2.1. Participants
The data come from a quantitative study using a survey of workers in small and medium-sized enterprises (SMEs—that is, enterprises with fewer than 249 workers) in Spain, carried out during the months of June and December 2016. In January of that year, there was a total of 3,228,747 SMEs in Spain, of which 55.45% were companies with no paid employees, which were consequently excluded from the study. The number of workers in the remaining SMEs ranged between one and nine for 40.37% of the enterprises, between 10 and 50 for 3.58%, and between 50 and 249 workers for 0.60% [
52]. The sample was made up of 1599 workers from 154 SMEs, and the average number of employees per company was 10. Regarding the age of the sample, 7.5% of these workers were younger than 25, 33.2% were between 25 and 35 years, 34.6% were between 35 and 45, 17.5% were between 45 and 55, and 7.1% were older than 55 years old. Regarding gender, 49% of the participants were men. As to educational level, 9.2% of the subjects had basic studies, 38.2% had a high school or vocational training degree, and 52.6% had completed higher studies. The professional status was divided into 8.1% unskilled workers, 75% in an administrative or technical post, 13.1% middle managers, and 3.9% managers. Concerning tenure, 15.2% of the sample had been working in the organization for less than one year, 49% had been working for over one year but less than 10 years, 22.7% had been working between 10 and 20 years, 9.5% between 20 and 30 years, and 3.6% for over 30 years. Finally, 83% of the sample worked full time, 15.5% part time, and about 1.5% had a different job situation.
2.4. Instruments
Job demands were evaluated by the job demands subscale of the job content questionnaire (JCQ) [
53], validated in its Spanish version by Escribà-Agüir, Más Pons, and Flores Reus [
54]. The items assess the amount of work that must be done in that job, the speed required when performing it, the complexity of the tasks, and the intensity with which the work must be performed. Examples of items are: “My job requires me to work very quickly”, “My job requires me to work hard”, “I am asked to do an excessive amount of work”, or “My work requires long periods of intense concentration”. The study of the adaptation to Spanish yielded a reliability of 0.74 for this version. Cronbach’s alpha in this study was 0.75.
Job resources were evaluated using the control and support subscales of the JCQ, in the Spanish version validated by Escribà-Agüir and collaborators [
53]. Job characteristics (the possibility of controlling the work, the use of competencies, the development of competences in the post, and the professional and emotional support received) were evaluated with 15 items. Examples of these items are: “My job often allows me to make my own decisions”, “My job requires a high level of competence”, “In my work, I perform varied activities”, or “I have the opportunity to develop my professional skills”. The original validation study found adequate reliability, ranging between 0.75 and 0.84. Cronbach’s alpha in this study was 0.85.
Personal resources refer to organizational identification, professional identification, and group identification. Following the procedure used in other studies [
16,
55], identification with each focus was measured with five items from Mael and Ashforth’s identification scale [
31]. We replaced the referent of the items to represent the respective foci (i.e., the team, profession, or organization). Sample items are “The successes of my organization/team/profession are my successes” and “When I talk about my organization/team/profession, I usually say ‘we’ rather than ‘they’”. Cronbach’s alpha values in the present study for each subscale were: 0.78 for organizational identification, 0.73 for professional identification, and 0.77 for group identification.
Organizational culture. We used the Spanish version [
56] of the FOCUS 93 questionnaire [
51], which assesses the frequency of certain features of one’s own organization. The underlying cultural model of this questionnaire was Quinn and Rohrbaugh’s [
50] competing values model. Based on two bidimensional axes (internal vs. external orientation and flexibility vs. control), the survey provides four organizational culture orientations: support, innovation, rules, and goals. Key concepts for support orientation are participation, cooperation, mutual trust, and team spirit. The innovation is characterized by the search for new information, creativity, openness to change, and anticipation. The rules emphasize respect for authority, rationality of the procedures, and the division of work. Goal orientation emphasizes performance indicators, accomplishment, and accountability [
51]. Support orientation includes eight items, innovation 12 items, rules six items, and goals 14 items. Examples of items are “How many people with personal problems are helped?”, “How often is constructive criticism accepted?” (support), “How often does your organization search for new markets for existing products?” (innovation), “How often are instructions written down?” (rules), and “How often is it clear how performance will be evaluated?” (goals). Reliability values in the original study were 0.91 for support, 0.69 for innovation, 0.77 for rules, and 0.76 for goals. In this study, Cronbach’s alpha values for each subscale are: 0.84 for support, 0.86 for innovation, 0.87 for rules, and 0.66 for goals.
Job Satisfaction. The brief index of affective job satisfaction (BIAJS, [
6]) was used. This brief scale is a measure of affective job satisfaction with just one factor composed of four items: “I find real enjoyment in my job”, “I like my job better than the average person”, “Most days I am enthusiastic about my job”, and “I feel fairly well satisfied with my job”. Responses are rated on a five-point Likert scale ranging from 1(strongly disagree) to 5 (strongly agree). Moreover, in order to reduce priming effects and acquiescent responses, the scale includes three distracter items: “My job is unusual” (between items 1 and 2), “My job needs me to be fit” (between items 2 and 3), and “My job is time-consuming (between items 3 and 4). In this study, the Cronbach’s alpha of this scale is 0.88.
