ORGANIZATIONAL BEHAVIORAND HUMAN PERFORMANCE31, 353--364 (1983)
The Role of Pay and Market Pay Variability in Job Application Decisions SARA L . RYNES
University of Minnesota
DONALD P.
SCHWAB AND HERBERT G . H E N E M A N
III
University of Wisconsin-Madison The influence of pay and three other attributes on job application decisions under two sets of market conditions were examined. Consistent with economic hypothesizing, results suggested that most respondents used noncompensatory strategies to evaluate pay in making their application decisions. Moreover, both model usage and the pay importance appeared to vary as a function of market pay variability. Results are discussed primarily in terms of likely limitations to the generalizability of findings obtained from studies of attribute importance.
Researchers have long been interested in how pay shapes individual behaviors and attitudes. One major line of inquiry in this regard has been on the attempt to determine how pay can be administered to influence such employee behaviors as absenteeism, turnover, and especially performance (e.g., Lawler, 1971; Locke, Feren, McCaleb, Shaw, & Denny, 1980). Continuing interest has also focused on the relative importance of pay vis-a-vis other job attributes such as the importance of work itself in influencing those behaviors (e.g., Herzberg, Mausner, Peterson, & Capwell, 1957; Lawler, 1971). In both instances, the research of industrial/ organizational psychologists has tended to focus on the role of pay in influencing people already inside organizations. The present study, in contrast, investigated the role of pay in individuals' job application decisions. Two objectives in particular motivated the research. The first involved an attempt to address the relative veracity of two competing models on how individuals process information about Sara L. Rynes is now at the New York State School of Industrial and Labor Relations, Cornell University. The material in this manuscript was prepared with support of the Graduate School, University of Wisconsin-Madison and from Grant 91-55-79-34 from the Employment and Training Administration, U.S. Department of Labor, under tile authority of Title II1, Part B, of the Comprehensive Employment and Training Act of 1973. Researchers undertaking such projects under government sponsorship are encouraged to express freely their professional judgment. Therefore, points of view or opinions stated in this document do not necessarily represent the official position or policy of the Department of Labor. 353 0030-5073/83 $3.00 Copyright© 1983by AcademicPress,Inc. All rightsof reproductionin any formreserved.
354
RYNES, SCHWAB, AND HENEMAN
multiple attributes in arriving at job choice decisions. Most previous studies have assumed, either explicitly or implicitly, that individuals evaluate job attributes in a compensatory fashion. In compensatory models, individuals are assumed to make trade-offs between various job attributes such that unfavorable levels of one attribute may be offset by more favorable levels of other characteristics. Previous research based on expectancy theory (Vroom, 1964), for example, has assumed a compensatory model with respect to job attributes, since choices are hypothesized to be determined by summations of Instrumentality × Valence estimates for all the relevant attributes associated with each job alternative. 1 Compensatory model usage is also hypothesized in classical economic theory (Rottenberg, 1956). Moreover, most decision sciences' research on job preferences has assumed compensatory relationships through the use of additive linear mathematical models to represent human combinatory processes (e.g., Zedeck, 1977; Krausz, 1978; Feldman & Arnold, 1978; Arnold & Feldman, 1981). Although it must be acknowledged that previous studies have generally achieved high predictability using additive linear models (see, e.g., Hoffman, 1960; Goldberg, 1968; Stahl & Harrell, 1981), there are nevertheless reasons for testing alternative models in the context of job search and choice. For example, Einhorn (1971) noted that decision makers often describe their decision strategies in ways that are inconsistent with an assumption of compensatory model usage. Second, although linear models have often shown superior predictability, there is nevertheless limited research showing significant nonadditive or nonlinear model usage by some individuals (e.g., Einhorn, 1971; Fischer, 1976; Arnold, 1981; Strand, Levine & Montgomery, 1981). Third, Hoffman (1960) has observed that greater predictability for a particular mathematical model does not ensure that that model accurately reflects the way decisions have actually been made. Perhaps the most important reason for examining noncompensatory models in the job choice context, however, is that they have been explicitly hypothesized and purportedly observed in the economics literature. Institutional labor economists, beginning with Reynolds (1951), concluded that unemployed j o b seekers rarely evaluate theattractiveness of alternative job offers in a compensatory fashion. Rather, he argued that job seekers establish minimally acceptable standards on one or two attributes (usually pay and/or type of work) 2 and then accept the first job to 1 According to the e x p e c t a n c y formulation, j o b s containing an attribute with either zero instrumentality or zero valence m a y still be c h o s e n if the s u m o f ( I × V) values for all o t h e r attributes is higher t h a n the analogous s u m s associated with other available alternatives. 2 In the case of p a y the minimal a m o u n t of c o m p e n s a t i o n required to accept a job offer h a s since been called the reservation wage.
