Technical Affairs Section

Mike Aamodt, Associate Editor

The "Dear Mike" question and answer part of this column will reappear. We did receive a couple of questions following the last ACN but because of timing issues we were not able to get them answered properly for inclusion in this issue. To those who provided the questions, thank you and be assured that you will soon see them answered here. To the rest of you, in order to have answers we need to have questions – so help us out. Please tell us the technical questions you have in the area of assessment to which we might provide answers. Let us hear from you. Send your questions to Mike Aamodt: phone: (703) 831-5513, fax: (703) 831-6113, e-mail: maamodt@runet.edu.

For this issue we have an article by Mark Nagy which answers the question: what do you do when you are dissatisfied with job satisfaction scales?

What To Do When You Are Dissatisfied With Job Satisfaction Scales: A Better Way to Measure Job Satisfaction

by Mark S. Nagy, Ph.D., Radford University

If you are feeling a little dissatisfied with current job satisfaction scales, you are not alone. Despite over 6000 articles on job satisfaction, many researchers such as Frank Landy (1989) suggest that we still know very little about job satisfaction. For example, most "common" people I've talked to assume that such things as job performance, turnover, and absenteeism are strongly related to job satisfaction. Yet, current measures of job satisfaction have failed to yield these strong relationships. In fact, most current measures show weak, if any, relationships with job performance, turnover and absenteeism. This lack of evidence suggests one of three things: 1) that these relationships, although intuitive, simply do not exist, 2) that our conceptualization of job satisfaction is wrong, or 3) that we have failed to adequately measure job satisfaction. I believe the reason we know so little about job satisfaction is that our measurement of the construct has been terribly poor. In the following pages, I will review problems associated with past approaches to measuring job satisfaction, and will present an alternative method of measuring the construct that should help us better understand the mysterious nature of job satisfaction.

Common Conceptualizations of Job Satisfaction

Before a person can measure a construct like job satisfaction, one must be able to articulate exactly what one is trying to measure. One of the earliest conceptualizations of job satisfaction was proposed by Frederick Herzberg and his associates in 1959. Herzberg believed that job satisfaction consisted of two distinct dimensions. One of these dimensions, called "hygiene" factors, involved the environmental surroundings of a job and included such extrinsic aspects as supervision, salary, interpersonal relations, working conditions, and status. Herzberg called the second dimension of job satisfaction "motivator" factors, which were related to job tasks, job content, and the intrinsic aspects of a job, and included such aspects as recognition for achievement, work itself, responsibility, and growth. Herzberg reasoned that satisfying hygiene factors cannot lead to job satisfaction, but may result in an avoidance of job dissatisfaction. However, satisfying motivator needs can lead to job satisfaction, but the absence of such factors cannot lead to job dissatisfaction.

Unfortunately, empirical evidence for Herzberg's theory is scarce. In fact, most of the support for Herzberg's theory has been gathered using his own, unique methods consisting of critical incident interviews. According to Godfrey Gardner (1977), these interviews asked subjects to recall instances when they had felt "exceptionally good" and, during a second interview, instances when they had felt "exceptionally bad" about their jobs. As several researchers have pointed out, using such a method may result in a self-serving bias in which individuals are likely to associate positive outcomes to themselves (making an internal attribution) and blame negative outcomes to extenuating circumstances (making an external attribution; Grigaliunas & Wiener, 1974; Gardner, 1977; Farr, 1977). Thus, Herzberg's conceptualization of job satisfaction, while unique, has not received much support in the job satisfaction literature and is thus of little value. However, its continued use even today may help explain why we know so little about job satisfaction.

Another, more popular conceptualization of job satisfaction has only one dimension but consists of responses to these two questions: 1) how much do I currently have, and 2) how much do I want. Generally, this discrepancy between what one has compared to what one wants is believed to be an indication of job satisfaction. In fact, many researchers, including Lyman Porter (1961), Lloyd Lofquist and Rene Dawis (1969), J.L Holland (1973), and Edwin Locke (1976) have proposed this discrepancy as a conceptualization of job satisfaction. In addition, Robert Kopelman (1979) noted that researchers in other fields, such as counseling psychology, tend to describe satisfaction as the difference between what one has versus what one desires.

