Which is a Better Predictor of Job Performance: Job Satisfaction or Life Satisfaction?

Will people who are happy with their jobs exhibit superior job performance? Researchers have examined the “happy worker is a productive worker” postulation for decades and concluded the relationship between job satisfaction and job performance is not as strong as one would expect. The current research investigates possible explanations, including operationalizations of the satisfaction and performance variables. Using information from 87 respondents and their supervisors, the addition of life satisfaction and organizational citizenship behavior to the model was tested. Results moderately supported the addition of life satisfaction to increase our ability to predict inand extra-role job performance. Review of Relevant Literature The speculation that job satisfaction is related to performance dates back to the early days of the field of Industrial/Organizational Psychology (e.g., Kornhauser & Sharp, 1932). Throughout the years, organizational theorists have advanced several viewpoints regarding the nature of the satisfaction-performance relationship: (1) satisfaction causes performance; (2) performance causes satisfaction; (3) the satisfaction-performance relationship is moderated by a number of other variables; or (4) both are caused by an exogenous variable. Some moderators examined in past research include the contingency of rewards (e.g., Jacobs & Solomon, 1977; Lawler, 1973), situational constraints (e.g., Bhagat, 1982; Herman, 1973), self-esteem (e.g., Jacobs & Solomon, 1977; Lopez, 1982), pressures for production (Triandis, 1959), and reciprocity norms (Organ, 1977). Each of these proposed relationships has been investigated by numerous researchers. Yet there remains no clear consensus on which explanation is most valid across all settings or the circumstances in which each explanation is most valid. Furthermore, the majority of empirical findings published on this topic point to the apparent conclusion that the relationship between satisfaction and performance is weak at best. Nonetheless, many researchers continue to pursue the investigation of this relationship largely because of the apparently commonsensical link between the two variables. While most people associated with the human relations movement focused on improving performance through satisfying employees’ needs, research results contradicted this underlying assumption (e.g., Katz, Maccoby, Gurin, & Floor, 1951; Katz, Maccoby, & Morse, 1950). These researchers found no differences in satisfaction between high and low performance groups, and, in some cases, concluded that those with lower performance levels reported higher satisfaction. Brayfield and Crocket (1955) reviewed existing literature and profoundly changed the nature of investigations regarding this relationship when they concluded that there was little evidence of a © 2006 Institute of Behavioral and Applied Management. All rights reserved. 20 connection between employee attitudes and their performance. Many researchers now consider the relationship between job satisfaction and performance among employees to be what Chapman and Chapman (1969) called an “illusory correlation” – a perceived relationship between two variables that we logically or intuitively think should interrelate, but, in fact, do not. Previous Research Findings Research examining the relationship between satisfaction and performance has been conducted since at least as early as 1945 (e.g., Brody, 1945), and the methodology utilized has varied greatly. Some researchers used established scales to measure job satisfaction, while some developed their own. Some used self-report ratings to assess performance, while others used peer or supervisor ratings. Populations studied include diverse groups, such as college students, homemakers, public employees, bus drivers, engineers, salespeople, military personnel, physicians, and business school faculty members (cf. Iaffaldano & Muchinsky, 1985). Estimated effect sizes range from r = -.39 (e.g., Podsakoff, Todor, & Skov, 1982) to r = .86 (Lawshe & Nagle, 1953). In a landmark book that is still widely cited today, Vroom (1964) analyzed 20 studies conducted between 1949 and 1963 that examined the relationship between satisfaction and performance and reported a median correlation of r = .14. This weak positive average correlation was one of the first substantial indications that the relationship between these two variables was not as strong as was previously believed. Since Vroom’s results were published, others have attempted to replicate his findings. Partly in response to criticisms of Vroom’s (1964) research on this topic, Petty and his colleagues (1984) used more sophisticated meta-analysis techniques to replicate Vroom’s finding. Their meta-analysis contained 15 studies used by Vroom (1964) and an additional 20 studies published between 1967 and 1982 that utilize individual-level measures of performance and satisfaction. When the researchers excluded the studies included by Vroom (1964), they calculated an estimated effect size of .23 based on the remaining 20 studies with a combined sample size of 3,140 respondents. A more recent meta-analysis investigating the relationship between job satisfaction and performance was conducted by Iaffaldano and Muchinsky (1985). This research investigated the strength of the relationship and attempted to draw a robust conclusion other researchers would not dispute by subjecting findings regarding the relationship between performance and satisfaction to more thorough meta-analytic procedures. Their research included 74 empirical studies published in 70 articles with a total sample size of 12,192 subjects and 217 effect sizes. In a finding surprisingly similar to Vroom’s (1964), they concluded that the estimated true correlation between performance and all types of satisfaction measures after correcting for the effects of sampling error and attenuation due to unreliable measurement of both satisfaction and performance is r = .17. As can be seen in Table 1, which summarizes the findings from these three metaanalyses, all three studies came to a very similar conclusion regarding the estimated strength of the relationship between job satisfaction and job performance. In aggregate, © 2006 Institute of Behavioral and Applied Management. All rights reserved. 