PAY AND JOB APPLICATION
355
meet or exceed those constraints. Following Reynolds, others (e.g., Sheppard & Belitsky, 1966; Sobel & Folk, 1965) have similarly concluded that unemployed job seekers have some reservation wage standard below which other attributes, however attractive, cannot induce a job acceptance. The existence of reservation wage strategies in choosing jobs has also been hypothesized in formal mathematical models developed by theoretical economists. Following Stigler's (1962) seminal work on the costs of market information, various optimal-stopping models have been developed that hypothesize the setting of reservation wages to maximize discounted lifetime earnings. (For a review of these models and their assumptions, see Lippman & McCall, 1976.) Although the origins (and most of the assumptions) of formal job search theory are quite distinct from those associated with the earlier work by institutional economists, both posit reservation wages to be the standard against which job offers are either accepted or rejected. Hence, the present research was designed to detect the possible use of noncompensatory, reservation wage strategies in decisions concerning whether or not to apply for job vacancies. The second major research question involved the potential importance of labor market parameters in job search and choice decisions. Specifically, economists such as Reynolds (1951) and Rottenberg (1956) have speculated that the importance of pay (or any other attribute) in job choice varies directly with the range of, or variance in, potential pay rates confronting the individual. Thus, they would hypothesize that in choice situations involving multiple job alternatives, the greater the range in pay offers, the greater the hypothesized effect of pay on the final decision. This hypothesis is potentially significant since, if true, it suggests that generalizations from any investigation involving the relative importance of job attributes are likely to be problematic in at least two ways. First, since markets for different occupations sometimes vary considerably in their characteristics, importance hierarchies derived from any particular occupational sample may generalize only to other job seekers facing similar market parameters. Thus, as an example, it may be erroneous to make inferences about the importance of pay to elementary schoolteachers (where pay varies little across alternatives) on the basis of information obtained from individuals in sales (where pay variability is much greater). What is probably less obvious, however, is that if the range of market parameters partially determines attribute importance, then any given individual's preference ordering may fluctuate in accordance with changes in characteristics of the market alternatives he or she generates. This possibility suggests that previous researchers reporting individual differences in pay importance (see Lawler, 1971, for a review) may have assumed that observed differences were due to stable individual charac-
356
RYNES, SCHWAB, AND HENEMAN
teristics when instead they may have resulted from differences in the kinds of markets confronted. For example, the widely reported finding that women have lower preferences for pay than do men (e.g., Jurgensen, 1978) may be attributable, at least in part, to the fact that women typically confront less variable pay distributions. Therefore, in an attempt to circumvent the potential confound of individual and market characteristics, the present study examined whether the job application strategies of a sample of occupationally homogeneous individuals changed when pay variability was experimentally increased (or decreased) between two time periods. The notion that attribute importance may be a function of market parameters suggests that it might also be interesting to investigate whether model usage might change in accordance with market variability. Thus, a tentative hypothesis was offered that noncompensatory job application strategies would occur more frequently under conditions of high (rather than low) pay variability (i.e., where the market offered both very highand low-paying alternatives). METHODOLOGY
Sample Subjects were 10 (5 female, 5 male) undergraduate students at a public midwestern university. They were recruited by mailing a request for participation to 6 males and 6 females randomly chosen from the population (N = 28) of seniors with a joint major in marketing and management. The request offered $35 for participation in the project and indicated that the respondent should apply only if s/he intended to seek full-time employment upon graduation. The first 5 individuals of each sex who applied were included in the sample. Subjects with a joint major in marketing and management were selected for study because they confront a wide range of starting salaries, 3 an important characteristic given the study's objective to investigate the impact of market wage variability. Procedure. Whereas most previous researchers have used self-report attribute ratings or rankings to study the role of pay in job choice, the present study employed a policy-capturing approach to examine questions of both pay importance and model usage. In policy capturing, hierarchies of attribute importance are derived by obtaining subjects' holistic evaluations of multiattribute alternatives, where levels of the attributes comprising those descriptions have been experimentally varied. The relative importance of the attributes is then Statistically inferred from Statistics obtained from the school placement office showed that salary offers to graduates with this major ranged from approximately $8500 to $18500 the previous year.