Finally, nearly all conceptualizations of job satisfaction involve some type of evaluation process. Edwin Locke (1976), for instance, defines job satisfaction as a "pleasurable or positive emotional state resulting from the appraisal of one's job or job experiences" (p. 1300); Arthur Brayfield and Harold Rothe (1951) suggested that job satisfaction is the "individual's attitude toward his work." (p. 307), while Pat Smith and her colleagues (1989) believed that job satisfaction was the extent to which "... the work environment (fulfills) the requirements of the individual" (p.21).

Consequently, it appears that job satisfaction can best be described as an evaluation process that considers what an individual has and what an individual wants. Armed with this solid conceptualization of job satisfaction, I will now evaluate how accurately popular methods of job satisfaction represent this conceptualization.

Measurements of Job Satisfaction

Global versus Facets Scales

There are two primary ways to measure job satisfaction. One method is to simply measure overall, or "global" job satisfaction. Many times, this overall measure is obtained by asking a single question, such as, "Overall, how satisfied are you with your job?". The other method of measuring job satisfaction involves "facet" scales that are used to measure such separate, or specific, areas of a job as satisfaction with supervision and satisfaction with pay.

While these differences may seem subtle, they may have important implications for predicting job satisfaction. For instance, facet scales may neglect some aspects of a job that are important to an employee. As a case in point, some facet satisfaction scales do not ask about working conditions, which may be important to employees. On the other hand, facet scales may also include job features that are not important to the employee, such as co-worker relations. Thus, facet scales may omit or add general areas that are or are not important to the employee.

On the other hand, asking global, overall evaluations of job satisfaction does allow employees to consider all relevant aspects of their jobs. For example, employees would consider working conditions if they were important to them while neglecting such unimportant aspects as co-worker relations. While this difference may seem subtle, it may help explain why global measures are better than facet measures at predicting organizational outcomes such as job performance and absenteeism (Iaffaldano & Muchinsky, 1985; Petty, McGee, & Cavender, 1984; Scott & Taylor, 1985).

Finally, given the fact that facet scales may be omitting relevant information to explaining job satisfaction while including other, irrelevant information, if it follows that combining or adding facet scales to get a global index of satisfaction is unwarranted. Yet, according to Vida Scarpello and John Campbell (1983), such combination commonly occurs in the measurement of job satisfaction. Consequently, using facet scales to predict organizational outcomes seems unwarranted.

Popular Job Satisfaction Scales

Job Descriptive Index. Without question, one of the most popular measurement devices of job satisfaction is the Job Descriptive Index (JDI). Developed in 1969 by Pat Smith and her colleagues, the JDI has been used in over 400 research publications (Smith et al., 1989). The JDI is used to determine satisfaction with five facets: work itself, coworkers, supervisor, pay, and promotion opportunities. The nature of the JDI stemmed from the belief that satisfaction is judged relative to an individual's perception of alternative jobs available to the person (Smith et al., 1989).

One methodological issue of concern is the use and allocation of points in the three point response used in the JDI scales. Instead of the customary 3, 2, 1 point distribution found on most three-point scales, the JDI allocates three points for a "yes" response, one point for a "?" response, and zero points for a "no" response. Apparently, the "?" response is more indicative of dissatisfaction than satisfaction (Cook, Hepworth, Wall, & Warr, 1981). However, in a 1982 study by Steven Johnson, Pat Smith, and Susan Tucker, the three point response format was compared to a five-point Likert scale using the same items from the JDI. Johnson's results indicated that the three-point scale was significantly negatively skewed in the supervision and co-worker subscales, and significantly positively skewed in the promotion subscale, whereas the five-point response had normal distributions for all five subscales. However, because there were no significant differences between the three-point and five-point scales, and because the three-item response format was easier to explain and easier for employees to use than the five-point scale, the authors lobbied for the continued use of the three-point scale. Yet, there was no mention of ceiling or floor effects existing in the three-point scales. It seems that in situations where there are no statistical differences between scales, biasing a distribution significantly in three of the five subscales would warrant the selection of the five-point scale over the three-point scale. Thus, it appears that the demand for efficiency is more of a concern to the authors of the JDI than the damaging potential of ceiling or floor effects on their obtained distributions.

Finally, the most important criticism of the JDI is that it does not follow its own conceptualization of job satisfaction. According to Pat Smith and her associates, the JDI was based on a definition that: "job satisfactions are feelings or affective responses to facets of the situation" (1989; p. 6). However, Smith et al. (1989) noted that asking employees for descriptions of their jobs revealed more frank and less defensive responses than asking employees regarding their feelings about their jobs. Consequently, they developed the JDI to emphasize job characteristics and not personal emotions about a person's job (Smith et al., 1989). Yet, they conceptualized job satisfaction as "feelings or affective responses". This contradiction certainly illustrates that the JDI is not a conceptually strong measure of job satisfaction. Consequently, this inadequate, albeit popular, measurement of the construct certainly helps explain why we know so little about the nature of job satisfaction.

