turnover was examined, the hypothesis being that those who take the initiative to find new work will be higher

This waa part of a larger research project performed by Torpie (1963).

Page 9: Some Correlates Of Risk Taking

LAWRENCE K. WILLIAMS 305

risk takers than those who choose to remain at a given job. A significant difference (.05 level) was found between those who had left voluntarily and those who remained on their jobs. The mean risk-taking score for those who had left the organization ( N = 29) was 3.44; for those who remained, the mean score was 2.43. The major difference between the groups was ac- counted for by the nine male respondents in the leaving group whose mean risk-taking score was 5.4 as against the compari- son group score of 3.0. For females the JPI scores were in the predicted direction, although the difference was not statisti- cally significant. It should be noted in this study that the men tended to give dissatisfaction with the job itself or intrinsic job problems as reasons for leaving while women more fre- quently gave reasons of an extrinsic nature. This result is con- sistent with the previous finding that reasonably high risk takers as compared to low risk takers are more concerned about the actual nature of their work than about the security of the environment.

Study 4 This study involved the examination of workers in a de-

pressed area who had lost their previous jobs as miners. A com- parison was made of those who had found subsequent em- ployment with individuals who remained unemployed! The 68 individuals who had located some form of employment had a mean risk-taking score of 3.26. This mean JPI score was sta- tistically different (.01 level) from that of 66 individuals who had not yet found a new job (mean = 2.16).

This finding would seem to confirm previous results, which indicated that the greater the degree of risk-taking propensity the more one is inclined to seek and accept jobs that are differ- ent from those that have been experienced before. In this par- ticular study the locating of a new job was undoubtedly af- fected by the individual's perception of the availability and appropriateness of alternative jobs. Those who found employ- ment, the higher risk takers, engaged in a greater number of search activities in finding employment, and looked at a

'A more detailed discussion of this research i reported in Williams, Folt- man, and Roaen (1963).

Page 10: Some Correlates Of Risk Taking

306 PERSONNEL PSYCHOLOGY

broader spectrum of jobs than did those who were lower risk takers.

Study 5 White-collar employees in a large light and power company

in Lima, Peru, were also administered the JPI index in con- junction with a general attitude questionnaire. A quasi scale (reproducibility = 32) was obtained. For these workers the in- dex seemed to have the same relationships and predictive value as it did in Study 1, which also involved white-collar utility workers. For the 180 respondents in this phase of the study, the previous findings that high risk-taking subjects are more likely to be interested in the intrinsic aspects of the job were again confirmed. For a high risk-taking population, the correlation between total job satisfaction and opportunities to do things a t which they were best was .95, while for low risk takers the correlation was .78. The difference between these two correla- tions was significant a t the .01 level.

Again, although exact comparative analyses were not pos- sible, it was found that high risk takers who were in jobs with a minimal amount of risk tended to be more dissatisfied than low risk takers. This was partially confirmed with regard to their feelings about promotion. When there was a very good chance of promotion, both high and low risk takers were highly satisfied. When there was a poor or no chance for promotion from low level jobs in the organization, high risk takers were significantly less satisfied than low risk takers.

Finally, mobility and moving in the organization were pre- sumably more important for high risk takers than low, with a significantly greater percentage ( P < .OOl) of the high risk takers indicating that they would like the job of supervisor.

The risk-taking items were administered also to blue-collar Peruvian workers, but it was impossible to obtain a reasonable scale from their responses.

Summary and Discussion This article has attempted to describe the utility of a risk-

taking index by showing its relationship to both attitudinal and behavioral variables. To this extent the report is in the

Page 11: Some Correlates Of Risk Taking

LAWRENCE K. WILLIAMS 307

nature of a report on congruent validity. The article also in- cludes a number of substantive findings, however, which should be evaluated in terms of “a theory of risk taking.” In this section, comments will be made first on the utility or nature of the scale and, secondly, on the findings and their im- plications for a risk-taking theory.