PAY A N D JOB A P P L I C A T I O N
357
the proportions of variance explained by each attribute in the subjects' overall evaluations. Although the relative superiority of either rating/ranking or policycapturing procedures has not been definitely established, a decision was made to use policy capturing on the basis of such hypothesized advantages as the possibility of weakened social desirability influences on responses (Arnold & Feldman, 1981) and the lower degree of self-insight required to obtain accurate results (Slovic & Lichtenstein, 1971). Moreover, although most previous policy capturing research on job choice has studied attribute importance (e.g., Zedeck, 1977; Feldman & Arnold, 1978), the method can easily be adapted to investigate questions of compensatory versus noncompensatory decision strategies. Additionally, the greater similarity of multiattribute evaluative tasks to real-world decision problems would appear to make policy capturing an attractive methodology for studying job choices. In the present study, multiattribute job descriptions were constructed after conducting initial interviews with several management and marketing professors and a number of students with the joint major. (None of the latter was included in the final sample.) These interviews revealed that starting salary, geographical location, promotional opportunities, and type of work were p r o b a b l y the most salient attributes to management-marketing majors in making job choices. Levels of these attributes were defined to create two sets of hypothetical job descriptions. In one, five salary levels were defined ($8800, $11,200, $13,600, $16,000, $18,400) to represent the widest pay distribution realistically available to the subjects. These were combined with three location levels (Houston, Minneapolis, San Francisco), two levels of work types (sales or management trainee), and two levels of average time to first promotion (1 or 3 years). A second set of descriptions with a narrower range of pay levels ($11,200, $12,400, $13,600, $14,800, $16,000), but identical levels of the other attributes was also created. Attribute levels in each set were combined to produce a 5 x 3 x 2 x 2 factorial design composed of 60 unique descriptions. These descriptions were replicated once to generate 120 descriptions in each pay distribution. Descriptions were administered to subjects, by set, in two sessions separated by at least 1 week. Half of the subjects evaluated descriptions containing the wide pay distribution first; half assessed the narrow distribution first? At the beginning of each session subjects were given an opportunity to examine the attributes and their levels and to evaluate five practice descriptions selected from the set. 4 A between-subjects test found no significant (p < .05) main effect for order or for order × attribute interactions.
358
RYNES, SCHWAB, AND HENEMAN
The practice descriptions and those used in the actual study called on subjects to respond " y e s " or " n o " to the question, "Would you attempt to obtain an interview for this job?" Although prior policy-capturing investigations have typically used some variant of job attractiveness measured on a 5- or 7-point scale, interview acceptances and rejections were examined here because perceptions of scaled attractiveness may not correspond with actual choices (see Einhorn, Kleinmuntz, & Kleinmuntz, 1979). Since our goal was to better understand the choices that people make about jobs, the dependent variable was felt to be a more direct reflection of the construct of interest. Interview (rather than job) acceptance was used as the stimulus since the latter decision would not generally be made on the basis of job descriptions only. Subjects were instructed to work through the 120 descriptions in each session in the order presented. Descriptions were randomly ordered, subject to the constraint that the levels of the first two attributes of adjacent descriptions not be identical. This ordering, used for all subjects and for both pay distributions, resulted in replicates never appearing closer than six descriptions from their original description. To encourage independent assessments of each alternative, subjects were not permitted to refer back to evaluations made on previous descriptions. Each of the two sessions took between 1 - 1 ½ hr to complete.