Minnesota Satisfaction Questionnaire and the Index of Organizational Reactions. Two other less popular facet measures of job satisfaction are the Minnesota Satisfaction Questionnaire (MSQ), developed by Lloyd Lofquist and Rene Dawis in 1969, and the Index for Organizational Reactions (IOR), developed by Randall Dunham, Frank Smith and Richard Blackburn in 1977. Both of these scales have documented evidence of validity in the research literature. Unlike the JDI, both of these scales have a five-point response format, and both involve a person's feelings about a job and not their descriptions about a job. However, both of these scales are quite long (the complete MSQ has 100 items, the IOR has 42 items), and, more importantly, both are missing what Rice and his colleagues (1989) call an integral component of the measurement of satisfaction: the discrepancy component. That is, while both scales ask employees how they feel about various aspects of their jobs, they do not frame those questions in a "how much do you have/how much do you want" framework. Consequently, whereas these scales appear to be conceptually stronger than the JDI, they are lengthy and they do not address the level of discrepancy one has with a particular facet of job satisfaction.

Job In General (JIG) and Faces Scales. Two popular global measures of satisfaction are the Job in General (JIG) and Faces Scales. The JIG was developed by the same authors who developed the JDI, namely Pat Smith and her associates in 1989. Like the JDI, the JIG uses the y/n/? three-point response format, with three points for a "yes" response, one point for a "?" response, and zero points for a "no" response. And, like the JDI, the JIG provides phrases or adjectives that a person uses when rating his/her job. Unlike the JDI, however, the JIG is based on feelings about one's job and not on descriptions of one's job. Although the scale is fairly new, the JIG has received evidence of construct validity (Smith et al., 1989). However, in a review of overall measures of job satisfaction, Vida Scarpello and John Campbell (1983) concluded that the best global rating of job satisfaction is a one-item, five-point scale that simply asks, "Overall, how satisfied are you with your job". Unfortunately, the JIG, with its 18 questions and three-point response format, does not fit the description of Scarpello and Campbell's ideal global measure of job satisfaction. Finally, like the JDI, the JIG adjectives and phrases omit the discrepancy element so frequently recommended in the conceptualization of job satisfaction (e.g. Rice, McFarlin & Bennett, 1989).

A scale that does have only one question when measuring global satisfaction is the Faces Scale developed by Kunin in 1955. As the name implies, the Faces Scale uses six facial expressions as its response format. One of the benefits of using the faces scale is that it is supposed to help those who cannot read complete the scale. However, a person must be able to read the question in order to answer the scale. In addition, like the JIG, the Faces Scale lacks the discrepancy component when measuring job satisfaction.

Discrepancy Measures. The notion of a discrepancy indicating levels of job satisfaction (or dissatisfaction) has been suggested throughout the job satisfaction literature. In fact, many researchers such as William Ronan and Edmond Marks (1973), Kenneth Teas (1981), and Robert Rice and his colleagues (1989) have stated that it would be difficult to propose a theory of job satisfaction (and, ostensibly, a measure job satisfaction) which did not incorporate some elements of discrepancy theory. Indeed, measures by Lyman Porter (1961), Edwin Locke (1976), and R.L. Holland (1973) have all incorporated the discrepancy notion into their measures of job satisfaction. Most of these measures involved asking two distinct questions: 1) how do you rate your current level?, and 2) how much more should there be?

Although asking two questions and then taking the difference is entirely consistent with popular conceptualizations of job satisfaction, there are some major methodological problems involved in this process. Perhaps the biggest problem is a restricted range of likely responses. For example, using a 1 (low) to 5 (high) scale, let's say Shelly rates her current pay as a 1. Her responses to the second question can range from a 1 to a 5. However, let's say Fred rates his current pay as a 4. Now, he must answer how much more should his pay be. Because he's already answered a 4 to the first question, chances are his response to the second question will be restricted to either a 4 or a 5. In other words, Fred is unlikely to report that his pay should be lower than it already is. Thus, answering the first question of "how do you rate your current pay" with a response greater than 1 will likely restrict the range of available responses to the second question.

