Using the Theory of Planned Behaviour to explain work-life balance program utilisation

Size: px
Start display at page:

Download "Using the Theory of Planned Behaviour to explain work-life balance program utilisation"

Transcription

1 Page 1 of 24 ANZAM 2009 Using the Theory of Planned Behaviour to explain work-life balance program utilisation Donald Ting School of Psychology, The University of Western Australia, Crawley, Australia tingt2@student.uwa.edu.au Dr Elliot Wood School of Psychology, The University of Western Australia, Crawley, Australia elliot@psy.uwa.edu.au Peter Sevastos School of Psychology, The University of Western Australia, Crawley, Australia peter@psy.uwa.edu.au

2 ANZAM 2009 Page 2 of 24 ABSTRACT Although work-life balance (WLB) programs are available in most organisations, they are seldom utilised. Organisational support, manager support, co-worker support, organisational time demand and career consequences are some barriers found to restrict WLB program utilisation. Instead of a direct relationship between these barriers and WLB program utilisation, it was proposed that barriers influence WLB program utilisation through the attitude, subjective norm and perceived behavioural control constructs of the Theory of Planned Behaviour. A path analysis using data from 242 employees of a State Government agency supported this hypothesis. Theoretical and practical implications of this study are discussed. Work-life balance (WLB) refers to the maintenance of balance between work and family responsibilities so that role conflicts between them are reduced (De Cieri, Holmes, Abbott & Pettit 2005). Individuals with healthy WLB experience less stress, and show greater satisfaction with life in general compared to those with poor WLB (Hobsor, Delunas, & Kesic 2001). With organisational benefits of WLB such as reduced turnover, and greater job satisfaction (Thomas & Ganster 1995), organisations are increasingly using WLB programs to offer a range of flexible work arrangements, such as accrued days off, career breaks, parental leave and telecommuting into the workplace. In a national WLB benchmarking study on 284 organisations throughout Australia (Managing Work/Life Balance, 2007), 75% of the surveyed organisations agreed that with WLB programs, there have been increases in employee motivation, satisfaction and engagement. Sixty-eight percent of the surveyed organisations agreed that through WLB programs, employees are able to better manage their work stress. On average, respondents reported a 7% reduction in turnover and a 9% reduction in absenteeism through WLB programs. On the flipside, studies have also reported that although WLB programs are available to employees, the utilisation rate of these programs has been relatively low (Nord, Fox, Phoenix, & Viano, 2002; Thompson, Beauvais, & Lyness, 1999). Using an Australian sample, De Cieri et al., (2005) found that out of 358 Australian organisations, 50% had less than 20% of their employees utilising the available WLB programs and only 6% of the organisations had more than 80% of employees using the available

3 Page 3 of 24 ANZAM 2009 WLB programs. As a result, researchers have begun to explore factors that inhibit WLB program utilisation. With such low utilisation, it seems fair to suggest that the maximum benefits are not being achieved from WLB programs. Researchers have since explored factors that can inhibit the utilisation of WLB programs and have frequently identified low organisational support, low manager support, low co-worker support, perceived negative career consequences and high organisational time demand as organisational barriers to WLB programs utilisation (e.g., De Cieri et al., 2005; Kirby & Krone, 2002; Nord et al., 2002; Smith & Gardner, 2007; Thompson et al., 1999). Organisational support refers to the extent in which an organisation is supportive and sensitive to employees needs to maintain WLB. Allen (2001) found that employees who perceived their organisation to be family-supportive utilised more family supportive programs in their workplace. Manager support refers to the extent to which managers are supportive and sensitive to employees family responsibilities. Research has shown that WLB programs can be ineffective if supervisors are not in favour of them (Thomas & Ganster, 1995). Other research on manager support has reported that supportive managers are associated with higher utilisation of WLB programs (Allen, 2001; Kirby & Krone, 2002; Smith & Gardner, 2007). Thompson et al. (1999) found managerial support to be positively related to WLB program utilisation. Smith and Gardner (2007) also found evidence supporting a positive relationship between WLB program utilisation and supervisor support. Co-worker support concerns support and sensitivity to fellow employees family responsibilities. Kirby and Krone (2002) found that unsupportive co-workers did not want others to utilise WLB programs mainly because it led to increase in their personal work responsibilities. In relation to the relationship between co-worker support and WLB program utilisation, researchers have found co-worker support to be related to higher program utilisation. For example, Kossek, Barber, and Winters (1999) found that managers with peers who used WLB programs, utilised more programs themselves. Some researchers have found co-worker support to have no influence on WLB programs utilisation (e.g., Smith & Gardner,

