Physical Demands, Job Resources, and Longer Work

Job resources were assessed with four scales: decision authority, social support, work-time control, and reward. Decision authority and social support were derived from the reliable and well-tested Demand–Control–Support-Questionnaire (DCSQ) (Chungkham et al., 2013 (link); Theorell et al., 1988 (link)), with response alternative on a 4-degree Likert scale ranging from 1: No, (almost) never to 4: Yes, often. Decision authority was measured with a 2-item-scale (what to do at work and how to do the work). The Scale reliability coefficient was 0.76. Social support was assessed with a 5-item-scale regarding pleasant atmosphere, understanding and cohesiveness among co-workers and managers. Cronbach's alpha was 0.85. Work-time control was measured with a 6-item-scale (Ala-Mursula et al., 2002 (link)), assessing the opportunities to influence the working time (length of work day, start and end times, taking breaks, running private errands during worktime, which days to work, and vacations) with five response alternatives ranging from 1: no, to a very small extent to 5: yes, to a large extent. Cronbach's alpha was 0.82. Rewards were measured by a 7-item-scale (Li et al., 2019 (link)) concerning job promotions (adequate salary, work and promotion prospects), esteem (receiving the earned acknowledgement, prestige and respect), and job security (including not expecting or experiences undesirable job changes) with response alternative on a 4-degree Likert scale ranging from 1: no, not at all to 4: yes, completely agree. Cronbach's alpha was 0.64.
A sum-index was estimated for each job resource. Thereafter, they were dichotomized into binary indicators according to their respective median values: decision authority: 0 “low” (37.5%; score range 2–5) or 1 “high” (62.5%; score range 6–8), social support: 0 “low” (51.5%; score range 5–15) or 1 “high” (48.5%; score range 16–20), work-time control: 0 “low” (52.5%; score range 6–13) or 1 “high” (47.5%; score range 14–30), and rewards: 0 “low” (44.7%; score range 7–17) or 1 “high” (55.3%; score range 18–28) (Table 1). Job resources were used as moderator variables in the association between physical demands or physical hazards and working longer.Table 1

Descriptive statistics and bivariate correlations with regard to the observations.

Table 1

	1	2	3	4	5	Nobs	Nlowest (%)	Cut-off value (range)
1 Physical demands	–					2729	53.39	10 (3–18)
2 Physical hazards	.279***	–				2696	53.75	2 (0–6)
3 Decision authority	−.035†	−.073***	–			2744	37.46	6 (2–8)
4 Social support	−.055**	−.147***	.144***	–		2669	51.48	16 (5–20)
5 WTC	−.028	.023	.244***	.100***	–	2593	52.45	14 (6–30)
6 Rewards	−.067**	−.101***	.181***	.306***	.157***	2262	44.69	18 (7–28)

Note. All variables are dichotomized. N_obs for number of observations (transitions). N_lowest for the percent of observations in the category below the cut-off value. *** for p < .001; ** for p < .01; * for p < .05; † for 0.05 ≤ p < .10.

Age category (59–63 years (0) versus ≥64 years (1) at baseline). This cut-off was based on the fact that the older age category included observations where the individual approached 65 years (representing the Swedish normative retirement age) and consequently two years later (at follow-up) had passed the normative retirement age.