Business-Unit-Level Relationship Between Employee Satisfaction,Employee Engagement, and Business Outcomes: A Meta-Analysis

James K. Harter

The Gallup Organization

Frank L. Schmidt

University of Iowa

Theodore L. Hayes

U.S. Immigration and Naturalization Service

Based on 7,939 business units in 36 companies, this study used meta-analysis to examine the relationshipat the business-unit level between employee satisfaction–engagement and the business-unit outcomes of customer satisfaction, productivity, profit, employee turnover, and accidents. Generalizable relationshipslarge enough to have substantial practical value were found between unit-level employee satisfaction–engagement and these business-unit outcomes. One implication is that changes in management practicesthat increase employee satisfaction may increase business-unit outcomes, including profit.

Locke, in his seminal 1976 review of the job satisfaction liter-ature, noted that more than 3,300 articles had been published onthe topic of job satisfaction. A search of PsycINFO for the years1976 through 2000 revealed at least another 7,855 publications onthe subject. Most job satisfaction studies (and subsequent meta-analyses) have focused on the individual employee level as a unitof analysis. For example, researchers have found positive linkagesbetween general workplace attitudes and individual performanceoutcomes (Iaffaldano & Muchinsky, 1985). A recent meta-analysisshowed a substantial relation between individual job satisfactionand individual performance (



.30; Judge, Thoresen, Bono, &Patton, 2001). Even though it is more common to study employeeattitude data at the individual employee level, studying data at thebusiness-unit level is critical because that is the level at whichemployee survey data are typically reported to client organizations.Business-unit-level research also provides opportunities to estab-lish linkages to outcomes that are directly relevant to most busi-nesses. Important outcomes such as customer loyalty, profitability,productivity, employee turnover, and safety variables are typicallyaggregated and reported at the business-unit level. A final advan-tage to reporting and studying data at the business-unit level is thatsingle-item scores are similar in reliability to subscale or dimen-sion scores at the individual level of analysis because each work-group item score is an average across many different individual-level scores, making single-item scores quite reliable. Thus,employee surveys reported at a business-unit level are more effi-cient and less dependent on survey length because item-levelmeasurement error is less of a concern.

Business-Unit Analyses

Several researchers have looked at how aggregate business-unit-level employee satisfaction, pride in service, and customer orien-tation relate to customer perceptions of service (Johnson, 1996;Reynierse & Harker, 1992; Schmit & Allscheid, 1995; Schneider,1991; Schneider, Ashworth, Higgs, & Carr, 1996; Schneider &Bowen, 1992; Schneider, White, & Paul, 1998; Ulrich, Halbrook,Meder, Stuchlik, & Thorpe, 1991; Wiley, 1991). Other researchershave studied employee attitudes aggregated at the company level(Denison, 1990), the school level (Ostroff, 1992), and the bank-branch level (Ryan, Schmit, & Johnson, 1996) in relation to otherorganizational outcomes such as financials and employee turnover.Although findings of causal direction are unresolved, preliminaryevidence in individual studies generally suggests that aggregateemployee attitudes have positive relations with customersatisfaction–loyalty and financials, and there is a negative relation-ship between employee attitudes and employee turnover. Onepotential problem with such business-unit-level studies is limiteddata due to a limited number of business units (the number of business units becomes the sample size) or limited access tooutcome measures that one can compare across business units. Forthis reason, many of these studies are limited in precision andstatistical power; as such, results from individual studies mayappear to conflict with one another. Meta-analytic techniques canprovide the opportunity to pool studies together to clarify thestrength of effects and their generalizability.Over the course of the past 30 years, researchers with TheGallup Organization have conducted thousands of qualitative fo-cus groups across a wide variety of industries. The approach

James K. Harter, The Gallup Organization, Lincoln, Nebraska; Frank L.Schmidt, College of Business Administration, University of Iowa; Theo-dore L. Hayes, U.S. Immigration and Naturalization Service, Washington,DC.We thank Donald O. Clifton, Sangeeta Badal, and the members of theGallup Workplace Management group for their input and support. Theopinions expressed in this article are solely the authors’, not those of theU.S. Government.Correspondence concerning this article should be addressed to James K.Harter, The Gallup Organization, 301 South 68th Street Place, Lincoln,Nebraska 68510. E-mail: jim_harter@gallup.com

268

underlying this research has been founded on what might be called

“

positive psychology

”

(e.g., Seligman & Csikszentmihalyi, 2000)

—

specifically, the study of the characteristics of successful employ-ees and managers and productive work groups. In developingmeasures of employee perceptions, Gallup researchers have fo-cused on the consistently important human resource issues onwhich managers can develop specific action plans. Throughout theworkplace research conducted by Gallup researchers, both quali-tative and quantitative data have indicated the importance of thesupervisor or the manager and his or her influence over theengagement level of employees and their satisfaction with theircompany. In Gallup

’

s research, items measuring aspects of theenvironment that the supervisor can directly influence explainmost of the variance in lengthier job satisfaction surveys andlengthier employee opinion surveys. This finding has been mir-rored in individual-level meta-analyses (e.g., Judge et al., 2001), inwhich the specific facet of satisfaction most highly related toperformance has been satisfaction with the supervisor.