4 ANZAM 2009 Page 4 of ). Howeever on closer examination of Smith and Gardner s study, items measuring co-worker support captured co-worker support in relation to general work issues (e.g., your co-workers are helpful in getting job done ) instead of capturing co-worker support in relation to WLB programs, a possible explanation for a null finding. Organisational time demands refer to the expectation that employees prioritise time for work commitments over non-work commitments. In organisations with high organisational time demand, employees are expected to work long hours and are often required to bring work home (Thompson et al., 1999). As a result, employees may feel that they are unable to make use of WLB programs even if they want to. In support of this view, Smith and Gardner (2007) found that employees who experienced high organisational time demand reported utilising WLB programs less. Perceived negative career consequences refer to perceptions about career consequences as a result of using WLB programs. As suggested by Perlow (1995), the time employees spend at work is often taken by managers to be a direct indicator of the employees effort and commitment to work. By utilising WLB programs, less time is spent at work, resulting in potential negative appraisals from managers which can jeopardise employees career. In support of this view, engineers at a Fortune 100 company indicated that they were reluctant to utilise WLB programs for the fear of negative career consequences (Perlow, 1995). Similarly, Almer, and Kaplan (2000) found that potential career damage due to using WLB programs discouraged employees from using the programs. Smith and Gardner (2007) provided further evidence when they found that employees who perceived greater career damage due to using WLB programs reported less utilisation of the programs. Ziemer and Wood (2007) also found that negative career consequences predicted the use of flexible working arrangements. Whilst research on WLB program utilisation converges on the relevance of key organizational barriers, researchers have until now presumed a direct model of barrier influence (e.g., Allen 2001; Smith &

5 Page 5 of 24 ANZAM 2009 Gardner 2007). However this ignores extensive theoretical work on the prediction of behaviour, specifically the Theory of Planned Behaviour (TPB) (Ajzen 1991) which proposes that, instead of a direct relationship between external factors and behaviour, predictions about behaviour need to consider instead a person s intention to perform a behaviour and their perceived behavioural control. Intention in turn is influenced by attitudes (the overall evaluation of the behaviour), subjective norms (the perceived social pressure to perform the behaviour) and perceived behavioural control as shown in the shaded area of Figure 1. These three TPB elements are influenced by external factors such as personal beliefs, social influences and environmental constraints. Several researchers have provided evidence of the mediating role of attitude, subjective norm and perceived behavioural control in predicting behaviour (e.g., Christian, Armitage & Abrams 2007) and the predictive utility of the TPB has been established in numerous studies (Armitage & Conner 2001). Insert Figure 1 here Less research on the TPB has focused on predicting behaviour in organisations, and none to date on WLB utilisation. This study aims to examine the utility of the TPB in explaining the link between barriers and WLB program usage. Since, according to Ajzen (1991), elements of the TPB are themselves influenced by external factors, we contend that such barriers may act as important external factors influencing elements of the TPB. Based on the TPB, we propose that barriers have an indirect effect on a person s WLB program utilisation, influencing intention and perceived behavioural control which in turn drive WLB program utilization (see the shaded area in figure 1). Proving an indirect model would allow HR practitioners to focus intervention strategies on areas that most strongly influence employees decisions to use WLB programs. That is, HR actions influencing direct antecedents like attitude, subjective norms and perceived behavioural control (rather than distal antecedents such as the barriers mentioned above) would, if the indirect model is correct, impact more strongly on program utilisation.

