A multilevel examination of work-life practices: is more always better?

During the past decade, researchers have begun to analyze the combined effects of multiple human resource practices (or bundles), sometimes under the rubrics of high-performance work systems or strategic HRM (e.g., Becker and Gerhart, 1996; Delaney and Huselid, 1996). However, as noted by Perry-Smith and Blum (2000), these studies have consistently excluded work-life human resource practices, most likely because employers ignore work-life support as a legitimate mainstream human resource issue (Kossek, 2005). More recently, researchers have begun to document the effects of specific work-life practices (also called work-family, family-responsive, or family-friendly practices). For example, Baltes et al. (1999) recently meta-analyzed the results of research on flextime and the compressed workweek. Other studies have focused on the effects of on-site child care (e.g., Goff et al., 1990; Kossek and Nichol, 1992), eldercare referral (Wagner and Hunt, 1994), and telecommuting (e.g., Bailey and Kurland, 2002; Duxbury et al., 1998; Igbaria and Guimaraes, 1999). Only in the past few years have researchers begun to analyze the effects of multiple work-life practices in concert, or in "bundles."

Two streams of research on multiple work-life practices have emerged: (1) studies examining the comparative effects of specific practices or types of practices and (2) studies examining the collective effects of multiple practices, typically using an index of the number of practices adopted. Both streams of research as well as the current study can be classified as "policy impact" research (Kossek, 2005) in which the relationship between policy access or use and attitudes and behaviors is examined. Because there has been very little research that focuses on the relationship between the number of different work-life practices offered by an organization and important outcomes of interest (e.g., organizational commitment), we examined the following research question: Is there a threshold level of practices such that additional practices have minimal incremental benefits? That is, if a bundle of work-life practices psychologically serves to signal to employees that their employer cares about them, is there some threshold number of practices that optimizes employee attitudes and perceptions? Although our work is theoretically grounded, we are primarily interested in whether there is an empirical relationship between the number of work-family practices offered by an organization and employee attitudes and perceptions that are related to important organizational outcomes. We focus on availability of programs rather than usage, as there is evidence that simply offering these practices can have a positive impact on employee attitudes, regardless of whether employees actually use the programs of policies (Grover and Crooker, 1995). In addition, because no prior research has examined these relationships using a multilevel design, we collected data at both the individual and work-group levels, and offer a conceptual justification for this added complexity.

In the following section, we review the two streams of research on multiple work-life practices, and describe the theory underlying our research. We then describe our methodology, including data collection from both employees and HR managers. After our results are presented, we offer suggestions for both future research and practice.

REVIEW OF THE LITERATURE AND CONCEPTUAL FRAMEWORK

A study by Meyer, Mukerjee and Sestero (2001) is representative of the first stream of research on work-life practices. Using a sample drawn from Working Mother magazine's "The 100 Best Companies for Working Mothers," Meyer et al. (2001) examined the association between profits and each of nine work-life practices. They found that profits were related to the availability of paid sick days and job sharing, although, surprisingly, the relationship with job sharing was negative. When they examined a combination of variables related to benefit usage and benefit levels available, profitability was positively related to the amount of adoption assistance and the percentage of employees working at home, and negatively related to employee usage of job sharing and on-site childcare. The models explained similar amounts of variance ([R.sup.2] = .28 and .27, respectively). Perry-Smith and Blum (2000) examined three clusters of work-life practices offered, and their effect on perceived organizational performance. The effects of specific practices were difficult to ascertain, though, because (1) the clusters combined multiple practices, (2) the clusters were not readily interpretable (e.g., forms of day care assistance emerged in two of the three clusters), and (3) results were not presented on a cluster-by-cluster basis. Grover and Crooker (1995) examined the effects of four work-life practices (flexible hours/schedules, information about childcare assistance, assistance with childcare costs, and maternity/paternity leaves) on attitudinal variables. Although flexible scheduling was (weakly) associated with higher affective organizational commitment (Beta = .045; p < .10), none of the other three work-life practices was related.

In the second stream of research, the effects of multiple practices have been examined by creating an index of the number of work-life practices adopted or used. Lambert (2000) examined relationships among the number of benefits used by individuals in one organization and (1) organizational citizenship behaviors and (2) perceptions of organizational support. Lambert's list of 20 benefits, though, was not exclusively limited to family-responsive initiatives. Nonetheless, perceived benefit usefulness was related directly to both dependent variables. Allen (2001) assessed the effects of ten specific variants of two major categories of work-life practices (four forms of flexible work arrangements and six forms of dependent care) on perceived organizational family supportiveness and supervisory support. Both the number of benefits used and the number offered were modestly related to perceived organizational family support and perceived supervisory support (correlations ranged from .16 to .19). Similarly, Anderson, Coffey and Byerly (2002) used two indices to study the availability of five dependent care practices and five schedule flexibility practices, and their effects on multiple outcome measures, including work-family conflict. The one significant path coefficient, between schedule flexibility and work-interferes-with-family conflict, was only--.07. Along the same lines, Batt and Valcour (2003) created indices related to employee access to four types of dependent care practices and five types of work schedule flexibility practices. Neither index was related to work-interferes-with-family conflict.

Konrad and Mangel (2000) examined the effects of the total number of work-life initiatives offered (up to 19) on productivity, and found a correlation of. 11. Haar and Spell (2004) examined a composite index of the past, present, and intended use of six work-life practices. The number of work-life practices presently used was unrelated to work-interferes-with-family conflict (WIF) and modestly related to family-interferes-with-work conflict (FIW). Greenberger, Goldberg, Hamill, O'Neil, and Payne (1989) created an index comprised of the number of formal work-life programs used (the maximum being 20) and related scores to attitudinal variables, including their own measure of organizational commitment. The total number of benefits used index was weakly related in job satisfaction among single women and organizational commitment among married women.

Summarizing research to date, a number of conclusions appear warranted. First, effect sizes have tended to be small using the operational definitions provided by (Cohen, 1992)--i.e., correlations of. 10 to .30; path coefficients of .02 to .15. Second, various approaches to studying practices have been employed, including examining practices individually, in sets classified by types of practices (e.g., clusters), of by a total index number. Third, very few studies have examined relationships between the number of work-life practices and three important outcomes in the organizational behavior literature: perceived supportiveness (one study), affective organizational commitment (one study), and work-family conflict (three studies). (These three outcome variables are examined in the present research.) Fourth, prior studies have examined the overall directionality of relationships, but none has examined outcomes in relation to increasing numbers of work-life practices offered, and whether there exists a threshold beyond which adding more practices has diminishing returns.