Job Satisfaction and Job Characteristics Meta Analysis

7/25/2019 Job Satisfaction and Job Characteristics Meta Analysis

1/11


2/11

Journal of Applied Psychology1985, Vol. 70, No. 2, 280-289

Copyright 1985 by the American Psychological Association, Inc.0021-9010/85/S00.75

A Meta-Analysis of the Relation of Job Characteristicsto Job Satisfaction

Brian T. Loher and Raymond A. NoeMichigan State University

Nancy L. MoellerUniversity of Maryland

Michael P. FitzgeraldArthur Young and Company Detroit, Michigan

Job enrichment is one method that has been used to increase employee satisfactionand work motivation. Hackman and Oldham's (1976) job characteristics modelhas served as the foundation for many job enrichment efforts. In particular, aconsiderable amount of research has been devoted to the study of the jobcharacteristics-job satisfaction relation. The purpose of this study was to statistically

determine, using meta-analysis procedures, the true relation between jobcharacteristics and job satisfaction. The role of growth need strength (GNS) as apossible moderator of this relation was also investigated. Results indicated amoderate relation between job characteristics and job satisfaction. This relationis stronger for employees high in GNS. Situational characteristics appear to bemore important in determining satisfaction for employees low in GNS.

Many theorists and researchers have arguedthat one way to increase employee perfor-mance and satisfaction is by enriching theemployee's job. Job enrichment seeks to im-prove both employee performance and satis-faction by bu ilding greater scope for personalachievement and recognition and greater op-portunity for individual achievement andgrowth into employees' jobs. Thus, job en-richm ent can be viewed as an organizationalintervention designed to restructure jobs withthe intent of making them more challenging,motivating, and satisfying to the individual.The current emphasis on job enrichment is

caused, in part, by the fact that today'semployees tend to bring more abilities, higherexpectations, and a greater desire for self-responsibility to the workplace than did theirpredecessors (Walton, 1972). To match thecharacteristics of the job to the needs of theindividual, job enrichment efforts have gen-erally applied theories that describe how the

The authors would like to thank Neal Schmitt, Dan

Ilgen, and two anonymous reviewers for their constructivecomments.

Requests for reprints should be sent to Brian T. Loher,Department of Psychology, Psychology Research Building,Michigan State University, East Lansing, Michigan 48824-1117.

interaction of job characteristics and individ-ual characteristics are related to organization-ally desirable outcomes.

The theoretical basis for many currentenrichment efforts is the Hackman and Old-ham (1975, 1976) job characteristics model.In the model (see Figure 1), specific jobcharacteristics such as skill variety (the degreeto which a job requires a variety of activitiesto carry out the work) and task significance(the degree to which the job has a substantialimpact on the lives of other people) affect theindividual's experienced meaningfulness ofthe work, experienced responsibility for work

outcomes, and knowledge about the resultsof his or her work activities (Hackman &Oldham, 1976). These three critical psycho-logical states have, in turn, been linked tosuch outcome variables as internal w ork mo-tivation, job satisfaction, absenteeism, turn-over, and work quality(Ford, 1969; Hackman,Oldham, Janson, & Purdy, 1975; Wanous,1974). As originally conceived, the job char-acteristics model also included the individualcharacteristic of growth need strength (GNS)as a moderator of the relation between thecharacteristics of the job and the outcomevariables (Hackman & Oldham, 1976).Hackman and Oldham assumed that one ofthe most important work values is the job

280
https://www.researchgate.net/publication/246874015_Motivation_Through_The_Work_Itself?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/246874015_Motivation_Through_The_Work_Itself?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/246874015_Motivation_Through_The_Work_Itself?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==


3/11

JO B CHARACTERISTICS-JOB SATISFACTION RELATION 281

Five Core PsychologicalCharacteristics

Skill variety

Task Identity

Task significance

States Outcomes

F

~?elings of

eanlngfulness

responsibility

results

High Intrinsic

motivation

High quality work

Hgh satisfaction

Low absenteeism and turnover

Employee Growth Need Strength

Figure 1. Relations between core job dimensions, criticalpsychological states, and on-the-job outcomes (fromHackman & Oldham, 1976).

incumbent's need for personal growth anddevelopment through his or her job. E mp loy-ees with high GNS should respond morepositively to jobs that have high levels of thefive core dimensions (task significance, skillvariety, etc.) than employees with low GNS.Various other individual characteristics suchas locus of control, knowledge, and skill havealso been hypothesized to moderate the out-comes of enriched jobs (Oldham, Hackman,& Pearce, 1976; Wanous, 1974).