The Scale. In exploring the utility of the JPI, it seemed nec- essary to make the demand that predictions be made for all classes or intervals in the index and for all classes or intervals available in the dependent variable. Collapsing the sample or population into high and low risk takers only, and exploring the relationship with a dependent variable, would provide a support for the hypothesis but not necessarily support for the instrument involved. The data indicated under Study 1 in particular was analyzed with the demand for making predic- tions for individuals at each point on the index. Despite the fact that the inventory fails to reach the reproducibility de- mand of a Guttman scale, there is empirical proof that it per- mits adequate predictions for classes of individuals repre- sented by each scale score.

It should be noted, however, that one merit of the inventory, the fact that it is easily administered within a pencil-and-paper questionnaire to workers, and the fact that risk taking is meas- ured only in terms of desired characteristics of a job, limits its utility with other populations. This scale has been used with high school students in a number of studies, and, although in nearly every instance predictions with regard to risk taking have been in the correct direction, they have not reached ap- propriate levels of significance. It is rather simple to explain this in retrospect in that individuals who have never had any real experience with jobs cannot make very realistic appraisals of alternatives with regard to them.

Finally, in commenting on the utility of the scale, it should be noted that the failure to obtain a reasonably reproducible scale for blue-collar workers in Peru is appropriately a reflec- tion of the conceptual differences between the two cultures in the perception of risk taking. In this light it should be noted that, while a reasonably reproducible quasi scale was obtained for white-collar workers, many of the items were rearranged in

Page 12: Some Correlates Of Risk Taking

308 PERSONNEL PSYCHOLOGY

terms of the marginals: i.e., an item which was endorsed by 20 percent of the United States employees might be endorsed by 70 percent of the Peruvian workers, while the reverse might be true of another item. Thus, in terms of comparative research, very rigid requirements with regard to scaling fea- tures of the instrument become necessary in order to under- stand the conceptual differences in a dimension or universe.

Risk Taking. The derivation of hypotheses relating the risk- taking measure to various dependent variables in the world of work has been relatively simple. The hypotheses have not been derived from a general theory of risk taking but rather have stemmed from the consideration of the dependent or choice variable and the prediction made that those who are high risk takers would be more likely to endorse the more uncertain alternatives, while those with a low risk-taking propensity would choose alternatives which reflect certainty or security. It has been found that high risk takers as compared to low risk takers are more likely to look for and obtain job opportunities which they think will be satisfying. Perhaps of greatest im- portance have been those findings relating to the intrinsic and extrinsic aspects of the work. In each case it was found that low risk takers were more concerned with the extrinsic character- istics of the work, and those who are high risk takers were more concerned with the intrinsic characteristics which pre- sent a direct challenge to them.

It is interesting to observe that most of the extrinsic incen- tives offered in an organization have 100 percent certainty to them as long as the person remains in the organization. Re- tirement, health, and welfare benefits, as well as promotional policies based on seniority, usually have 100 percent certainty. The intrinsic incentives, such as jobs with responsibility, vari- ety, as well as pay and promotion systems based on merit as distinct from seniority, are all of a much lower order of cer- tainty.

It should also be noted that the research has not examined the perception of probabilities on the part of the respondents. There is still the possibility that high risk takers do not nec- essarily take more calculated risks than do low risk takers, but rather that high risk takers do not perceive the same amount