Analyses Within-subjects analyses were performed, with analysis of variance (ANOVA) used as the principal technique to study subjects' evaluations of the job descriptions. 5 Since the attribute levels were combined to form an orthogonal design, eta squares (,/2) were then computed to estimate the relative importance of the four attributes, and their interactions, in the job interview decisions. The impact of the pay distribution on pay importance was investigated by comparing the size of pay eta squares in the wide and narrow conditions, by subject. Evidence of noncompensatory choice behavior was studied as follows. Because the five pay levels were crossed with all other attribute levels, there were 24 (12 original and 12 replicate) jobs at each pay level in both salary range sets. Subjects' interview acceptance rates were determined at each of the five pay levels. Rejection of all 24 interviews at certain pay levels, regardless of the levels of the other attributes, was taken as evidence of the use of noncompensatory reservation wage Due to the d i c h o t o m o u s nature of the d e p e n d e n t variable nonhomogeneity of variance, all analyses were also performed using based on maximum likelihood estimation procedures. Because none affected in any substantial way, present findings are reported in the A N O V A terminology.
and the r e s u l t a n t logistic regressions of the results was more conventional
PAY A N D JOB A P P L I C A T I O N
359
strategies in making decisions. In addition, interview rejection rates were compared across pay conditions to determine whether the use of noncompensatory decision strategies appeared to increase as the market wage distribution was expanded.
RESULTS An initial analysis was performed to determine the reliability of subjects' interview decisions. Responses to the original descriptions were correlated with those of the replicate descriptions, by subject, within each pay distribution. On average, original and duplicate decisions correlated .90 in the wide distribution and .75 in the narrow one. Analysis of variance results tended to support the hypothesis that greater pay variability across job alternatives increases the importance of pay as estimated through policy capturing. Specifically, the mean eta square for salary in the wide distribution was .65 (range: .43-.82) as compared with .40 (range: .21- .56) in the narrow one. Moreover, in only one instance (Subject 5) did the variance explained by salary in the narrow distribution exceed that in the wide one. An across-subjects sign test indicated this directional pattern to be statistically significant (p < .01). One possible explanation for these results concerns differences in the obtained reliabilities across conditions. Because subjects' responses were generally more consistent in the wide than the narrow salary range, one might argue that the latter eta squares were correspondingly attenuated. To examine this possibility, eta squares in the narrow condition were treated as r2's and "corrected" by estimating their value ff the dependent variable had been measured with the reliability obtained in the wide salary range condition (NunnaUy, 1978). Results of this analysis showed that although the difference between eta square pairs was generally reduced, none changed direction. Although the preceding analysis suggests that policy-capturing estimates of relative importance fluctuate with differences in the ranges of attribute manipulations, it leaves unanswered the question of whether differences in attribute variability somehow influence individuals' evaluations of particular attribute l e v e l s . 6 Thus, two additional analyses were performed to address this question. First, two ANOVAs (one wide range and one narrow) were again performed for each subject, but this time using only the 72 descriptions in each condition that had pay levels common to both treatments ($11,200, $13,600, $16,000). Resultant eta squares for the two conditions were then compared to see whether the proportions of variance explained by pay were different across wide and narrow distributions when pay levels were 6 We thank an a n o n y m o u s reviewer for raising this question.