6 ANZAM 2009 Page 6 of 24 Based on prior research, we posit a specific relationship between certain barriers and attitude, norms and perceived behavioural control. With regards to organisational support and manager support, since low levels of organisational and managerial support have previously been found to reduce the perceived control employees have in utilising WLB programs (Thomas & Ganster 1995) we hypothesize that support from the organisation and managers towards WLB will most strongly influence perceptions of behavioural control. On the other hand, social learning theory suggests that social norms develop as employees learn about the social environment through the observation of others (Cialdini 2003). Employees observe their co-workers and adopt relevant beliefs about using WLB programs - when coworkers are perceived to be unsupportive of WLB, it creates strong social pressure against the use of WLB programs (see Kossek, Barber & Winters 1999). Therefore, we suggest that co-worker support towards WLB influences subjective norms about WLB program use. With long working hours, employees do not have the opportunity to make use of WLB programs even if desired (Smith & Gardner 2007). Such a loss of control when organisational time demand is high suggests this barrier influences perceived behavioural control over WLB program utilisation. Finally, it is expected that both negative career consequences and support from the organisation and managers influence attitudes towards WLB program use. Since attitudes towards a behaviour are generally formed based on the associated consequences of that behaviour (Ajzen, 1991), employees who perceive negative career consequences from using WLB programs can be expected to form negative attitudes towards WLB program use (Ziemer & Wood 2007). In an organizational context, if the organisation or managers are not supportive of WLB, the consequences of program use are also likely to be seen as negative, leading to negative attitudes towards using WLB programs (Smith & Gardner 2007).

7 Page 7 of 24 ANZAM 2009 METHOD Sample An online survey was sent to 772 employees in a State government agency. The 242 responses consisted of 114 males (47.1%) and 128 females (52.9%), mean age of 36 to 40 years old. Slightly more than half (54.1%) had no dependants and 70.2% were either married or in a relationship. The sample closely corresponded to the demographics of the agency. Measures Barriers Items measuring organisational support, manager support, career consequences and time demand were adapted from Thompson, Beauvais & Lyness (1999) and measured with six, five, three and four items respectively on a 7-point Likert scale (strongly disagree to strongly agree). Co-worker support was measured using modified manager support items (references to managers changed to coworkers ). Negatively-worded items were recoded - high scores reflected conditions supportive of program use. TPB elements Items measuring attitude, subjective norm and perceived behavioural control were based on standard wordings (Ajzen 2006). Attitude was measured using a single stem Utilising WLB programs available in my organisation is followed by five items of bipolar adjectives (e.g., Good-Bad ) on a 7-point Likert scale with the adjectives at each end. Three, four, and three items respectively measured subjective norm, perceived behavioural control, and intention, on a 7-point Likert scale (strongly disagree to strongly agree). Program utilisation One item measured program utilisation, In the course of the past 3 months, how often have you utilised the following WLB programs in your organisation. Twelve different options