Page 13: Some Correlates Of Risk Taking

LAWRENCE K. WILLIAMS 309

of risk in an alternative as do low risk takers. High risk takers may assign a completely different set of probabilities to an ar- ray of alternatives than do low risk takers, and both of them may engage in an activity or be motivated towards an incen- tive if the subjective payoff is approximately fifty percent. In Study 1, high risk-taking respondents indicated that there were more alternative jobs in the community than did low risk tak- ers despite the fact that objectively their situation was the same in terms of training, background, and in the nature of the environment. Thus, high risk takers may have been as- signing a reasonably comfortable probability to finding an- other job whereas low risk takers were assigning extremely low odds to the possibility of finding another job. The implications of this in terms of changing an organization is the fact that we may not want to manipulate incentives but rather manipulate or change the perception of a possible payoff of incentives. Adding an additional bonus to accomplish a certain activity may run actually counter to the organization’s intentions if, in the process of increasing the size of the incentive, they lower the perceived odds of obtaining it on the part of low risk takers. In a similar fashion, increasing the amount of incentive of an extrinsic incentive, for example, when the probability is al- ready 100 percent, may do nothing to motivate additional be- havior on the part of high risk takers who are predisposed to entertain alternatives that have some degree of challenge in- herent in the process.

In general, this set of studies has suggested that risk taking as a measure of individual propensity, and as a way of de- scribing choices or possible behavior in the work setting, does allow for meaningful predictions and allows one to consider in- dividuals and organizational events in the same variable lan- guage. References ATKINSON, J. W. “Motivational Determinants of Risk-Taking Behavior.”

Psychological Review, LXIV (19571,359-372. ATKINSON, J. W. (Editor). Motives in Fantasy, Action and Society. Princeton :

D. Van Nostrand, 1958. ATKINSON, J. W., BASTIAN, J. R., EARL, R. W., AND LITWIN, G. H. “The

Achievement Motive, Goal-Setting, and Probability Preferences.” J O U ~ Q ~ of Abnormal and Social Psychology, LX (lW), 27-36.

Page 14: Some Correlates Of Risk Taking

310 PERSONNEL PSYCHOLOGY

ATKINSON, J. W. AND LITWIN, G. H. “Achievement Motive and Test Anxiety Conceived as Motive to Approach Success and Motive to Avoid Failure.” Journal of Abnormal and Soeial Psychology, LX (1960), 52-63.

EDWARDS, W. “Probability-Preferences Among Bets With DifTering Expected Values.” American Journal of Psychology, LXVII (1954), 56-67. (a)

EDWARDS, W. “The Reliability of Probability Preferences.” American Journal of Psychology, LXVII (1954),68-95. (b)

GOODENOUGH, W. H. “A Technique For Scale Analysis.” Educational and Psy- chological Measurement, IV (1944), 179-190.

GUTTMAN, L. “The Cornell Techniques For Scale and Intensity Analysis.” In C. Churchman, R. Ackoff, and M. Wax (Editors), Measurement of Con- sumer Interest. Philadelphia : University of Pennsylvania Prem, 1947. Reprinted in Educational and Psychological Measurement, VII (19471, 249-279. (a)

GUTTMAN, L. “Suggestions For Further Research in Scale and Intensity Analy- sis of Attitudes and Opinions.” International Journal of Opinion and Attitude Research, I (19471,30435. (b)

MANN, F. c. AND WILLIAMS, L. K. “Some Effects of the Changing Work En- vironment in the Office.” Journal of Social Issues, XVIII (1962),90-101.

SLOVIC, P. “Convergent Validation of Risk-Taking Measures.” Journal of Abnormal and Social Psychology, LXV (1962), 68-71.

STOUFFER, S. A., GUTTMAN, L., SUCHMAN, E. A., LAZARSFELD, P. F., STAR, SHIRELY A., AND CLAUSEN, J. A. “Studies in Social Psychology in World War 11.” In Social Science Research Council (Editors), Volume 4, Meas- urement and Prediction. Princeton, N. J. : Princeton University Press, 1950.

TORPIE, A. L. “Job Satisfaction and its Relationship to Turnover.” Unpub- lished Master’s thesis, Cornell Unversity, 1963.

WILLIAMS, L. K. “The Measurement of Risk-Taking Propensity in an Indus- trial Setting.” Unpublished Ph.D. thesis, University of Michigan, 1960.

WILLIAMS, L. K., FOLTMAN, F. F. AND ROSEN, N. A. Some Psychological Cor- relates of a Depressed Area. Reprint Series No. 139. Ithaca, N. y.: Cor- nell University, 1963.