360
RYNES, SCHWAB, AND HENEMAN
held constant. A tentative hypothesis was made that the proportions of variance explained by the $11,200"$16,000 pay levels would be lower in the wide range condition, since those levels represented only a part of the wide pay range distribution but contained all of the pay variability in the restricted condition. For the second analysis, each subject's evaluation of jobs paying $11,200, $13,600, and $16,000 were combined across treatments and analyzed in a single analysis of variance, with experimental condition (wide vs narrow pay range) treated as another (orthogonal) factor in the design. If people evaluate identical pay levels differently across different market conditions, this should be reflected by the presence of significant pay × pay distribution interactions in the combined analysis. Comparisons of eta squares across pay treatments showed that, on average, the $11,200-$16,000 pay range accounted for 53% of explained variance in the narrow range and only 42% in the wide one (see Table 1). Moreover, more variance was explained in the narrow than in the wide range for 8 of the i0 subjects. Thus, it appears that the influence of any given range of attribute values on overall policy-capturing evaluations decreases when that range represents only a part, rather than all, of the available distribution. Results from the combined ANOVAs suggested a similar tendency for subjects to make stronger differentiations between $11,200 and $16,000 in the narrow, than in the wide, condition. Pay x pay distribution interactions were statistically significant (p < .05) for four subjects. Examination of the cell means revealed that, in general, more interviews were rejected at $11,200 in the narrow condition (where that was the lowest possible salary) than in the wide condition (where some alternatives offered only $8,800). Similarly, more $16,000 interviews were accepted in the narrow than in the wide range (where one-fifth of the jobs promised $18,400). These trends are summarized in Table 1, which shows average percentages of interview acceptances for all subjects at each common salary level.
TABLE1 AVERAGE PERCENTAGES OFINTERVIEW ACCEPTANCESACROSS PAY RANGE TREATMENTS
$11,200
$13,600
$16,000
Average proportion of variance explained
20% 15%
57% 58%
80% 84%
.42 .53
Salary Level Range Wide Narrow
361
PAY AND JOB APPLICATION
Turning to the second major focus of the research, it is suggested in Table 2 that a majority of subjects were using noncompensatory decision strategies in at least one of the experimenal conditions. As can be seen, seven subjects in the wide pay distribution and three in the narrow one rejected all interviews at certain pay levels across all combinations of nonpecuniary attributes. The larger number of consistent rejections at certain salary levels in the wide range supports the hypothesis that not only attribute importance, but also model usage, is likely to be influenced by the degree of variability in market parameters.
DISCUSSION For those with a substantive interest in the issue of how individuals decide whether or not to pursue a job offer, two principal findings emerge from the present research. First, job application decisions appear to be TABLE 2 INTERVIEW ACCEPTANCERATES BY SALARY LEVEL Pay distribution Wide (Narrow)
Salary level $8,800 ($11,200)
$11,200 ($12,400)
$13,600 ($13,600)
$16,000 ($14,600)
$18,400 ($16,000)
0.0 (4.2) 0.0 (0.0) 8.3 (62.5) 0.0 (4.2) 4.2 (20.8) 0.0 (0.0) 4.2 (16.7) 0.0 (4.2) 0.0 (37.5) 0,0 (0.0)
0.0 (4.2) 0.0 (0.0) 83.3 (70.8) 12.5 (29.2) 41.7 (62.5) 0.0 (0.0) 33.3 (37.3) 0.0 (20.8) 4.2 (70.8) 25.0 (20.8)
12.5 (13.0) 0.0 (20.4) 100.0 (95.8) 58.3 (54.2) 75.0 (100.0) 62.5 (16.7) 75.0 (54.2) 66.7 (79.2) 20.8 (91.7) 95.8 (79.2)
45.8 (29.2) 20.8 (0.0) 100.0 ( 100.0) 95.8 (54.2) 91.7 ( 100.0) 66.7 (58.3) 91.7 (79.2) 100.0 (79.2) 83.3 (95.8) 100.0 (95.8)
79.2 (54.2) 100.0 (50.7) 100.0 ( 100.0) 100.0 (91.7) 100.0 ( 100.0) 66.7 (58.3) 100.0 (95.8) 100.0 ( 100.0) 95.8 (100.0) 95.8 ( 100.0)
Subject One Two Three Four Five Six Seven Eight Nine Ten
Note. Figures without parentheses indicate interview acceptance rates in the wide pay distribution; those in parentheses represent the narrow pay distribution. All figures are shown in percentages.