21 the conclusions drawn by these researchers, and many others, substantiate the presence of a positive, but very weak, relationship between job satisfaction and job performance. Table 1: Estimated Effect Sizes for the Satisfaction-Performance Relationship at the Individual Level of Analysis from Three Meta-Analytic Reviews Study Estimated Effect Size # of Studies Included Level of Analysis Total N 95% Confidence Interval* Vroom (1964) r = .14 15 Individual 3,333 .106 to .174 Petty, McGee, & Cavender (1984) r = .23 20 Individual 6,372 .205 to .255 Iaffaldano & Muchinsky (1985) r = .17 74 Individual 12,19 2 .152 to .188 *NOTE: Confidence intervals calculated using total sample size (N) Proposed Explanations for the Satisfaction-Performance Relationship If one accepts the conclusion evidenced by researchers such as Vroom (1964) and Iaffaldano and Muchinsky (1985), it becomes important to seek an accurate explanation for this apparent lack of relationship. In the following section, several explanations for the findings pertaining to the relationship between these two variables are discussed. From this discussion, several key empirical hypotheses are derived, which will be tested in the current study. Operationalization of satisfaction Regarding Vroom’s (1964) and Iaffaldano and Muchinsky’s (1985) conclusions that there is a weak relationship between job satisfaction and performance, Wright and Cropanzano (2000) proposed “this relegation may be premature” (p. 85). They suggested the observed relationship between these two variables is not strong owing to a flaw in the manner in which satisfaction has been operationalized. Rather than operationalize this variable using a narrow measure, such as job satisfaction, Wright and Cropanzano proposed using a more comprehensive measure of happiness which captures satisfaction with various aspects of life. Indeed, when they used a measure of happiness instead of job satisfaction, they found a stronger relationship with performance (r = .32). Similarly, Staw, Sutton, and Pelled (1994) analyzed data collected through the University of Michigan’s Survey Research Center and found employee positive emotion was significantly related to work achievement in the form of supervisor ratings of performance (r = .16) and pay (r = .24). While the theoretical basis for examining the relationship between life satisfaction and job performance appears sound, replication of findings such as these is necessary to strengthen our confidence in these conclusions and correct for any possible sampling errors. In particular, replication of the study by Wright and Cropanzano (2000) would be beneficial because these researchers based their conclusions on a small sample (n = © 2006 Institute of Behavioral and Applied Management. All rights reserved. 22 47) of a specialized group of employees (e.g., human service workers), used only an affective measure of well-being, and used a single item to assess the supervisor rating of job performance. In Staw et al.’s (1994) work, positive emotion was assessed using multiple sources (e.g., self-report and observer-report), however, the majority of the items focused on affective reports of signs of depression. In addition, while the researchers used supervisor evaluations to assess job performance, they experienced a very large drop-out rate of supervisors between time 1 and time 2 of their data collection process. Thus they were able to use data for only 60 employees to generate their estimates of the effect size. The current study is designed, in part, to test the theoretical proposition underlying studies such as these, namely, that life satisfaction (a.k.a. happiness) makes an important contribution to the prediction of job performance beyond that of job satisfaction. Operationalization of performance Several researchers have suggested the reason strong consistent results for the relationship between job satisfaction and performance have not been found is attributable to the manner in which performance is measured (e.g., Fisher, 1980; Organ, 1977). Specifically, they argued that performance has typically been measured in a narrow manner, which focuses on the degree to which an employee reaches either a quantity requirement or satisfies some preconceived standard for effective behavior (i.e., in-role performance). They advocate that, instead, performance should be conceptualized as not only a satisfaction of preconceived requirements, but should also include the performance of “spontaneous and innovative behavior,” which Kahn and Katz (1960) felt was important to organizational effectiveness. Kopelman, Brief, and Guzzo (1990) suggested three categories of pertinent behaviors for measuring organizational effectiveness: (1) attachment – attending to and staying in the organization; (2) performance – job-related tasks and activities comprising of the employee’s formal organizational role; and (3) citizenship – cooperation and collaborative efforts. Part of the argument for using organizational-level measures of performance, instead of individual-level measures, is that individual performance may be constrained or facilitated by situational factors that can confound the relationship, such as budgetary support, time availability, job related information, or individual factors (ability, effort, or personality). Furthermore, several researchers have pointed out that most performance measures are restricted in range because of organizational