The supportive supervisory scale: psychometric properties in Chinese health care aides samples

The SSS has 15 items and includes two parts. The first part is labeled “Respect Uniqueness” and the second part “Being Reliable” [9]. Answer options are “never”, “seldom”, “occasionally”, “often”, “always”, which successively represent the score of “1, 2, 3, 4, 5”.

The SSS was translated from English into Chinese using Brislin’s translation model [10]. The steps for sinicization of SSS are shown in Fig. 1. Firstly, two bilingual researchers separately translated the original SSS into Chinese. The discrepancies between these two translations were reviewed and discussed comprehensively, and formed a single version, which was then translated back into English by another bilingual researcher. The retroversion was repeatedly compared with the original SSS scale and the Chinese expressions were adjusted accordingly. Throughout the Chinese version, “supervisor” was replaced by “nurse supervisor” to make the items more understandable since HCAs report directly to registered nurses in LTCFs in China. During this procedure, the translation validity index (TVI) was used to assess the translation equivalence of versions. It used a 4-point Likert scale (1 = uncorrected, 2 = needs major modification on equivalent item, 3 = equivalent but needs minor modification, and 4 = equivalent). In this study, three language experts were recruited to compare the SSS in English and Chinese. The items were revised until a TVI score of 4 was achieved. The revised version of SSS was pilot tested with a convenience sample of 30 HCAs in a LTCF in Suzhou to evaluate whether the Chinese version of SSS was easy to understand. Language expression was adjusted if HCAs felt it was difficult to understand. After the pilot test, the Chinese version of SSS (SSS-C) was finalized for the test of its psychometric properties.

The study was conducted in 4 LTCFs in Suzhou, China. Health care aides meeting the following criteria were enrolled in the study: (i) working in the LTCFs for more than 3 months; (ii) able to give written consent. A total of 300 participants completed the scale. 41.4% of the HCAs were less than 50 years old and 54% were 51–60 years old; 85.3% were female; 100% of the respondents were employed full-time. The mean number of years that the respondents have worked in the LTCF was less than 5 years (n = 241, 80.3%).

After giving written consent, HCAs were asked to fill out the survey questionnaire independently and anonymously in the nursing station or staff lounge, without their supervisor being present. It was guaranteed that the supervisors had no access to the responses. The researchers remained in the room and were available to answer questions when necessary. Neither compensation nor remuneration was offered to the participants.

Analyses were performed by IBM SPSS Statistics 21. Participant characteristics and major variables were summarized by descriptive statistics (Table 1). Reliability was tested by internal consistency (Cronbach’s alpha), split-half reliability (Spearman-Brown coefficient) and test–retest reliability (Pearson correlation coefficient) (reliability coefficient ≥ 0.7 was acceptable) [11]. Construct validity was examined by exploratory factor analysis (EFA) (principal component analysis with varimax rotation). Scree plot, Kaiser criterion (eigenvalue ≥ 1.0), and clinical interpretability were considered in determination of factor solution (When the factor loading ≥ 0.40, the item can be considered in the factor) [12]. Discriminative validity was assessed by examining if supervisory support varied between the different facilities by one-way analysis of variance. Multiple comparisons, using Bonferroni’s procedure, were completed to compare every pair of facilities. The significant level was 0.05[13].

The 15-item SSS-C had satisfactory internal consistency (Cronbach’s α coefficient = 0.852), split half reliability (Spearman-Brown coefficient = 0.834) and test–retest reliability (Pearson correlation coefficient = 0.784) (see Table 2). Corrected item–total correlation and Cronbach’s α coefficient if item deleted for SSS-C are demonstrated in Table 3. The item-to-item correlations were positive, in the 0.083–0.541 range (Table 4).

Construct validity was examined by EFA, and the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy for this analysis was satisfactory (= 0.887), indicating that the sample size for the EFA was adequate. The result of the Bartlett’s test of sphericity was significant (χ² = 1125.922, P < 0.001), demonstrating a sufficient correlation to perform factor analysis. Using the Kaiser criterion, three factors were initially identified, and total variance explained was 48.128%. However, since the communality of item1(My supervisor recognizes my ability to deliver quality care), 4(My supervisor tries to understand my point of view when I speak to them), 5(My supervisor tries to meet my needs in such ways as informing me of what is expected of me when working with my residents) and 7(My supervisor keeps me informed of any major changes in the work environment or organization) was less than 0.4, they should be deleted in theory [14]. After deletion of these four items and a second exploratory factor analysis, three factors could still be extracted, accounting for 55.654% of the total variance (KMO = 0.843, χ² = 806.668, P < 0.001) (Table 5).

Mean item scores of the 15-item SSS ranged from 3.79 to 4.26 (SD = 0.715 to 1.033; Table 3). The four items under Factor I represented “the supervisors’ need to build connections with staff that involved respecting them as individuals” [9], therefore, this factor was labeled “building connections with staff”. Factor II including three items represented “the supervisors were available to staff to listen and respond to their concerns, and that they kept staff informed of what was new on the unit” [9], therefore, this factor was labeled “being dependable”. Factor III was labeled “being empathic” because the items loaded on this factor represented “the supervisors try to understand their point of view, recognize and accommodate expressed needs, recognize their abilities, and help them develop” [9]. Coefficient alpha reliabilities on the items comprising the three factors were: Factor I = 0.723, Factor II = 0.588 and Factor III = 0.663. The coefficient alpha for the 11-item SSS was 0.816.

The discriminant validity of the SSS-C varied significantly between sites. For example, within Facility D, scores of the SSS were significantly higher than those for Facility C (F = 4.791, p < 0.005; see Table 6).