This restriction of range problem could result in spurious relationships or the attenuation of significant relationships (Johns, 1981; Gati, 1989). In short, using two questions to get a difference score produces empirical findings that are useless. Consequently, most discrepancy measures of job satisfaction, while conceptually strong, are methodologically weak.

To summarize, many current job satisfaction scales fall short in adequately measuring job satisfaction. Whereas the JDI, the MSQ and the IOR seem to be conceptually weak, the method conceptually strongest, the discrepancy measure, has suffered from methodological problems. While these shortcomings may help explain why we know so little about job satisfaction, their continued use offers little hope of ever improving our knowledge of job satisfaction. However, there is a new, improved way of measuring satisfaction that not only is conceptually strong, but also is methodologically sound.

A Better Way to Measure Job Satisfaction

A possible solution to the restriction of range problem in discrepancy scores was suggested by Toby Wall and Roy Payne in 1973. They suggested that the way to avoid a restricted range of responses is to combine the two questions into one question. For job satisfaction, this would entail combining two questions about a job facet into one question. For instance, instead of asking two separate questions, "how satisfied are you with your current pay" and "how much more should your pay be", we should simply ask one question, "how much more should your pay be than it is now?". Using this strategy would eliminate the restricted range of responses observed when asking two separate questions, but would still be consistent with discrepancy conceptualization of job satisfaction.

In addition, the best way to measure facet satisfaction may be to ask about an overall evaluation of a particular facet. All of the facet scales reviewed in this paper (the JDI, MSQ, and IOR) ask several different questions about a particular facet, such as pay.

Moreover, recall that facet scales may include facets that are unimportant to an employee while omitting facets that are important to an employee. In much the same way, asking specific questions in a facet scale may reflect questions that are not important to the employee while omitting other questions that are important to an employee. For instance, a question about pay satisfaction may ask if one's pay is high, but may not ask if one's pay is fair. If pay fairness is an important issue to the person completing the survey, he/she would not be able to express this on a pay facet scale. Consequently, asking an employee to rate his/her pay using an overall assessment such as, "how much more should your pay be than it is now" would allow the employee to consider all relevant aspects of his/her pay, be it fairness, amount, or some other aspect that may not have been addressed in a facet measure of job satisfaction.

A final component that should be included when measuring facet job satisfaction is the level of importance a person attaches to each facet. For instance, pay may not be important to Shelly, but it may be important to Fred. Thus, while both may indicate they are dissatisfied with their pay, the only person we should be concerned about is Fred, since pay is important to him. Although I have found in my own research (Nagy, 1995) that importance levels have little to do with facet satisfaction, importance levels appear to have a lot to do with overall satisfaction. In fact, I have found that multiplying importance scores and discrepancy scores for several facets is a very good indication of overall satisfaction (Nagy, 1995). Plus, obtaining information about importance levels for a number of facets may provide very helpful information to organizations. For example, realizing that pay is not an important facet to a majority of the workforce could alter reward strategies used by an organization.

There are several benefits of using this new way of measuring job satisfaction. First, using the new method is easy for employees to answer. Unlike other facet scales, which use adjectives or phrases that may be confusing (e.g., "makes me content" on the JDI), this new method is straightforward: "how much more should there be than there is now?". Second, this new way of measuring job satisfaction is efficient. At most (if one includes importance levels), there are only two questions per facet. As a comparison, the JDI uses as many as 18 adjectives or phrases for one facet, while the IOR uses as many as six questions for one facet. A third benefit of this new measurement method is that it is very flexible. Whereas other facet scales have predetermined facets, the new method can be modified to measure just about any facet. For instance, if job security is an area of concern, the new method can be used to create a question regarding job security (e.g., "how much more job security should you have than you have now"); current facet scales do not have this capability. Finally, asking about importance levels can be very informative, as they can be used to indicate overall job satisfaction as well as areas that are important (and not important) to employees.

In summary, this new method of measuring job satisfaction, which entails combining two discrepancy questions into one question and asks about importance levels of each facet, satisfies both of the criteria of measuring job satisfaction -- it is consistent with popular conceptualizations of job satisfaction and it is also methodologically sound. In addition, this new way of measuring job satisfaction is easy, efficient, flexible, and informative. Hopefully, use of this new measure of job satisfaction will allow us to better understand the true nature of job satisfaction.

Questions or comments about this article can be directed to Dr. Mark S. Nagy, Department of Psychology, Radford University, Radford, VA, 24142, (540) 831-5519, e-mail: mnagy@runet.edu

References

Brayfield, A.H. & Rothe, H.F. (1951). An index of job satisfaction. Journal of Applied Psychology, 35:307-311.