The results of research studies attemptingto evaluate the relation between job charac-teristics and employee outcomes have beenequivocal at best (White, 1978), possibly dueto the impact of situational variables thatvary across organizations (e.g., climate, workgroup norms). Also, individual characteristicsbesides GNS (e.g., authoritarianism or locusof control) and demographic variables (e.g.,

urban/rural socialization, educational level)may act as moderators (Pierce & Dunham,1976; Roberts & Click, 1981; White, 1978).

The typical procedure used to study mod-erators of the relation of job characteristicsand employee outcomes involves subgroupanalysis; participants are split into two ormore groups based on their standing on themoderator of interest. Correlations betweenjob characteristics and outcome variables arethen calculated and compared across groups(Stone, 1976). In the Hackman and Oldham(1975) m odel, the individual characteristic ofGNS was intended to function as a moderator.Indeed, GNS has emerged as something of amoderator in that GNS has generally influ-

enced the magnitude rather than the sign ofthe focal relationship (Aldag, Barr, & Brief,1981, p. 427). However, inconsistent findingshave plagued this facet of research. Because

of these inconsistent findings, Roberts andGlick (1981) strongly questioned the utilityof including GNS as a moderator of the jobcharacteristics-employee outcome relation-ship.

From a practical standpoint, the evidencedoes not appear sufficient to unequivocallysupport the view that increasing the presenceof core job dimensions (i.e., enriching thejob) results in beneficial organizational andemployee outcomes. Job enrichment mayresult in desirable outcomes for one group ofemployees (high GNS) and may be moresuccessful in organizations with certain char-acteristics.

The purpose of this study is to apply meta-analytical techniques to determine (a) thestrength of the relation between job charac-teristics and one employee outcome, job sat-isfaction, and (b) whether the relation betweenjob characteristics and job satisfaction ismoderated by GNS.

Schmidt and Hunter (1977) have demon-strated how meta-analysis can be used toestimate the true relation between variablesthrough identifying the extent to which vari-ance in observed correlation coefficientsacross studies is due to statistical artifactssuch as sampling error and unreliability inmeasurement. Furthermore, these procedureshave also been applied to subgroup analyses.First, meta-analysis procedures are applied

to the entire sample. I f a substantial amountof residual variance remains after correctionsfor statistical artifacts, the sample is separatedinto two or more subgroups on the basis oftheir score on the moderator variable ofinterest. Schmidt and Hunter (1977) defineda substantial amount of residual varianceas being when 25% or more of the totalobserved variance is not accounted for bystatistical artifacts. Mean weighted correla-tions and variance estimates are then calcu-lated for each subgroup. Statistical evidencecan be taken as support for the moderatingeffect when the average correlation variesbetween subgroups and residual variance inthe correlation coefficient approaches zero
https://www.researchgate.net/publication/232569795_Individual_Differences_and_Reactions_to_Job_Characteristics?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/232569795_Individual_Differences_and_Reactions_to_Job_Characteristics?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/237950714_The_Moderating_Effect_of_Work-Related_Values_on_the_Job_Scope-Job_Satisfaction_Relationship?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/237950714_The_Moderating_Effect_of_Work-Related_Values_on_the_Job_Scope-Job_Satisfaction_Relationship?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/232591217_Development_of_a_General_Solution_to_the_Problem_of_Validity_Generalization?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/232591217_Development_of_a_General_Solution_to_the_Problem_of_Validity_Generalization?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/232591217_Development_of_a_General_Solution_to_the_Problem_of_Validity_Generalization?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/232569795_Individual_Differences_and_Reactions_to_Job_Characteristics?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==https://www.researchgate.net/publication/237950714_The_Moderating_Effect_of_Work-Related_Values_on_the_Job_Scope-Job_Satisfaction_Relationship?el=1_x_8&enrichId=rgreq-29008712-8123-45d4-86a5-5f30f9c3be46&enrichSource=Y292ZXJQYWdlOzIzMjUwMjY2ODtBUzoxOTk2ODYyMDA3OTUxNDZAMTQyNDYyMDMwODg2OA==


4/11

282 LOHER, NOE, MOELLER, AND FITZGERALD

within the subgroups (Hunter, Schmidt, &Jackson, 1982).

MethodCompiling the Distributionof Observed Correlations

The first step in the cumulative analysis was to establishcriteria to define which studies should be included.Hackman and Oldham (1976) developed the Job Diag-nostic Survey (JDS) to measure the facets of jobs thatare present in their job characteristics model. Therefore,the first inclusion criterion was that only studies thatused the JDS or a JDS-like measure of perceived taskcharacteristics were included in the analysis. A secondcriterion pertained to the measure of job satisfaction.Only job satisfaction scales designed to measure thefeelings or subjective reactions of respondents were in-cluded. In addition, only studies reporting individualresponses rather than aggregates and zero-order correla-tions between job characteristics and job satisfaction wereincluded.