Hypotheses

The hypotheses examined in this study were as follows:

Hypothesis 1

: Business-unit-level employee satisfaction and engage-ment will have positive average correlations with the business-unitoutcomes of customer satisfaction, productivity, profit, employeeretention, and employee safety.

Hypothesis 2

: The correlations between employee satisfaction andengagement and business-unit outcomes will generalize across orga-nizations for all business-unit outcomes. That is, these correlationswill not vary substantially across organizations, and in particular,there will be few if any organizations with zero or negativecorrelations.

Method

Independent Variable Measures

This study used an instrument developed from studies of work satisfac-tion, work motivation, supervisory practices, and work-group effective-ness. The instrument, the Gallup Workplace Audit (GWA; The GallupOrganization, 1992

–

1999), is composed of an overall satisfaction itemplus 12 items that measure employee perceptions of work characteristics.These 13 items were developed to measure employee perceptions of thequality of people-related management practices in business units. Thecriteria for selection of these questions came from focus groups, research,and management and scientific studies of the aspects of employee satis-faction and engagement that are important and influenceable by the man-ager at the business-unit or work-group level. This article presents ameta-analysis of studies conducted by The Gallup Organization to calibratethe instrument

’

s relatedness to business-unit outcomes, generalizabilityacross organizations, and usefulness in differentiating more effective work groups from less effective ones in relation to a variety of desirable businessoutcomes.The term

employee engagement

refers to the individual

’

s involvementand satisfaction with as well as enthusiasm for work. Aside from theoverall satisfaction item, the GWA

’

s 12 items (Items 1

–

12 in Table 1)measure processes and issues that are actionable at (i.e., under the influenceof) the work group

’

s supervisor or manager (further elaborated on in theDiscussion section). Although these 12 items explain a great deal of thevariance in what is defined as

“

overall job satisfaction

”

in the literature andalthough as a composite measure they have high convergent validities(Mount, Colbert, Harter, & Barrick, 2000) with overall job satisfactionmeasures such as the Brayfield

–

Rothe Satisfaction Index (Brayfield &Rothe, 1951), we refer to them as measures of employee engagement todifferentiate these actionable work-group-level facets from the more gen-eral theoretical construct of

“

job satisfaction.

”

The GWA items are antecedents of personal job satisfaction and otheraffective constructs. As Kahn (1990) suggested, broadly defined constructssuch as job involvement (Lawler & Hall, 1970; Lodahl & Kejner, 1965),organizational commitment (Mowday, Porter, & Steers, 1982), or intrinsicmotivation (Deci, 1975) add to understanding employee perceptions of themselves, their work, and their organization. However, these understand-ings are too general to be easily applied in practice because they exist at adistance from the day-to-day experiences of employees within their work situation. That is, employees are proud of their company and satisfied withtheir job in part because their basic needs are met fairly consistently. As inKahn

’

s conceptualization, we see engagement occurring when individualsare emotionally connected to others and cognitively vigilant. Employeesare emotionally and cognitively engaged when they know what is expectedof them, have what they need to do their work, have opportunities to feelan impact and fulfillment in their work, perceive that they are part of something significant with coworkers whom they trust, and have chancesto improve and develop. Having a measurement tool with items that makesense to employees and managers is critical to employees

’

and managers

’

acceptance of the instrument

’

s results and for their motivation to takeaction as a result of feedback based on such items.The GWA was designed to reflect two broad categories of employeesurvey items: those measuring attitudinal outcomes (satisfaction, loyalty,pride, customer service intent, and intent to stay with the company) andthose measuring or identifying issues within a manager

’

s control that areantecedents to attitudinal outcomes. The GWA includes 1 outcome itemreferring to overall satisfaction with one

’

s company that can be seen as ageneralized summary of specific affect-based reactions to work (see alsoLocke, 1976, pp. 1300

–

1319). A meta-analysis by Wanous, Reichers, andHudy (1997) demonstrated that, even at the individual level, single-itemmeasures of overall satisfaction have moderate reliabilities (approximately.60). At the business-unit level, where responses are averaged across manyindividuals, the reliability of single items is higher (as we show later in theResults section).