Cook, J.D., Hepworth, S.J., Wall, T.D., & Warr, P.B. (1981). The Experience of Work. London: Academic Press.

Dunham, R.B., Smith, F.J., & Blackburn, R.S. (1977). Validation of the Index of Organizational Reactions with the JDI, MSQ, and Faces scales. Academy of Management Journal, 20:420-432.

Farr, R. M. (1977). On the nature of attributional artifacts in qualitative research: Herzberg's two-factor theory of work motivation. Journal of Occupational Psychology, 50:3-14.

Gardner, G. (1977). Is there a valid test of Herzberg's two-factor theory? Journal of Occupational Psychology, 50:197-204.

Gati, I. (1989). Person-environment fit research: Problems and prospects. Journal of Vocational Behavior, 35:181-193.

Grigaliunas, B., & Wiener, Y. (1974). Has the research challenge to Motivation-Hygiene Theory been conclusive? An analysis of critical studies. Human Relations, 27:839-871.

Herzberg, F., Mausner, B, & Snyderman, B. (1959). The motivation to work. New York: John Wiley and Sons, Inc.

Holland, J. L. (1973). Making vocational choices: A theory of careers. Englewood Cliffs, NJ: Prentice-Hall.

Iaffaldano, M. T., & Muchinsky, P. M. (1985). Job satisfaction and job performance: A meta-analysis. Psychological Bulletin, 97:251-273.

Johns, G. (1981). Difference score measures of organizational behavior variables: A critique. Organizational Behavior and Human Performance, 27:443-463.

Johnson, S.M., Smith, P.C., & Tucker, S.M. (1982). Response format of the Job Descriptive Index: Assessment of reliability and validity by the multi-trait, multi-method matrix. Journal of Applied Psychology, 67:500-505.

Kopelman, R. E. (1979). Directionally different expectancy theory predictions of work motivation and job satisfaction. Motivation and Emotion, 3:299-317.

Kunin, T. (1955). The construction of a new type of attitude measure. Personnel Psychology, 8:65-67.

Landy, F. J. (1989). Psychology of work behavior. (Fourth edition). Pacific Grove, CA: Brooks/Cole.

Locke, E.A. (1976). The nature and causes of job satisfaction. In M.D. Dunnette (Ed.), Handbook of Industrial and Organizational Psychology. Chicago: Rand McNally.

Lofquist, L.H. & Dawis, R.V. (1969). Adjustment to Work: A Psychological View of Man's Problems in a Work-Oriented Society. Appleton-Century-Crofts, New York.

Petty, M. M., McGee, G. W., & Cavender, J. W. (1984). A meta-analysis of the relationship between individual job satisfaction and individual performance. Academy of Management Review, 9:712-721.

Porter, L. W. (1961). A study of perceived need satisfactions in bottom and middle management jobs. Journal of Applied Psychology, 45:1-10.

Rice, R. W., McFarlin, D. B., & Bennett, D. E. (1989). Standards of comparison and job satisfaction. Journal of Applied Psychology, 74:591-598.

Ronan, W. W., & Marks, E. (1973). Continuing problems in exploring the structure of job satisfaction. Journal of Vocational Behavior, 3:279289

Scarpello, V., & Campbell, J. P. (1983). Job satisfaction: Are all the parts there? Personnel Psychology, 36:577-600.

Scott, K. D., & Taylor, G. S. (1985). An examination of conflicting findings on the relationship between job satisfaction and absenteeism: A meta-analysis. Academy of Management Journal, 28:599-612.

Smith, P.C., Kendall, L.M., & Hulin, C.L. (1969). The Measurement of Satisfaction in Work and Retirement. Rand-McNally, Chicago.

Smith, P.C., Balzer, W., Josephson, H.I., Lovell, S.E., Paul, K.B., Reilly, B.A., Reilly, C.E., & Whalen, M.A. (1989). Users' Manual for the Job Descriptive Index (JDI) and the Job in General (JIG) Scales. Bowling Green State University: Bowling Green, Ohio.

Teas, R. K. (1981). A within-subject analysis of valence models of job preference and anticipated satisfaction. Journal of Occupational Psychology, 54:109-124.

Wall, T. D., & Payne, R. (1973). Are deficiency scores deficient? Journal of Applied Psychology, 58:322-326.