Academic and practitioner-oriented journals in theareas of psychology, human resource management, andorganizational behavior were reviewed for studies thatmet our inclusion criteria. The studies included in themeta-analysis are listed in the Appendix. The data re-corded from each study included: (a) sample size, (b) abrief description of the sample, (c) the name an d type ofsatisfaction measure used, (d) reliabilities (when reported)of the JDS task dimensions and satisfaction measure,and (e) the reported correlations between the JDS taskdimensions and job satisfaction. Studies that met ourinclusion criteria were classified as (a) studies of thedirect relation between job characteristics and job satis-faction (Type O or overall studies), and (b) studies of therelation between job characteristics and job satisfactionas moderated by GNS (Type M or moderated studies).

Data AnalysisBoth Type O and Type M studies were included in the

analysis for establishing the true relationship betweenjob characteristics and job satisfaction. If job character-istics-job satisfaction correlations were reported for high-and low-GNS groups but not for a combined sample,then each GNS group within the Type M studies wastreated as if it were a separate study. Only Type Mstudies were included in the moderator analysis.

Initially, separate analyses were conducted of the relationbetween each of the five task dimensions and job satis-faction. This was done to ascertain the true relationshipbetween each task characteristic and job satisfaction andto determine which, if any, of the task characteristicswere more strongly related to job satisfaction. A secondanalysis of the job characteristic-job satisfaction relationwas conducted using the average of correlations betweeneach of the JDS task dimensions and the job satisfactionmeasure. The rationale for combining the task dimensionswas based on the findings that the intercorrelationsamong task dimensions are generally high and positive(Hackman & Lawler, 1971). Additionally, Aldag et al.(1981) found that the dimensionality of the JDS remainsquestionable in view of the number of studies reporting

factor structures that are inconsistent with the originalfive factor solutions proposed by Hackman and Oldham.

The zero-order correlations between the task dimensionsand the job satisfaction measures are presented in Table1 for the Type O studies. The resulting average correlationbetween the job characteristics dimensions and job sat-isfaction (job characteristics index) is presented in thelast column of Table 1 for each study. The analogouscorrelations for Type M studies are presented in Table2. The mean estimates presented in Tables 1 and 2 arelikely to be underestimates of the job characteristic-jobsatisfaction relation for two reasons. First, these valuesare not corrected for unreliability and, second, the jobcharacteristics index is a mean rather than a compositevalue, which does not take into account the intercorre-lations among dimensions.

The first step in the meta-analysis was to calculate theaverage sample-weighted correlation between job com-plexity and satisfaction using the Schmidt-Hunter pro-cedures. The sample-weighted correlation coefficient givesmore weight to correlations that presumably have theleast sampling error (i.e., those with the largest samplesize), thereby giving the best estimate of the truerelationship between the variables of interest (Schmidt& Hunter, 1977).

The calculation of the weighted variance of the samplecorrelation coefficients and the calculation of the errorvariance w as the next step in the analysis. The weightedvariance of the sample correlation coefficients is theobserved variance in the distribution of the correlationcoefficients. The error variance is the variation in thecorrelations across studies that would be expected tooccur as a result of sampling error.

Sampling error is only one of several artifactual sourcesof variance that m ay account for differences in observedcorrelations across studies (Schmidt, Hunter, Pearlman,& Shane, 1979). Two other such artifactual sources thatwere of interest in the present study were differencesamong studies in predictor reliability and differencesamong studies in criterion reliability. Error of measure-ment has the effect of lowering the observed correlationbetween job characteristics and job satisfaction. Therefore,correction for unreliability in the measures of the jobcharacteristics, the job characteristics index, and satisfac-tion represented the next step in the analysis. For the jobcharacteristics index, the reliability for each study wascalculated by averaging across the reliabilities of the fivetask dimensions. The predictor and criterion reliabilitiesfor Type O and Type M studies are presented in Table3. Not all of the studies in the analysis reported reliabilitiesfor both the JDS task dimensions and/or the satisfactionmeasure (unmatched data). Therefore, the interactiveformula (Hunter, Schmidt, & Jackson, 1982) was usedto obtain the variance of the true score correlations aftercorrecting for unreliability in the JDS and job satisfactionmeasures.

The final step in the analysis was to calculate the 95%confidence interval for the true score correlation betweenjob characteristics and job satisfaction.

Results

A total of 28 studies that met the inclusioncriteria were found. A ll but one of the studies


5/11

JOB CHARACTERISTICS-JOB SATISFACTION RELATION 283

were published in the following journals:Academy of Management Journal, Adminis-trative Science Quarterly, Human RelationsJournal of Applied Psychology, Journal of

Business Research Organizational Behaviorand Human Performance, and Personnel Psy-chology. Table 4 presents the characteristicsof the 28 studies.