Table 1

Items Comprising the Gallup Workplace Audit

Overall Satisfaction

—

On a five-point scale, where

“

”

extremelysatisfied

and

“

”

extremely dissatisfied,

how satisfied are you with(Name of Company) as a place to work?1. I know what is expected of me at work.2. I have the materials and equipment I need to do my work right.3. At work, I have the opportunity to do what I do best every day.4. In the last seven days, I have received recognition or praise fordoing good work.5. My supervisor, or someone at work, seems to care about me as aperson.6. There is someone at work who encourages my development.7. At work, my opinions seem to count.8. The mission/purpose of my company makes me feel my job isimportant.9. My associates (fellow employees) are committed to doing qualitywork.10. I have a best friend at work.11. In the last six months, someone at work has talked to me about myprogress.12. This last year, I have had opportunities at work to learn and grow.

Note.

–

269

BUSINESS-UNIT-LEVEL META-ANALYSIS

The 13 GWA statements are presented in Table 1. As a currentstandard, these statements are asked of each employee (median partic-ipation rate



77%) with six response options (1



strongly disagree



strongly agree

; the sixth response is unscored and is

don’t know/ does not apply

). As a satisfaction item, the 1st item is scored on asatisfaction scale rather than on an agreement scale. As a total instru-ment (sum or mean of Items 1

–

12), the GWA has a Cronbach

’

s alphaof .91 at the business-unit level of analysis (



4,172 business units).The meta-analytic convergent validity of the equally weighted mean (orsum) of Items 1

–

12 to the equally weighted mean (or sum) of additionalitems added to longer surveys measuring commonly known facets of job satisfaction and engagement is .91 (also at the business-unit level of analysis;



8,127 business units). The observed correlation of themean (or sum) of Items 1

–

12 with the overall satisfaction item averages.77 (at the business-unit level), and the true score correlation is .91. Assuch, additional facets related to overall satisfaction are likely to bestatistically redundant with Items 1

–

12. The above findings provideevidence that the GWA, as a composite measure, captures the generalfactor in longer employee surveys and most of the variance in overallsatisfaction assessments. (We further discuss the statistical relevance of additional antecedent items in the Discussion section.) In studying theappropriateness of analyzing the 12 antecedent items as a unidimen-sional construct, we conducted a factor analysis across business unitsand companies and found the ratio of the first to second eigenvalues tobe 5.9 times the ratio of the second to third eigenvalues (sufficientaccording to a definition of unidimensionality provided by Lord, 1980).

Description of the Data

The Gallup database contains 42 studies conducted in 36 independentcompanies. In each GWA study, 1 or more of the GWA core items wereused (83% included all 12 items, and 90% included 11 or more items), andstarting in 1997, all items were included in all studies. For each businessunit, its score on the employee engagement variable was the average acrossthe GWA items that were administered. Its score on overall satisfactionwas its mean score on the single overall satisfaction item. Satisfaction datawere aggregated at the business-unit level and were correlated withbusiness-unit performance measures within each company: customersatisfaction

–

loyalty, profitability, productivity, turnover, and safety inci-dents. Again, the level of analysis in these studies was the business unit, notthe individual employee.Pearson correlations were calculated to estimate the correlation of business-unit mean of employee satisfaction

–

engagement with each of these five general business outcomes (described below). Correlations werecalculated across business units within each company, and these correlationcoefficients were entered into the meta-analysis database. We then calcu-lated mean validities, standard deviations of validities, and validity gener-alization statistics for these two composite indices for each of the fivebusiness-unit outcome measures.Tables 2 and 3 provide summaries of the studies included in thismeta-analytic study. Table 2 presents the number of companies, businessunits, and respondents represented in each industry. There was consider-able variation in type of industry represented; companies from 21 indus-tries provided studies. Each of the five general government-industry clas-

Table 2

Number of Companies, Business Units, and Respondents by Industry Type

Industry type Companies Business units RespondentsFinancialDepository 3 1,163 8,656Security 2 69 2,606ManufacturingFood 2 35 2,781Instrument 1 8 164Paper 1 118 35,479Printing 1 14 420RetailAutomotive 1 80 1,384Building materials 1 42 4,340Eating 5 316 16,999Entertainment 1 106 1,051Food stores 2 184 16,483Furniture 1 275 28,175Miscellaneous 1 634 14,753ServicesBusiness 1 20 600Education 3 200 1,747Health 2 334 13,675Hotels 3 167 6,549Recreation 1 14 288Transportation and public utilitiesTrucking 1 96 6,213Communication 2 4,039 35,964Electrical 1 25 187Total financial 5 1,232 11,262Total manufacturing 5 175 38,844Total retail 12 1,637 83,185Total services 10 735 22,859Total transportation and public utilities 4 4,160 42,364Total 36 7,939 198,514

270

HARTER, SCHMIDT, AND HAYES