attempts to constrain the work behavior and performance of employees (e.g., Johns, 1991; Staw, 1984). It is possible that employees experiencing dissatisfaction at work will react in ways other than lower performance that will not be observed in traditional studies of the satisfaction-performance relationships, such as filing a grievance, attempting to improve their performance, or asking for a transfer (Locke & Latham, 1984). Specifically, Fisher (1980) suggested that outcomes such as attendance, predictability, compliance, following of rules, cooperation, drug use, sabotage, and so forth also be included in the measurement of individual organizational performance. Similarly, Organ (1988) proposed that job performance should also contain a measure of the extent to which employees engage in organizational citizenship behaviors (OCBs), which he defined as behaviors which are (1) discretionary; (2) not directly or explicitly recognized © 2006 Institute of Behavioral and Applied Management. All rights reserved. 23 by the formal reward system; and (3) in the aggregate, promote the effective functioning of the organization. Organ’s definition of OCB goes far beyond the typical measure of job performance and measures the following additional factors: altruism, courtesy, sportsmanship, conscientiousness and civic virtue. At the time Organ (1988) introduced this concept, he declared that research suggests “OCB and job satisfaction are bound together in a robust relationship.” The first two empirical studies which examined the relationship between job satisfaction and OCB did, indeed, support this proposition that the correlation between satisfaction and organizational citizenship behavior was stronger than the correlations between satisfaction and performance published in previous studies (Bateman & Organ, 1983; Smith, Organ, & Near, 1983). Since that time, more than 50 studies have been published that investigate at least one of the five dimensions of OCB and its relation to job satisfaction. In a meta-analysis of 55 studies in this area, Organ and Ryan (1995) calculated average effect sizes for the five components of OCB. While most of the studies they examined included only measures of altruism and compliance, there were five studies that reported results for the other three aspects of OCB; however, these findings are less conclusive. Overall, their findings support the conclusion that there is a robust relationship between satisfaction and OCB and that this relationship is, in fact, stronger than the relationship between satisfaction and in-role performance (r = .24 for altruism; r = .22 for compliance; r = .24 for courtesy, r = .21 for civic virtue; r = .28 for sportsmanship). The uncorrected estimated effect sizes for the relationships between satisfaction and these five measures of OCB exceed the uncorrected estimate calculated by Iaffaldano and Muchinsky (1985); moreover, the 95% confidence intervals for these findings (.20 to .27 for altruism, .18 to .25 for compliance, .20 to .29 for courtesy, .16 to .26 for civic virtue, and .23 to .32 for sportsmanship) do not overlap with the 95% confidence interval for Iaffaldano and Muchinsky’s uncorrected estimate (.123 to .169), with the exception of a slight overlap with the confidence interval for civic virtue. One could argue, however, that while these effect sizes are larger than those reported by Vroom (1964) and Iaffaldano and Muchinsky (1985), they do not significantly differ from that reported by Petty et al. (1984) (r = .23). Organ and Ryan (1995), however, feel their results can be more accurately compared to the effect size estimate reported by Iaffaldano and Muchinsky (1985) because their finding is considered more robust due to the fact that their research not only covered a longer time period, but also included a considerably larger number of respondents (N = 12,192) than did Petty et al. (1984) (N = 3,140). In addition, Organ and Ryan (1995) calculated an average correlation of r = .38 when all measures of OCB across all studies were collapsed and treated as an overall measure of the variable. In sum, they conclude their research provides support for the hypothesis that measures of OCB would be more related to satisfaction than would in-role job performance, but add the caveat that these findings are likely to hold up better for nonmanagerial, nonprofessional employees due to the fact that most research related to OCBs is conducted on these types of samples. © 2006 Institute of Behavioral and Applied Management. All rights reserved. 24 Relationship Between Job Satisfaction and Life Satisfaction As discussed previously, there is empirical support for the idea that if one measures overall life satisfaction (i.e. happiness or overall well-being), rather than job satisfaction, there is likely to be a stronger relationship with performance. Prior to investigating the relative strength of the relationship between life satisfaction and job performance, however, it is beneficial to examine the nature of the relationship between job satisfaction and life satisfaction. Most research investigating this relationship is correlational, thus preventing an accurate determination of the direction of causality in the relationship. Initially, in a meta-analysis containing 34 studies with a combined sample size of 19,811, Tait, Padgett, and Baldwin (1989) reported the estimated relationship between job satisfaction and life satisfaction to be .44. To determine more accurately the direction of causality for this strong positive relationship, Judge and his colleagues (1993) used structural equation modeling and cross-lagged time designs. When looking at job and life satisfaction at the same point in time, Judge and Watanabe (1993) found evidence of a reciprocal relationship, but when lagged correlations were examined, there was evidence that life satisfaction was a significant predictor of job satisfaction