The number of studies in which the impactof job enrichment as an organizational inter-vention was assessed were quite small. Onlythree studies involved an actual change inthe job process or technology. A number ofother studies that reported the results of

actual job enrichment interventions werefound, but these articles did not report thestatistical information needed in order to beincluded in the present study (e.g., correla-

tions between task dimensions and job satis-faction). The studies were almost equally splitinto those in which the sample was perform-ing the same job N = 12) and those in whichdata were collected across a number of dif-ferent jobs N = 16). Whereas the JDS wasthe prevalent measure of job characteristics,no measure of job satisfaction was used con-sistently across all studies. Finally, only asmall number of studies examined the prob-

Table 1Correlations Between Job Characteristics and Job Satisfaction for Type O Studies

2.

3.4.

5.6.

7.

8.

9.

10.11.12.

13.14.

15.16.17.18.

19.20.21.

Study

Al dag& Brief (1975)Arnold & House

(1980)Brief & Aldag ( 1978)Caldwell & O'Reilly

(1982)Dunham (1977)Evans, Kiggundu &

House (1979)Griffin (1981) I

IIHackm an & Lawler

(1971)Hackman & Oldham

(1976)Katz(1978a)Katz (1978b)Katz & Van Maanen

(1977)Kiggundu (1980)O'Reilly, Parletee, &

Bloom (1980)Orpen (1979)Rousseau (1977)Rousseau (1978)Schmitt, Coyle, White,

& Rauschenberger(1978)

Schmitt & White 1

Sims &Szi lagyi ( 1976)Walsh, Taber, & Beehr

(1980) III

III

N

104

120155

88784

343129171

208

65830602094

3500138

7636

199271

411860766

48696

232

T.I.

.34

.37

.31

.41

.16

.15

.38

.66

.20

.22

.21

.22

.24

.34

.23

.36

.26

.05

.14

.24

.30

.28

.32

.33

T.S.

.43

.21

.27

.20

.26

.21

.24

.25

.23

.21

.33

.35

.54

.18

.27

.33

S.V.

.32

.28

.43

.34

.18

.16

.54

.69

.38

.32

.23

.23

.28

.57

.33

.27

.58

.38

.42

.31

.54

.32

.12

.20

Aut.

.51

.41

.26

.51

.27

.29

.35

.71

.39

.38

.28

.27

.40

.60

.39

.28

.48

.47

.40

.28

.26

.31

.24

.38

Fdbk.

.37

.43

.26

.4321

.24

.66

.66

.28

.38

.26

.26

.36-.06

.46

.32

.37

.37

.26

.17

.26

.12

.28

.20

JobCharacteristics

Index

.394

.34

.315

.392

.204

.22

.483

.68

.313

.302

.244

.246

.302

.332

.348

.316

.446

.29

.298

.266

.340

.258

.24

.278

Note. T.I. = Task Identity; T.S. = Task Significance; S.V. = Skill Variety; Aut. = Autonomy; Fdbk. = Feedback.* Unpublished study via personal communication.b Three separate studies in one publication.


6/11


lem of common method variance. The averagecorrelation between the measures of growthneed strength, task characteristics, and jobsatisfaction was .05, which suggests that the

correlations were not significantly inflatedbecause of the reliance on self-report mea-sures.

The results of the analysis of the relationbetween each of the five skill dimensions andjob satisfaction are presented in Table 5. Thecorrelations after correcting for sampling errorand reliability in the measures range from.46 for Autonomy to .32 for Task Identity.Controlling for statistical artifacts accountedfor less than 75% of the observed variance inall of the task characteristics, which suggeststhat the job characteristic-job satisfaction

relationship is moderated by other variables.An examination of the 95% confidence inter-vals indicated that no one task characteristicnecessarily has a stronger relationship with

job satisfaction than any other (see Figure 2).This gives further evidence for the lack ofdimensionality of the JDS, that is, the JDSmay be a measure of the overall complexityof the job rather than of specific job charac-teristics. These results support combining thetask dimensions in order to establish therelation between job complexity and jobsatisfaction.

The results of the analysis of the job char-acteristic-job satisfaction relation based onthe average of correlations between each ofthe JDS task dimensions (job characteristic

Table 2Correlations Between Job Cha racteristics and Job Satisfaction for Type M Studies

Study

High growth need strength1. Armenakis, Field, Holley,Bedeian & Ledbetter(1977)

2. Brief &Aldag( 1975)3. Griffin (1982)4. Griffin (1981) I

II5. O'Reilly & Caldwell

(1979)6. Orpen(1979)7. Pokorney, Gilmore, &

Beehr (1980)

8. Umstot, Bell, & Mitchell(1976)9. Wanous(1974)

Low growth need strength1. Armenakis, Field, Holley,

Bedeian, & Ledbetter(1977)

2. Brief & Aldag ( 1975)3. Griffin (1982)4. Griffin (1981) I

H5. O'Reilly & Caldwell

(1979)6. Orpen (1979)7. Pokorney, Gilmore, &

Beehr (1980)8. Umstot, Bell, & Mitchell

(1976)9. Wanous(1974)

N

2835486565

3718

44

5037

2735386464

3718

54

5037

T.I.