8 ANZAM 2009 Page 8 of 24 offered by State government agencies were listed (e.g., flexi-time, parental/maternal leave). Following previous TPB research using past behaviour as a proxy to actual behaviour, (Johnson & Hall 2005), frequency of use was measured on a 5-point scale ( Never to More than 3 times per week ). Statistical Treatment Exploratory factor analyses To test independence and loading of items, factor analyses using maximum likelihood extraction method with promax rotation were conducted. As the measure of barriers and the measure of TPB were from different original scales, barriers items and TPB items were analysed separately. A factor loading cut-off criterion of 0.32 (Tabachinick & Fidell 1996) was used to determine the factor solution and items that did not load clearly on the relevant factor were excluded (due to space restrictions, item loadings are available from the authors). The factor analysis of barrier items revealed a 4 factor solution. Interpretation showed that three of the four factors were related to the barriers intended to be measured in this study, explaining 55.17% variance and labeled co-worker support (α = 0.97); time demand and career consequences (α = 0.85); and organisational and managerial support support (α = 0.88). The fourth factor (support for elderly care responsibilities) did not fit with the theoretical background and was excluded from further analysis. Factor analysis of TPB items revealed a 4-factor solution explaining a total of 60.46% of the variance. Factor 1 was interpreted to represent intention (α = 0.93), factor 2 attitude (α = 0.84), factor 3 perceived behavioural control (α = 0.81), and factor 4 subjective norm (α = 0.75). With the exception of program utilisation, composite scores for each measure were calculated by averaging scores across corresponding items. For program utilisation, responses for each WLB program was first converted to a binary score where 0 indicated no utilisation and 1 indicated utilisation, regardless of usage frequency. Binary scores were then summed to give a composite score for program utilisation. For example, a score of 8 indicated 8 of the 12 programs were utilised at least once in the past 3 months.

9 Page 9 of 24 ANZAM 2009 Model development Adhering to the theoretical base and measurement results, the proposed indirect model was revised to accommodate a shift from five theoretical barriers to three measured barriers (see Figure 2). Insert Figure 2 here To support the theoretical argument that barriers would not influence intention more than attitude, subjective norm and perceived behaviour control, three nested models were first developed and their fit indices compared: a baseline model testing barriers and TPB elements together, a barrier model testing barriers only, and an element model (see Figure 3) testing TPB elements only. If attitude, subjective norm and perceived behavioural control influenced intention more so than did barriers, the barrier model would be a poorer fit than the baseline whereas the TPB element model would be comparable in fit to the baseline. Insert Figure 3 here To directly test the fit of the indirect TPB model compared to the direct barrier model, a second baseline model was developed incorporating the TPB and linking barriers directly to utilization, then tested against a nested indirect model (Figure 4). Next, the baseline model was compared with a nested direct model (see Figure 5). If the indirect model were supported, the direct model would be of a poorer fit than the baseline model whereas the indirect model would be comparable to baseline. The indirect model would also fit significantly better than the direct model. Insert Figure 4 here

10 ANZAM 2009 Page 10 of 24 Insert Figure 5 here All models were tested with path analysis using AMOS 7.0. Chi-square values were used for model comparisons. Preferred threshold for fit indices was set at >0.95 for CFI, <.06 for RMR, and <.07 for RMSEA (Hu & Bentler 1999). To test specific relationships between barriers and TPB elements, regression weights were examined. Significance level was set at p<.05. RESULTS Due to space restrictions, means, standard deviations and correlations are not presented here but are available from the authors upon request. Whilst barriers were weakly correlated with intention (r =.17 to.20, p<.01), contrary to expectations, they were not significantly correlated with utilization (r =.02 to.09, p>.05). Model Testing Path analysis was initially used to test the TPB element model (see Figure 6). As expected, the results showed good model fit (χ2 = 1.11, df = 2, p >.05, RMR =.02, CFI = 1.00, RMSEA =.00 (.00,.11)). Regression weights indicated that subjective norm and perceived behavioural control positively influenced intention however attitude was not significantly related to intention. Intention but not perceived behavioural control was significant in predicting program utilisation. Insert Figure 6 here

