A Meta-Analytic Review of the Relationship Between Absence and Turnover

Atul Mitra, G. Douglas Jenkins, Jr., and Nina GuptaDepartment of Management, College of Business Administration, University of ArkansasA meta-analysis that combined 33 correlations from 17 studies found a corrected average correla-tion of .330 between absence and turnover. The type of absence measure did not moderate thecovariation between absence and turnover, but industry type and study duration did act as modera-tors. The results are discussed in the context of the "withdrawal" approach to understandingemployee behaviors.Absenteeism and turnover are two of the most popular out-comes studied in organizational research. It is hardly surprisingthat there is considerable debate about the true relationshipbetween these outcomes. One school of thought argues thatunder certain circumstances, absenteeism and turnover reflectdifferent manifestations of the same underlying construct (w#/z-drawal) and sees merit in examining them together (e.g., Beehr& Gupta, 1978; Gupta & Jenkins, 1982; Rosse, 1988; Rosse H Wolpin & Burke, 1985). An opposing school ofthought is strongly critical of this approach (e.g., Mobley, 1977,1982; Price, 1977; Price & Mueller, 1981; Steers & Mowday,1981), arguing that each behavior is unique and should be exam-ined individually. In this article, we use a meta-analytic ap-proach to begin resolving this debate. Specifically, we explorewhat light the bulk of research evidence sheds on the ab-senteeism-turnover covariation. This exploration should alsoclarify the relative validity, or lack thereof, of the entire with-drawal construct.Researchers have proposed three relationships between ab-senteeism and turnover. The first position asserts that there is acontinuum of withdrawal behaviors progressing from ab-senteeism to turnover (Herzberg, Mausner, Peterson, & Cap-well, 1957), the second asserts that absenteeism and turnoverare alternatives to each other (Hill & Trist, 1953), and the thirdasserts that there is no relationship between the two (March S Wolpin & Burke, 1985).Opponents of the withdrawal construct assume that there isno inherent relationship between absenteeism and turnoverand offer several supporting arguments. For instance, Mobley(1982) claimed that the term withdrawal has an implicit conno-An earlier version of this article received the Best Paper Award at the1990 Southern Management Association meeting in Orlando, Florida.We thank Jack Hunter, David Roach, Bob Rodgers, Ellen Whitener,and three anonymous reviewers for helping to untangle some of themeta-analysis complexities and providing helpful comments on earlierdrafts. We particularly want to thank one anonymous reviewer forsuggesting the unemployment analyses.Atul Mitra is now at Arkansas College, Batesville, Arkansas.Correspondence concerning this article should be addressed to G.Douglas Jenkins, Jr., Department of Management, College of BusinessAdministration, University of Arkansas, Fayetteville, Arkansas 72701.tation of escape or avoidance but that both turnover and ab-senteeism can be motivated by attractions of alternatives ratherthan avoidance. Thus, the term is clearly not descriptive of allturnover and absenteeism behaviors. This position is bolsteredby the results of an early landmark review (Porter & Steers,1973): Of 22 cases in which antecedents of turnover and ab-senteeism were studied concurrently, only six significant rela-tionships occurred in the same direction for both absenteeismand turnover. With very few conditions under which with-drawal behaviors are manifested, the utility of the withdrawalconstruct is limited (Mobley, 1982), and the antecedent pro-cesses of absenteeism and turnover rather than those of thestructural relationships between these behaviors are empha-sized.Proponents of the withdrawal construct summon several ar-guments to support the underlying commonalities between ab-senteeism and turnover. Rosse and Hulin (1985) noted thatthere is sufficient evidence to conclude that job attitudes under-lie a spectrum of withdrawal or adaptive behaviors. Likewise,Rosse and Miller (1984) reasoned that both of these behaviorshelp reduce dissatisfaction with the work role. Gupta and Jen-kins (1991) contended that examining absenteeism and turn-over individually suffers from criterion contamination and cri-terion deficiency in that both absenteeism and turnover en-compass voluntary and involuntary behaviors, and in that thetwo behaviors might serve as alternatives, depending on organi-zational and other constraints. Furthermore, the information-processing literature indicates that people react more stronglyto negative than to positive information (e.g., Webster, 1982)such that a single negative outcome may overshadow a multi-tude of positive outcomes (Leon, 1981). That is, noxious organi-zational stimuli may in fact have a greater impact on an em-ployee's decision to withdraw than do attractions outside theorganization. In short, this position advocates the study of mul-tiple behaviors in conjunction rather than each behavior in iso-lation.At least three reviews of the literature examined the relation-ship between absenteeism and turnover (Gupta & Jenkins,1980; Lyons, 1972; Muchinsky, 1977). All three concluded thatempirically, absenteeism and turnover are positively related,particularly at the individual level of analysis. Methodologicalproblems in the way research evidence was cumulated in thesereviews, however, reduce confidence in their conclusions