.47

.40

.48

.52

.73

.53

.29

.40

.27

.30

-.10.33.13.18.57

.31

.31

.17

.67-.07

T.S.

.41

.27

.64

.70

.29

.18

.29

.28

S.V.

.28

.47

.53

.56

.82

.87

.19

.34

.70

.50

.09

.35-.03

.60

.53

.63

.14

.09

.73

.15

Aut.

.80

.53

.74

.52

.72

.85

.46

.42

.71

.59

.03

.35

.23

.11

.73

.58 02

.23

.41-.09

Fdbk.

.66

.36

.84

.44

.71

.65

.28

.63

.76

.41

.38

.36

.13

.42

.63

.31

.09

.15

.48

.10

JobCharacteristics

Index

.553

.440

.648

.510

.745

.662

.298

.486

.628

.450

.10

.348

.115

.328

.615

.424

.140

.186

.514

.023

Note. T.I. - Task Identity; T.S. - Task Significance; S.V. = Skill Variety; Aut. = Autonomy; Fdbk. = Feedback.


7/11

JO B CHA RACT ERISTICS-JOB SATISFACTION RELATIO N 285

index) are presented in Table 6. The sample-weighted correlation coefficient was .39 aftercorrecting for attenuation due to unreliabilityin the task characteristic and job satisfaction

measures. The observed variance in the cor-relations was .0059. After correcting for vari-ance due to sampling error and unreliabilityin the measures, the variance in the correla-tions w as still .0028. These statistical artifac tsaccount for 53% of the observed variance inthe correlation coefficients. This indicatesthat 47% of the observed variance is due toother factors, one of which may be GNS. Ifboth job characteristics and job satisfactionhad been perfectly measured, we w ould expectto find a true distribution centered on .39,

with a standard deviation of .0712 and a 95%confidence interval ranging from .25 to .53.

The large amount of variance that re-mained unexplained in the overall analysis

indicated the necessity for subgroup analyses.As a result, subgroup analyses were perform edfor the high- and low-GNS groups (Type Mstudies). Results of the analysis of the influ-ence of GNS on the job characteristics-jobsatisfaction relationship are presented in Table7. The sample-weighted correlation for thehigh GNS group was .57. The observed vari-ance for the high GNS group was .0135. The true relationship between job characteristicsand job satisfaction was .68 after correctingfor unreliability of measurement. Approxi-

Table 3Reliability Data for J ob C haracteristics and Job Satisfaction Measuresfor Type O and Type M Studies

Mean jobsatisfaction

1. Brief* Aldag (1975)2. Griffin (1982)3. Pokorney, Gilmore, &

Beehr(1980)4. Wanous (1974)

.63

.93

.76

Type M studies

.62 .84

.73

.69

.91

.72

.63

.92

.75

Study T.I. T.A. S.V. Aut. Fdbk. r 'yy

Type O studies

1.2.3.4.

5.

6.

7.8.9.

10.11.

12.13.

Aldag & Brief (1975)Brief & Aldag (1978)Dunham (1977)Evans, Kiggundu, &

House (1979)Griffin (1981) I

IIHackman & Lawler

(1977)Katz(1978b)Kiggundu (1980)O'Reilly, Parlette, &

Bloom (1980)Rousseau (1978)Schmitt, Coyle, White, &

Rauschenberger (1978)Schmitt & WhiteWalsh, Taber, & Beehr

(1980) III

III

.63

.72

.52

.91

.97

.77

.72

.62

.59

.76

.61

.53

.55

.69

.63

.72

.50

.73.59

.36

.70

.57

.62.76

.53

.89

.96

.90

.83

.78

.63

.74

.51

.50

.41

.37

.69

.73

.53

.89

.96

.77

.75

.63

.54

.75

.51

.53

.67

.64

.63

.75

.38

.86

.96

.75

.71

.70

.62

.77

.48

.36

.73

.72

.64

.74

.49

.89

.96

.80

.75

.66

.55

.74

.54

.48

.59

.61

.87

.79.93

.76

.74

.88

.67

.79

.70

.74

.73

.69

.64

.90

.74

.94

.76

Note. T.I. = Task Identity; T.S. = Task Significance; S.V. = Skill Variety; Aut. = Autonomy; Fdbk. = Feedback.