Organizational Citizenship Behavior. We used a Spanish adaptation [
57] of the organizational citizenship behavior scale designed by Lee and Allen [
58]. This 16-item instrument has the advantage of evaluating both OCBs directed at the organization (OCB
O) and OCBs directed at individuals (OCB
I), unlike other scales that only focus on the perspective of the institution. Therefore, it provides two separate, but correlated measures of the behaviors aimed at benefiting peers and companies. Subjects report how often they perform the actions described in the items at work on a five-point Likert-type response format ranging from 1 (infrequently) to 5 (frequently). Example items are: “I show an interest in the organization’s image” and “I dedicate time to helping others who have problems related or not to the tasks”. The reliability analysis showed satisfactory internal consistency for the scale in previous studies [
59]. In our sample, the Cronbach’s alpha value of the OCB
O subscale is 0.90 and of the OCB
I subscale it is 0.87.
Health. Self-rated health status was assessed with four items, following the procedure of the survey of health, aging, and retirement in Europe (SHARE) [
60]. Questions from the self-completion questionnaire (nr. 4, options a, d, m, and l) were included, two of which assess depression and exhaustion and the other two happiness and high energy. The first two items were reversed so as to indicate a global measure of good health. The Cronbach’s alpha of this scale was 0.70 in this study.
All the scales were responded on a Likert-type response scale ranging from 1 (totally disagree/never/infrequently) to 5 (totally agree/always/frequently).
Along with the questions relating to these scales, information was gathered on the following sociodemographic characteristics, which are the control variables in our model: age (0: younger than 25 years; 1: between 25 and 35; 2: between 35 and 45; 3: between 45 and 55; 5: over 55 years), gender (0: men; 1: women), education (0: basic studies; 1: high school degree or vocational training; 2: higher education), professional category (0: unskilled worker; 1: administrative staff or technician; 2: middle management; 3: manager), tenure (0: less than one year; 1: between 1 and 10 years; 2: between 10 and 20 years; 3: between 20 and 30 years; 4: over 30 years), and work situation (1: full-time and 2: part-time).
2.5. Data Analysis
Firstly, the participants’ scores in each of the scales were calculated. For this purpose, we applied the generalized partial credit model (GPCM) [
61], developed under the assumptions of the item response theory [
62] for situations with ordinal items. The scores were then transformed to a scale with a mean of 250 and standard deviation of 25.
In the second phase, we estimated the multivariate multilevel model [
63], in which all four response indicators were entered at once: (1) job satisfaction, (2) OCB
O, (3) OCB
I, and (4) health. We specified a two-level model, in which the employees (level 1) are grouped into organizations (level 2).
The Mplus version 8 (Muthén & Muthén, Los Angeles, CA, USA) [
64] program was used to estimate the models and, to facilitate the interpretation of results, the quantitative predictors on workers’ level were group-mean centered, and the predictors on organization’s level were grand-mean centered. Regarding the sociodemographic characteristics, “male,” in the case of gender, and the initial categories of the ordinal variables (age, level of studies, professional category, and tenure) were taken as reference categories. Finally, to identify the workers’ job situation, we created two dummy variables (full-time and part-time).
Data Aggregation
Given that the dimensions of organizational culture were measured at the worker level, we estimated the mean value of the organization in these indicators in order to analyze its relationship with the wellbeing perceived by employees of different organizations. Analysis of the data’s adequacy to be aggregated was carried out by evaluating the inter-rater agreement (IRA) and intraclass correlation. Thus, we calculated the within-group interrater reliability RWG (j) index for multiple item scales [
65], the average deviation (AD) index [
66], and the intra-class correlation coefficients (ICC
(1) and ICC
(K)).
The results show that, except for two organizations (RWG
(36) = 0.49 and RWG
(148) = −0.05), the RWG (j), values for the predictor support were higher than 0.70, and higher than 0.80 for 96% of the SMEs. Similar results were observed for the other predictors: innovation (all the values exceeded 0.79), rules (approximately 4.5% of the organizations obtained values between 0.7 and 0.8, obtaining higher values in the remaining cases), and goals (all the organizations presented values higher than 0.72, and in 98% of the companies, higher than 0.83). The above values revealed a strong measure of agreement among the workers [
67].
A similar trend was observed when analyzing the AD indices. In the case of predictor support, 91.56% of the organizations did not exceed the critical value suggested by Burke and Dunlap [
68] for five-point Likert-type response scales, that is, 0.83 (c/6 = 5/6). For the predictor’s innovation, rules, and goals, these values were, respectively, 93.51%, 94.81%, and 90.26%. We note that only three organizations in the support dimension (AD
(100) = 1.03, AD
(36) = 1.11, and AD
(148) = 1.19) and one in the goal dimension (AD
(36) = 1.03) presented AD indices higher than 1 [
66].
According to the intra-class correlation coefficients, the ICC(1) values were higher than 0.37 in all cases for the predictors support, innovation, and goals, and higher than 0.36 for the dimension rules, so they could all be considered as appropriate (higher than 0.25). The ICC(K) values were higher than 0.83 in all cases.
Considering the calculated indices and coefficients, the data’s adequacy to be aggregated was justified. This aggregation was performed for all the organizations and indicators analyzed, apart from two companies in which unsatisfactorily low levels of agreement were observed, both in the Rwg index and the AD index. These two organizations were omitted from the analysis, so the final sample was composed of 1579 workers from 152 SMEs.