11 Page 11 of 24 ANZAM 2009 Comparison of baseline, barrier and TPB element model fit (see Table 1), showed the baseline model was a better fit than the barrier model (χ2 difference = , df = 3, p<.001) but not significantly different to the element model (χ2 difference = 5.79, df = 3, p> 05). Insert Table 1 here Fit statistics for the baseline, direct and indirect models (Table 2) showed the baseline was a better fit than the direct model (χ2 difference = , df = 9, p<.001) but not significantly different to the indirect model (χ2 difference = 2.20, df = 3, p>.05). Direct paths between barriers and utilisation were nonsignificant (due to space restrictions, only the indirect model path diagram is presented below in figure 7), Chi-square comparison between indirect and direct models showed the indirect model to be a better fit (χ2 difference = , df = 6, p<.001) supporting an indirect effect of barriers on utilisation through the TPB. Insert Table 2 here Insert Figure 7 here In the indirect model, 5.9% of the variance in program utilisation was accounted for by intention and 44.1% of the variance in intention was accounted for by attitude, subjective norm and perceived behavioural control. Organisational and managerial support predicted attitude (β =.17, p<.05) and perceived behavioural control (β =.24, p<.01) as did time demand and career consequences (β =.42, p<.001, and β =.38, p<.001 respectively) but not subjective norm. Co-worker support predicted subjective norm (β =.30, p<.001) but failed to predict attitude and perceived behavioural control. Based on regression weights, the total indirect effect on utilisation of organisational and managerial support (β =.12) and of time demand and career consequences (β =.16) was traced through perceived behavioural

12 ANZAM 2009 Page 12 of 24 control and intention and the total indirect effect on utilisation of co-worker support was traced through subjective norm and intention (β =.29). DISCUSSION The aims of this study were to examine the indirect effects of organisational barriers on WLB programs utilisation through the TPB and to examine the relationships between specific organisational barriers and attitude, subjective norm and perceived behavioural control. The results showed that a TPB model generally fitted well with the data. An indirect model explained the relationship between barriers and program utilisation better than a direct model. That barriers were not directly related to intention and program utilisation supports Ajzen s (1991) view that behaviour is influenced by individual beliefs which are in turn affected by external factors. While contrary to Thompson et al. (1999) and Smith and Gardner (2007) who found organisational barriers to be directly related to utilization, in the present study, the effects of other variables were partialled out, representing a more stringent test of the relationship. However, the mediating role of the TPB between barriers and program utilisation was not confirmed. For the TPB to be confirmed as a mediator, barriers would need to be related to program utilisation (Baron & Kenny 1986). However despite reliable measures, attitude did not predict intention and perceived behavioural control did not predict program utilisation. The results suggest instead an influence of barriers on TPB elements and an independent influence of social pressure and perceived control on program utilization. When it comes to predicting program utilisation, this study shows that barriers do not influence intention and utilisation directly. Rather, barriers influence program use indirectly through the TPB. From a practical standpoint, HR practitioners can use these findings to focus attention directly on the perceived social pressure on employees to use WLB programs and on employees own perceived control. For example, reminding employees that WLB are an entitlement rather than an incentive may be an appropriate first step in increasing perceptions of control.

13 Page 13 of 24 ANZAM 2009 The results of this study may not generalise outside the public sector (particularly if organizations offer different programs), the measures in this study were all self-report, and a longitudinal design would provide more robust evidence of the causal nature of these relationships, however results do suggest that the relationship between previously considered barriers and WLB program usage is more complex than first thought. Other factors such as financial needs, or work commitments could be included in future studies to test their influence on a person s utilization of WLB programs. Future research should at least take into account this complexity in considering ways to improve utilisation of WLB programs.