8/11


Table 4Characteristics of Studies Used in Meta-Analysis

InterventionYes 3

No 25SampleMultiple jobs 16Same job 12

Measure of task characteristicsJDS 23JCI 5

Measure of satisfactionAlderfer ERG Scale 2Brayfield & Rothe 3Hackman & Lawler 3JDIWork 2JDSGeneral Satisfaction 4

Kunin Faces Scale 2Minnesota Satisfaction Questionnaire 2Other (e.g., Index of Organization Reactions,

Yale Job Inventory, Lawler & Hall) 10Assess common method variance

Yes 4No 24

Note. JDS = Job Diagnostic Survey; JCI = Job Charac-teristic Inventory; JDI = Job Descriptive Index; ERG =Existedness, Relatedness, Growth.

mately 97% of the observed variance wasaccounted for by sampling error and mea-surement unreliability. The lower and upper

limits for the 95% confidence interval were.64 and .72.

The sample-weighted correlation coefficientfor the low-GNS group was .32, and theobserved variance was .0368. After correctingfor variance due to sampling error and un-reliability, variance for the low-GNS groupwas still .0169 (approximately 46% of theobserved variance). This indicates that forthe low-GNS group, other factors besidesGNS may be affecting the relation betweenjob characteristics and job satisfaction. Aftercorrecting the sample-weighted correlationfor attenuation due to unreliability in themeasures of job characteristics and job sat-

isfaction, the correlation between the twovariables was .38, with the 95% confidenceinterval ranging from .08 to .68.

Discussion

The results of the analysis involving studiesof the job characteristics-job satisfaction re-lation appear to answer our question regardingthe level of the relation between job charac-teristics and job satisfaction. The correlationbetween the job characteristics index and jobsatisfaction is about .39. The relation betweeneach of the task characteristics and job sat-

Table 5Results of the Meta-Analysis by Skill Dimension

Variable T.I. T.S. S.V. Aut. Fdbk.

Sample-weighted correlation

coefficientObserved varianceVariance due to sampling

errorVariance corrected for

sampling errorVariance due to

unreliability in themeasures

Sample-weighted correlationafter correcting forunreliability in themeasures

Variance of distribution oftrue score correlations

Upper 95% confidence limitLower 95% confidence limit

.24

.006403

.002177

.004226

.000663

.32

.006389

.48

.16

.25

.003923

.001983

.002440

.000835

.38

.003587

.50

.26

.30

.013691

.002030

.011661

.001624

.41

.018664

.68

.14

.34

.009454

.001912

.007542

.001346

.46

.011106

.67

.25

.29

.011311

.002041

.009270

.001968

.41

.013863

.64

.18

Note. T.I. = Task Identity; T.S. = Task Significance; S.V. = Skill Variety; Aut. = Autonomy; Fdbk. = Feedback.Total sample size was 15,542.


9/11

JOB CHARACTERISTICS-JOB SATISFACTION RELATION 287

Table 6Results of the Overall Mela-Analysis

Sample-weighted correlationcoefficient .29

Observed variance .005882Variance due to sampling error .002058Variance corrected for sampling

error .003824Variance due to unreliability in the

measures .001046Sample-weighted correlation after

correcting for unreliability in themeasures .39

Variance of distribution of true scorecorrelations .005068

Upper 95% confidence limit .53Lower 95 % confidence limit .25

Note. Total sample size was 15,542.

isfaction ranges from .32 (task identity) to.46 (autonomy). Some unexp lained variancein the correlations between studies remainsafter correcting for variance due to the arti-facts of sampling error and unreliability inthe predictor and criterion measures. Thislevel of correlation should certainly offer

some support for efforts to increase job sat-isfaction through the use of job enrichment.However, the results from the moderatedstudies warn that simply enriching a job willnot necessarily hold the same amount ofbenefit for everyone.

The results for the studies with GNS as amoderator are particularly intriguing. We cannow state with some confidence that growthneed strength (GN S) acts as m oderator of therelation between job characteristics and jobsatisfaction. The correlation between jobcharacteristics and satisfaction is .68 for per-

Task Identity

Task Significance

Skill Variety

Autonomy

Feedback

.26i

1.64

OO .10 .20 .30 .40 SO .60 .70

Figure 2. 95% confidence intervals for task characteristics.

sons who are high on GN S and about .38 forpersons who are low on GNS. The variancein the correlations for the high GNS groupis essentially eliminated after controlling forsampling error and measurement unreliabil-ity. However, the variance in the correlationsbetween job characteristics and job satisfac-tion is still large for the low-GNS group , evenafter correcting for sampling error and un-reliability in the JDS and job satisfactionmeasures. Apparently, factors that do notaffect the relation between job characteristics

and job satisfaction for persons with highgrowth need strength do come into play forpersons who are low on growth need strength.The model in Figure 3 illustrates how sucha relation migh operate.

