Surviving repeated waves of organizational downsizing: The recency, duration, and order effects associated with different forms of layoff contact

Abstract

In this paper we examine: (1) recency and duration effects of layoff contact; and (2) the order effects associated with different types of layoff contact experiences. Workers employed by a large company engaged in repeated waves of downsizing completed questionnaires in 1997, 1999, and 2003. Using data only from workers who experienced indirect or direct layoff contact at each time period (N=460), we found some evidence that recent direct experiences were associated with significant group differences for intent to quit and depression. There was also some evidence to suggest that a single direct layoff experience still affected workers’ levels of job security, even when this experience occurred some 6 years prior to the Time 3 measurement. The largest within-group changes in scores over time were typically found among workers experiencing an indirect experience followed by a direct experience, suggesting that the order of events impacted worker job security, intent to quit, and depression. For workers experiencing back-to-back direct downsizing, the rate of decline slowed for depression. These findings are examined in light of the stress vulnerability and resiliency hypotheses.

Keywords:

• Downsizing
• psychological health
• resiliency survivors
• stress

This research was supported by Grant no. AA10690-02 from the National Institute of Alcohol Abuse and Alcoholism of the National Institutes of Health.

Notes

1. To examine the degree of similarity between the original 3,700 targeted employees and the 776 who participated in all three waves of the survey, we compared gender and paycode data between the two. Other demographic data were not available to us from company records. This analysis revealed that our sample was roughly similar to the 3,700 in terms of gender (22% for the company, 20.8% for the sample). In terms of paycode, however, we had a lower rate of participation among blue-collar workers. Company records data on the 3,700 employees (sample percents in parentheses) revealed that 50% were hourly workers (31.3%), 17% were salaried-exempt (20.9%), 10% were salaried-nonexempt (16.6%), 10% were managerial (13.6%), and 13% were engineers (17.6%). Aside from leaving the company, we cannot be certain as to the reasons for nonparticipation. We assume that these reasons are varied, ranging from lack of motivation to little time to complete a lengthy questionnaire. We also note, however, that cuts were roughly targeted at all employee paycodes; thus, unequal survival rates across paycodes is not a likely reason for the differences.

2. Using data from all participants who completed the Wave 3 survey, principal components analysis on the short form of the CES-D revealed a strong single factor solution. One factor explained over 64% of the total variance, and there was a sharp decrease in the Eigenvalues between the first (4.50) and second (.667) factor; communality estimates ranged from .49 to .74. Mirowsky and Ross (1989) report a correlation of .92 between this short and the full version of the CES-D (p. 187).

3. Because only three groups logically existed at Time 2, we report the repeated measures results for the II, ID, and DI groups between Times 1 and 2.

4. As seen in Table IV, T2 versus T3 and T1 versus T3 comparisons for all groups other than the III group had sample sizes ranging from 22 to 50. Examination of statistical power tables (Cohen, 1988; Statistical power of the t test for one sample or two related samples, n.d.) revealed that for an alpha level of .01, to achieve a power level of .80, for a medium effect size (d=.50) required a sample size of 50. A sample size of 20 required a large effect size (slightly greater than d = .80) to achieve statistical power of .80. A sample size of 30 required d =.66 to reach a power of .80. Thus, for many of our comparisons, it was likely that only large effect sizes would be reliably detected.

5. Using a more conventional type of analysis, we did conduct several 2 (I vs. D at T1) × 2 (I vs. D at T2) × 2 (I vs. D at T3) MANCOVAs in an effort to isolate the impacts of layoff contact at each time period. We found a significant main effect for Time 3 contact on intent to quit, matching the results shown in Table III. For Time 3 depression, we obtained a significant 2 (I vs. D at T1) × 2 (I vs. D at T2) interaction that revealed significantly lower depression scores for those in the DD condition. Last, for Time 3 job security, we obtained a 2 (I vs. D at T1) × 2 (I vs. D at T2) interaction that approached significance (p=.06), suggesting that those with “mixed” experiences at Times 1 and 2 (I then D or D then I) fared worse at Time 3 than those with either II or DD histories. Because of the very small cell sizes, however, we determined that this analytic strategy was untenable (i.e., for DDD N =3, and for DDI N=9).

6. Because workers without repeated downsizing contact were removed from the analyses, we wish to acknowledge that the between-group differences reported in Table III may fail to reflect the substantial impact downsizing experiences have on these outcome measures. Independent samples t tests that compared workers with repeated contact at each time period (i.e., the sample studied in this paper, N=460) with the rest of the workers in the panel study (N=306) found significant group differences on job security (Times 1, 2, and 3), intent to quit (Times 2 and 3), and depression (Times 2 and 3). In all cases, participants with repeated contact at each time period were significantly less secure and had greater reported levels of intent to quit and depression.