five years later. A reasonable conclusion from this research would be that by making people happier with their lives overall, organizations are also increasing the job satisfaction of their employees and can, in turn, reap the benefits of having employees with high job satisfaction. In contrast, using structural equation modeling, Judge and Hulin (1993) found support for the hypothesis that the relationship between job satisfaction and subjective wellbeing (e.g., happiness) is reciprocally causal. They summarized the implications of their findings by stating: “to spend the majority of one’s waking hours at a dissatisfying job appears to have a significant influence on the overall experienced well-being in one’s life. Conversely, spending an equal amount of time at a satisfying and perhaps challenging and motivating job has a strong positive influence on one’s subjective well-being. The role of work and work satisfaction in one’s life should not be underestimated.” (Judge & Hulin, 1993, p. 414) Their investigation showed that individuals with happy lives were much more likely to be satisfied with their jobs (r = .41, p < .01, one-tailed; n = 253) and vice versa (r = .31, p < .01, one-tailed; n = 253). An examination of the difference between the chi-square values for the two models tested revealed that the two correlation coefficients were not significantly different, thus demonstrating that subjective well-being is not a significantly better predictor of job satisfaction than job satisfaction is of subjective well-being; subjective well-being is both a significant antecedent and a consequence of job satisfaction. Overall, the findings of the Judge and Watanabe (1993) and Judge and Hulin (1993) studies suggest the strong positive relationship between job and life satisfaction would best be interpreted as a bi-directional path. The implication of findings such as these is that it is possible that an employee who is dissatisfied with his or her job, but very satisfied with his or her life, may perform well on the job. Conversely, an employee who © 2006 Institute of Behavioral and Applied Management. All rights reserved. 25 is very satisfied with his or her job, but very dissatisfied with his or her life, may perform poorly on the job. Current Research Hypotheses In aggregate, the aforementioned research offers multiple potential explanations for the weak relationship between job satisfaction and job performance. Furthermore, it implies several suggestions for future research. While research exists that has investigated the superiority of broader measures of satisfaction and of broader measures of performance, no research exists that has attempted to examine both of these issues in a single empirical study. Overall, the current research proposes the relationship between satisfaction and performance is stronger than the weak positive correlation found by researchers such as Vroom (1964) and Iaffaldano and Muchinsky (1985) and that this correlation can be found to be substantially larger when the operationalization of the two variables is broader than in past research. Based on findings reviewed above from Wright and Cropanzano (2000) and Staw et al. (1994), it is expected that life satisfaction will be a better predictor of the outcome variables than is job satisfaction. The following hypotheses, therefore, are advanced: Hypothesis 1: (a) The relationship between life satisfaction and in-role job performance will be stronger than the correlation between job satisfaction and in-role job performance. (b) The relationship between life satisfaction and OCB will be stronger than the relationship between job satisfaction and OCB. (c) The relationship between life satisfaction and the combined measure of performance (in-role performance and OCB) will be stronger than the relationship between job satisfaction and the combined measure of performance. Research by individuals such as Organ and Ryan (1995) offer preliminary evidence that there is a significant positive relationship between job satisfaction and organizational citizenship behavior; therefore, the following hypotheses regarding the relationships between the predictor variables and organizational citizenship behavior are extended: Hypothesis 2: (a) The relationship between job satisfaction and OCB will be stronger than the relationship between job satisfaction and in-role job performance. (b) The relationship between life satisfaction and OCB will be stronger than the relationship between life satisfaction and in-role job performance. (c) The relationships between life satisfaction and job satisfaction and the combined measure of OCB and in-role job performance will exceed the relationships either satisfaction measure has with either individual performance assessment. Vroom (1964) discussed the possibility that the association between job satisfaction and job performance is affected by other variables not controlled for by the researchers. Fairly convincing evidence has been offered by Tait et al. (1989), Judge and Watanabe (1993), and Judge and Hulin (1993) suggesting there is a strong positive relationship between life satisfaction and job © 2006 Institute of Behavioral and Applied Management. All rights reserved. 26 satisfaction. To examine the possibility that life satisfaction operates as an intervening variable in the relationship between job satisfaction and performance, it must be determined that there is a relationship between life satisfaction and job satisfaction, therefore, the following hypothesis is proposed: Hypothesis 3: There will be a positive relationship between life satisfaction and job satisfaction. Table 2 summarizes the hypothesized relationships between the variables. Table 2: Anticipated Relative Strength of Hypothesized Relationships Job Satisfaction Life Satisfaction In-Role Performance + ++ Organizational Citizenship Behavior ++ +++ Combined Performance Measure +++ ++++