The more complex and enriched a job is,the more likely the high-GNS person whopossesses a high need for personal growthand development (Hackman & Oldham,1976) is to be satisfied with that job. Incontrast, for low-GN S employees, who haveless need for growth and development, thepresence of certain external situationa l char-

Table 7Results of the Moderator Analysis

Variable High GNS

Note. Total sample size for high-GNS group was 427. For low-GNS group, sample size was 424.

Low GNS

Sample-weighted correlation coefficientObserved varianceVariance due to sampling error

Variance corrected for sampling errorVariance due to unreliability in the measuresSample-weighted correlation after correcting for unreliability in the measuresVariance of distribution of true score correlationsUpper 95% confidence limitLower 95% confidence limit

.57

.013517

.010721

.002796.002458

.68

.000482

.72

.64

.32

.036792

.019083

.017709.000764

.38

.024179

.68

.08


10/11


Figure 3. A model of the proposed relation between jobcomplexity, GNS, and job satisfaction.

acteristics (such as work group or manage-ment support for enrichment activities) maybe necessary if the core job dimensions areto increase employee satisfaction. That is, theopportunities an enriched job offers, in andof themselves, may not be recognized orcared about by low-GNS employees. Whatmay be more important is how the employee'swork group views these opportunities. If thework group is supportive of enriched work,this may help to enhance the employee'ssatisfaction with a more complex job. Future

research is necessary to investigate hypothesesabout moderating situational characteristicsfor low-GNS persons.

An alternative explanation is that satisfac-tion influences how individuals describe theirjobs (the authors thank an anonymous re-viewer for this suggestion). Satisfied individ-uals may see their jobs as more complexthan dissatisfied individuals. The direction ofarrows in Figure 3 may be reversed.

Finally, the characteristics of the studiesreported in the literature are particularlydisturbing. Few empirical studies of actualjob enrichment interventions have been re-ported in the professional literature. Morestudies evaluating the impact of actual changesin job characteristics on employee attitudesand performance need to be conducted. Ad-ditionally, more attention needs to be directedto the impact of common method varianceon the job characteristics-job satisfaction re-lation. The use of multiple methods (e.g.,interviews, unobtrusive observation) to assess

the job characteristics-job satisfaction relationwill help to determine the extent to whichcommon method variance results in spu-riously inflated correlation coefficients.

ReferencesAldag, R. J., Barr, S. H., & Brief, A. P. (1981). Measure-

ment of perceived task characteristics. PsychologicalBulletin, 90 , 415-431.

Ford, R. N. (1969). Motivation through the work itself.New \brk: American Management Association.

Hackman, J. R., Lawler, E. E. (1971). Employeereactions to job characteristics. Journal of AppliedPsychology, 55 , 259-286.

Hackman, J. R., & Oldham, G. R. (1975). Developmentof the job diagnostic survey. Journal of Applied Psy-

chology, 60, 159-170.Hackman, J. R., & Oldham, G. R. (1976). Motivationthrough the design of work: Test of a theory. Organi-zational Behavior and Human Performance, 16 , 250-279.

Hackman, J. R., Oldham, G. R., Janson, R., & Purdy,K. (1975). A new strategy for job enrichment. CaliforniaManagement Review, 57-71.

Hunter, J. E., Schmidt, F. L., & Jackson, G. B. (1982).Meta-analysis: Cumulating research findings acrossstudies. Beverly Hills: Sage.

Oldham, G. R., Hackman, J. R., & Pearce, J. L. (1976).Conditions under which employees respond positively

to enriched work. Journal of Applied Psychology, 61 ,395-403.Pierce, J. L., & Dunham, R. B. (1976). Task design: A

literature review. Academy of M anagement Review, 4,83-97.

Roberts, K. H., & Click, W. (1981). The job characteristicsapproach to task design: A critical review. Journal ofApplied Psychology, 66 , 193-217.

Schmidt, F. L., & Hunter, J. E. (1977). Development ofa general solution to the problem of validity general-ization. Journal of Applied Psychology, 62 , 529-540.

Schmidt, F. L., Hunter, J. E., Pearlman, K., & Shane,G. S. (1979). Further tests of the Schmidt-Hunter

Bayesian validity generalization. Personnel Psychology,32 , 257-281.Stone, E. F. (1976). Moderating eifect of work-related

values in the job scope-job satisfaction relationship.Organizational Behavior and Human Performance, 15 ,147-167.

Walton, R. E. (1972). How to counter alientation in theplant. Harvard Business Review 50, 70-81.

Wanous, J. P. (1974). Individual differences and reactionsto job characteristics. Journal of Applied Psychology,59, 616-622.

White, J. K. (1978). Individual differences in job quality-worker response relationship: Review, integration, and

comments. Academy of Management Review, 3, 267-

280.


11/11

JOB CHARA CTERISTICS-JOB SATISFACTION RELA TION 289

Appendix

Studies Included in Meta-Analysis

Aldag, R. J., & Brief, A. P. (1975). Impact of individualdifferences on employee affective responses to taskcharacteristics. Journal of Business Research, 3, 311-321.