362
RYNES, SCHWAB, AND HENEMAN
influenced in a number of ways by parameters of the market distribution confronted by the job seeker. Second, decisions whether or not to apply for job interviews often appear to be made on the basis of noncompensatory decision rules. Each of these findings, in turn, suggests a number of methodological cautions with respect to future job choice research. The contention of institutional labor economists (e.g., Reynolds, 1951; Rottenberg, 1956) that the role of any given attribute in job choice is a function of its variability was supported in at least two ways. First, eta squares from the policy-capturing analyses confirmed the a priori hypothesis that the proportion of job application variance explained by pay is a direct function of pay variability. Second, subsequent analyses revealed that job seekers evaluate particular levels of pay somewhat differently, depending upon the magnitude of other pay levels available for comparison. In terms of future research, these results suggest the need to choose attribute levels that accurately reflect subjects' appropriate market parameters or there will be little chance of obtaining externally valid results. Although some previous researchers appear to have been cognizant of the importance of market realism in policy-capturing research (e.g., Zedeck, 1977), others have employed rather superficial definitions of attribute levels (e.g., "presence or absence" of each attribute; Feldman & Arnold, 1978). Furthermore, the demonstrated impact of market parameters on obtained estimates of attribute importance suggests that future researchers attempt to avoid confounding differences in the markets confronted by individuals with variations in personal characteristics (e.g., race or sex) in their explanations of observed differences in preference hierarchies, r With respect to the issue of compensatory versus noncompensatory model usage, the finding that a majority of subjects appeared to evaluate employment opportunities in relation to a reservation wage standard implies that preferences for nonpecuniary attributes may be irrelevant to individuals' job choice decisions until minimal pay constraints have been met or exceeded, s It should be noted, however, that whether or not reservation wage-setting strategies are actually observed in a given job choice situation may depend on a number of factors. For example, in the present research the extent of observed reservation wage behavior varied as a function of the market pay distribution, with minimal pay constraints being more in evidence in the wider pay condition 7 As noted earlier, evidence that women report pay to be of less importance than men in job choice decisions is common. However, in the present study a between-subjects analysis revealed that the average pay r/2 was identical for males and females in the wide range and (insignificantly) higher for females in the restricted range. 8 Shapira (1981) recently observed similar refusals to make certain attribute trade-offs among a sample of currently employed individuals.
PAY AND JOB APPLICATION
363
than in the narrow one. Furthermore, although the issue was not directly tested in the present research, one might also hypothesize that had nonpecuniary attributes been defined at more favorable (and unfavorable) levels, reservation wage behaviors would have been observed less frequently, if at all. At the very least, however, present results suggest that future policy-capturing researchers be cognizant of the fact that assumptions of linear compensatory model usage may be inappropriate in studies of job interview and subsequent choice decisions. Given the potential implications of reservation wage-setting behaviors for job choice outcomes, additional research into the nature and extent of noncompensatory decision strategies would appear warranted. Previous theorizing about decision making in general, and job choice in particular, has suggested a number of additional variables that may be related to noncompensatory model usage. Economic job search theory, for example, suggests that reservation wage levels are a function of the costs of job search, as well as the mean and variance of the relevant pay distribution (Lippman & McCall, 1976). Other previously suggested hypotheses are that noncompensatory models are more likely to be used under conditions of high uncertainty (e.g., Reynolds, 1951; March & Simon, 1958) or high task complexity (Shapira, 1981; Simon, 1979), or where market constraints necessitate the use of sequential (as opposed to simultaneous) choice strategies (e.g., Reynolds, 1951; Einhorn, Kleinmuntz, & Kleinmuntz, 1979). In combination, these hypotheses suggest that differences in market characteristics associated with various occupations may be an important influence on individuals' job search and choice strategies. In summary, present results suggest that both the importance of pay and the nature of its role in decision strategies are likely to vary in accordance with the market pay distribution. Moreover, previous job choice theorizing and research suggest that attribute importance and the extent of noncompensatory model usage are likely to vary as a function of job seeker and nonpecuniary market characteristics as well. Thus, results from the present study underscore the need for caution in making generalizations about the role of pay (and, by implication, other attributes) in decisions involving employment. REFERENCES Arnold, H. J. A test of the validity of the multiplicative hypothesis of expectancy-valence theories of motivation. Academy of Management Journal, 1981, 24, 128-141. Arnold, H . J . , & Feldman, D.C. Social desirability response bias in self-report choice situations. Academy of Management Journal, 1981, 24, 249-259. Einhorn, H. J. Use of nonlinear, noncompensatory models as a function of task and amount of information. Organizational Behavior and Human Performance, 1971, 6, 1-27. Einhorn, H. J., Kleinmuntz, D. N., & Kleinmuntz, B. Linear regression and process-tracing models of judgment. Psychological Review, 1979, 86, 465-485.
