A pre- and post-intervention study was conducted. There was no control group because it was difficult to set due to the condition of the participant organizations. Therefore, we put this study as a pilot study to investigate the effectiveness of our job crafting intervention preliminarily. The study protocol was registered retrospectively at the UMIN Clinical Trials Registry (UMIN-CTR) (ID = UMIN000024062). The ethics review board of The University of Tokyo Graduate School of Medicine/Faculty of Medicine approved the procedures before the start of the study (the reference number: 10749). This manuscript was reported according to the TREND statement checklist .
This study was carried out from August to November 2015 at two private companies in Japan. We recruited all full-time managers in a manufacturing company (n = 54) and managers who belonged to seven selected departments at a private psychiatric hospital (n = 25). Participants were approached by a contact person in their own company or hospital using an e-mail invitation or a poster. The inclusion criterion for participants was regular (full time) employment; workers with non- regular (part time) employment and reemployment in a temporary position were excluded.
The intervention program, which was based on the job crafting theory of Wrzesniewski and Dutton (2001) , was developed through literature review [20–25], discussion with occupational health professionals, and interview with employees on how they craft their own job in their working lives, such as changing the relationships with others or reframing the significance of their work. One of the unique features of the program, compared to previous job crafting interventions [26–28], was that it contains three aspects of job crafting (i.e., task, human relation, and cognition), which may be important to work engagement. The program consisted of two 120-min sessions with a two-week interval between them and was conducted by one researcher (the first author) and one clinical psychologist. The two training sessions were held in a room at the worksite in a group setting with 9–13 participants during working hours in the company, separately for department 1 (first session n = 11, second session n = 9), department 2 (n = 12 and 13, respectively), and department 3 (n = 12 and 12, respectively). The sessions were conducted outside of working hours in the hospital (n = 13 and 10, respectively). In the first session, following the introduction of the idea of job crafting including task, relational or cognitive crafting, by the researcher, participants learned the concept of job crafting from a case study (30 min), shared their personal crafting stories in their own working lives (30 min), and made their own individual job crafting plans (task crafting, relational crafting, and cognitive crafting) for the next two weeks (30 min). They were also provided with a homework booklet for a job crafting exercise. During the two weeks between the first and second sessions, they were encouraged to implement their job crafting plans. In the second session, each participant reviewed his/her own job crafting plan individually (15 min); the participants then shared their reflections as a group (30 min), discussed what job crafting would be feasible and sustainable to practice (30 min), and, finally, made a modified job crafting plan (30 min). Participants who could not attend first session were distributed the material of the session and asked to make their job crafting plan and practice it before they attend the second session. Thus, they could attend the second session. There was no incentive offered for participation in the program.
All data were collected using a web-based self-report questionnaire at baseline, post-intervention, and one-month follow-up.
Work engagement was assessed using the Japanese version of the Utrecht Work Engagement Scale (UWES), which has been reported to be reliable and valid . It comprises 9 items assessing vigor (3 items; e.g., ‘At my work, I feel bursting with energy.’), dedication (3 items; e.g., ‘My job inspires me.’), and absorption (3 items; e.g., ‘I get carried away when I am working.’). All items were rated on a seven-point Likert scale ranging from 0 (never) to 6 (always), and the total score for each subscale was divided by the number of items to get an average score.
Psychological distress was measured using the Brief Job Stress Questionnaire (BJSQ) , comprising 15 items assessing irritation (3 items; e.g., ‘I feel anger’), fatigue (3 items; e.g., ‘I feel very tired’), anxiety (3 items; e.g., ’I feel uneasy’), and depression (6 items; e.g., ‘I feel depressed’). All items were measured on a four-point Likert scale ranging from 1 (never) to 4 (almost always), and the total score was calculated by dividing the sum of item scores by the number of items in the present study.
Manipulation check of the intervention
Job crafting was assessed using a scale developed by Sekiguchi and his colleagues  based on the conceptualization by Wrzesniewski and Dutton (2001) , which has been reported to be reliable and valid . It comprises 12 items assessing three subscales: task crafting (4 items; e.g., ‘Add or reduce tasks so that my job can be performed more smoothly’), relational crafting (4 items; e.g., ‘Actively interact with people through my job’), and cognitive crafting (4 items; e.g., ‘Reframe my job as significant and meaningful’). All items were measured on a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree), and the total score, as well as each subscale score, was calculated by dividing the sum of item scores by the number of the items.
Demographic variables included age, gender (0 = male, 1 = female), marital status (0 = married, 1 = not married), educational attainment (0 = university or higher, 1 = high school or some college), and working hours per week were collected. These variables could be possibly confounded in the association between work-related variables and employee well-being [1, 21, 23, 32, 33]. A close investigation of the responses to work hours revealed that some respondents reported less than 35 h per week, while all respondents were employed full-time. These responses were coded as missing values.
The estimated sample size was 66 participants to detect an effect size (Cohen’s d) of 0.35 or greater for work engagement, at an alpha error rate of 0.05 (two-tailed) and a beta error rate of 0.20 using the G*Power 3 program [34, 35]. No previous studies have reported an effect size of the job crafting intervention program for work engagement. However, referring to a previous study  that showed the effect size (Cohen’s d) for self-efficacy and job-related affective well-being, it seemed reasonable to set 0.35 as an expected effect size in our job crafting intervention program.
The analysis was performed at the individual level. A mixed model for repeated measures conditional growth model analysis over time (baseline, post-intervention, and one-month follow-up) was conducted to test the effect of the intervention. The mixed model is useful to understand changes in human behavior over time because of higher-level clustering unit (i.e., time) [36, 37]. First, we applied several mixed models to the data: random intercept and random slope; random intercept only; and random slope only. A converged model that showed the smallest AIC (Akaike Information Criterion), an indicator of goodness of fit of the model, was selected. If these mixed models did not converge, a fixed model was used. The linear mixed model in SPSS Statistics 22.0 (SPSS Inc, Chicago, IL) was used. Also, the effect sizes and the 95 % confidence intervals (95 % CIs) were calculated using Cohen’s d only among those who completed the questionnaire at post-intervention or follow-up, although the effect sizes may be biased due to drop-outs [38, 39]. Values of 0.2, 0.5, and 0.8 are generally interpreted as being suggestive of small, medium, and large effects, respectively .