14 ANZAM 2009 Page 14 of 24 REFERENCES Ajzen I (1991) The theory of planned behavior, Organizational Behaviour and Human Decision Processes 50: Ajzen I (2006) Constructing a TpB questionnaire: Conceptual and methodological considerations. Retrieved February 18, 2008 from Allen T (2001). Family-supportive work environments: The role of organisational perceptions, Journal of Vocational Behaviour, 58: Almer E & Kaplan S (2000) Myths and realities of flexible work arrangements The CPA Journal, 70: Armitage C & Conner M (2001) Efficacy of the theory of planned behaviour: A meta-analytic review, British Journal of Social Psychology, 40: Baron R & Kenny D (1986) The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51: Christian J, Armitage C & Abrams D (2007) Evidence that theory of planned behaviour variables mediate the effects of socio-demographic variables on homeless people s participation in service programmes. Journal of Health Psychology, 12: Cialdini R (2003) Crafting normative messages to protect the environment. Current Directions in Psychological Science, 12(4): De Cieri H, Holmes B, Abbott J, & Pettit T. (2005) Achievements and challenges for work/life balance strategies in Australian organizations. International Journal of Human Resource Management, 16: Hobsor C, Delunas L & Kesic D (2001) Compelling evidence of the need for corporate work/life balance initiatives: Results from a national survey of stressful life-events. Journal of Employment Counselling, 38:

15 Page 15 of 24 ANZAM 2009 Hu L & Bentler P (1999) Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria verses new alternatives, Structural Equation Modeling, 6: Johnson S & Hall A (2005) The prediction of safe lifting behavior: An application of the theory of planned behavior, Journal of Safety Research, 36: Kirby E & Krone K (2002) The policy exists but you can t really use it : communication and the structuration of work-family policies, Journal of Applied Communication Research, 30: Kossek E, Barber A & Winters D (1999) Using flexible schedules in the managerial world: The power of peers, Human Resource Management, 38(1): Managing Work Life Balance (2007). Work/life initiatives: The way ahead report on the year 2007 survey. Roseville, NSW: Managing Work/life Balance International. Nord W, Fox S, Phoenix A & Viano K (2002) Real-world reactions to work-life balance programs: lessons for effective implementation, Organizational Dynamics, 30: Perlow L (1995) Putting the work pack into work/family, Group and Organization Management, 20: Smith J & Gardner D (2007) Factors affecting employee use of work-life balance initiatives, New Zealand Journal of Psychology, 36: Tabachinick B & Fidell L (1996) Using multivariate statistics (3rd ed.). Harper Row: New York. Thomas L & Ganster D (1995) Impact of family-supportive work variables on work-family conflict and strain: a control perspective, Journal of Applied Psychology, 80: Thompson C, Beauvais L & Lyness K (1999) When work-family benefits are not enough: the influence of workfamily culture on benefit utilization, organisational attachment, and work-family conflict, Journal of Vocational Behaviour, 54: Ziemer K & Wood E (2007) Testing the importance of flexible working arrangements and barriers to their use. Paper presented at the 16th Annual World Business Congress, Maastricht, Netherlands, July.

16 ANZAM 2009 Page 16 of 24 Figure 1. TPB and proposed indirect model.

17 Page 17 of 24 ANZAM 2009 Figure 2. Proposed indirect model with three barriers

18 ANZAM 2009 Page 18 of 24 Figure 3. Paths for the baseline (solid line), barrier (dotted line) and TPB element (hatched line) models.

19 Page 19 of 24 ANZAM 2009 Figure 4. Baseline model (dotted and solid lines) and indirect model (solid lines only).

20 ANZAM 2009 Page 20 of 24 Figure 5. Direct model with barriers and TPB as direct predictors of utilisation.

21 Page 21 of 24 ANZAM 2009 Figure 6. Path model of Theory of Planned Behaviour.

22 ANZAM 2009 Page 22 of 24 Figure 7. Path model of indirect model (variances expressed as percentages).

23 Page 23 of 24 ANZAM 2009 Table 1. Fit Statistics for the Baseline, Barrier and Element Models χ 2 df CFI RMR RMSEA Null Baseline (.00,.08) Barrier (.24,.31) Element (.00,.09)

24 ANZAM 2009 Page 24 of 24 Table 2. Fit Statistics for Baseline, Indirect and Direct Models. χ 2 df CFI RMR RMSEA Null Baseline (.13,.21) Indirect (.11,.18) Direct (.25,.30)