364
RYNES, SCHWAB, AND HENEMAN
Feldman, D. C., & Arnold, H. J. Position choice: Comparing the importance of organizational and job factors. Journal of Applied Psychology, 1978, 63, 706-710. Fischer, G. W. Multidimensional utility models for risky and riskless choice. Organizational Behavior and Human Performance, 1976, 17, 127-146. Goldberg, L. R. Simple models or simple processes? Some research on clinical judgment. American Psychologist, 1968, 23, 483-496. Herzberg, F., Mausner, B., Peterson, R. O., & Capwell, D. F. Job attitudes: Review of research and opinion. Pittsburgh, Penn.: Psychological Service of Pittsburgh, 1957. Hoffman, P. J. The paramorphic representation of clinical judgment. Psychological Bulletin, 1960, 57, 116-131. Jurgensen, C. E. Job preferences (What makes a job good or bad?) Journal of Applied Psychology, 1978, 63, 267-276. Krausz, M. A new approach to studying worker job preferences. Industrial Relations, 1978, 17, 91-95. Lawler, E. E. III. Pay and organizational effectiveness. New York: McGraw-Hill, 1971. Lippman, S., & McCall, J. The economics of job search: A survey. Part I. Economic Inquiry, 1976, 14, 155-190. Locke, E. A., Feren, D. B., McCaleb, V. M., Shaw, K. N., & Denny, A. T. The relative effectiveness of four methods of motivating employee performance. In K. D. Duncan, M. M. Gruneberg, & D. Wallis, (Eds.), Changes in working life. New York: wiley, 1980. Pp. 363-388. March, J., & Simon, H. Organizations. New York: Wiley, 1958. Nunnally, J. C. Psychometric theory (Rev. ed.), New York: McGraw-Hill, 1978. Reynolds, L. G. The structure of labor markets. New York: Harper and Brothers, 1951. Rottenberg, S. On choice in labor markets. Industrial and Labor Relations Review, 1956, 9, 183-199. Shapira, Z. Making trade-offs between job attributes. Organizational Behavior and Human Performance, 1981, 28, 331-355. Sheppard, H. L., & Belitsky, A. H. The job hunt. Baltimore: Johns Hopkins Press, 1966. Simon, H. A. Rational decision making in business organizations. American Economic Review, 1979, 69, 493-513. Slovic, P. & Lichtenstein, S. Comparison of Bayesian and regression approaches to the study of information processing in judgment. Organizational Behavior and Human Performance, 1971, 6, 649-744. Sobel, I., & Folk, H. Labor market adjustments by unemployed older workers. In A. M. Ross (Ed.), Employment policy and the labor market. Los Angeles: Univ. of California Press, 1965. Stahl, M. J., & HarreU, A. M. Modeling effort decisions with behavioral decision theory: Toward an individual differences model of expectancy theory. Organizational Behavior and Human Performance, 1981, 27, 303-325. Stigler, G. Information in the labor market. Journal of Political Economy, Supplement, 1962, 70, 94-105. Strand, R., Levine, R., & Montgomery, D. Organizational entry preferences based upon social and personnel policies: An information integration perspective. Organizational Behavior and Human Performance, 1981, 27, 50-68. Vroom, V. H. Work and motivation. New York: Wiley, 1964. Zedeck, S. An information processing model and approach to the study of motivation. Organizational Behavior and Human Performance, 1977, 18, 47-77. RECEIVED; March 11, 1982.