Armenakis, A. A., Field, H. S., Jr., Holley, W. H., Jr.,Bedeian, A. G., & Ledbetter, B., Jr. (1977). Humanresource consideration in textile work redesign. HumanRelations, 30, 1147-1156.

Arnold, H. J., & House, R. J. (1980). Methodologicaland substantive extensions to the job characteristicsmodel of motivation. Orga nizational Behavior andHuman Performance, 25 , 161-183.

Brief, A. P., & Aldag, R. J. (1975). Employee reactions

to job characteristics: A constructive replication. Journalof Applied Psychology, 60, 182-196.

Brief, A. P., & Aldag, R. J. (1978). T he job characteristicsinventory: A n examination. Academy of ManagementJournal, 21 , 659-670.

Caldwell, D. F., & O'Reilly, III, C. A. (1982). Taskperceptions and job satisfaction: A question of causality.Journal of Applied Psychology, 67 , 361-369.

Dunham, R. B. (1977 ). Reactions to job characteristics:Moderating effects of the organization. Academy ofManagement Journal, 20, 42-65.

Evans, M . G., Kiggundu, M. N., & House, R. J. (1979).A partial test and extension of the job characteristics

model of motivation. Organizational Behavior an dHuman Performance, 24 , 354-381.Griffin, R. W. (1981). A longitudinal investigation of task

characteristics relationships. Academy of ManagementJournal, 24 , 99-113.

Griffin, R. W. (1982). Perceived task characteristics andemployee productivity and satisfaction. Human Rela-tions, 35 , 927-938.

Hackman, J. R., & Lawler, E. E. (1971). Employeereactions to job characteristics. Journal of AppliedPsychology, 55 , 259-286.

Hackman, J. R., & Oldham, G. R. (1976). Motivationthrough the design of work: Test of a theory. Organi-

zational Behavior and Human Performance, 16 , 250-279.Katz, R. (1978a). The influence of job longevity on

employee reactions to task characteristics. HumanRelations, 31 , 703-726.

Katz, R. (1978b). Job longevity as a situational factor injob satisfaction. Administrative Science Qua rterly, 23 2),204-223.

Katz, R., & Van Maanen, J. (1977). Th e loci of worksatisfaction: Job, interaction, and policy. Human Re-lations, 30, 469-486.

Kiggundu, M. N. (1980). An empirical test of the theory

of job design using multiple job ratings. Human Re-lations. 33, 339-351.O'Reilly, C. A ., & Caldwell, D. (1979). Informational

influence as a determinant of task characteristics andjob satisfaction. Journal of Applied Psychology, 64 ,157-165.

O'Reilly, C. A., Parletee, G. N., & Bloom, J. R. (1980).Perceptual measures of task characteristics: The biasingeffects of differing frames of reference and job attitudes.Academy of Management Journal, 2 3, 118-131.

Orpen, C. (1979). The effects of job enrichment onemployee satisfaction, motivation, involvement, andperformance: A field experiment. Human Relations,

32, 189-217.Pokorney, J., Gilmore, D., & Beehr, T. (1980). Jobdiagnostic survey dimensions: Moderating effect ofgrowth needs and correspondence with dimensions ofjob rating form. Organizational Behavior and Huma nPerformance, 26 , 222-237.

Rousseau, D. (1977). Technological differences in jobcharacteristics, employee satisfaction, and motivation:A synthesis of job design research and sociotechnicalsystems theory. Orga nizational Behavior and HumanPerformance, 19, 18-42.

Rousseau, D. (1978). Characteristics of departments,positions, and industries: Contexts for attitudes and

behavior. Administrative Science Quarterly, 23 , 521-540.

Schmitt, N., Coyle, B., White, J. K., & Rauschenberger,J. (1978). Background needs, job perceptions, and jobsatisfaction: A causal model. Personnel Psychology, 31,889-901.

Sims, H. P., & Szilagyi, A. D. (1976). Job characteristicsrelationships: Individual and structural moderators.Organizational Behavior and Human Performance, 17 ,211-230.

Umstot, D., Bell, C. H., & M itchell, T. R. (1978). Effectsof job enrichment and task goals on satisfaction andproductivity: Implications for job design. Journal of

Applied Psychology, 61, 379-394.Walsh, J. I., Taber, T. D., & Beehr, T. A. (1980). An

integrated model of perceived job characteristics. Or-ganizational Behavior an d Human Performance, 25 ,252-267.

Wanous, J. P. (1974). Individual differences and reactionsto job characteristics. Journal of Applied Psychology,59, 616-622.

Received May 25, 1984Revision received July 30, 1984

Documents

Job Satisfaction and Job Characteristics Meta Analysis