I used five-point Likert-type scales ranging from 1, “strongly disagree,” to 5, “strongly agree” for all substantive variables. Since all measures were originally constructed in English, I employed translation and back-translation procedures (Brislin, 1986) to ensure that the English and traditional Chinese versions of all items were comparable at a high degree of accuracy. All items are shown in the Appendix A-F.
Furthermore, I counterbalanced the measure order of direct reports’ questionnaires to reduce potential biases caused by item priming effects. According to Podsakoff, MacKenzie, Lee, and Podsakoff (2003), an item priming effect refers to the fact that “asking questions about particular features of the work environment may make other work aspects more salient to respondents than those work aspects would have been if the questions had not been asked
in the first place.” For instance, if I ask direct reports to rate their supervisors’ authentic leadership behaviors first, the direct reports’ subsequent response regarding PS value congruence is primed by the previous ratings of authentic leadership. Hence, I counterbalanced the measure order as follows: the mediating variable (i.e., PS value congruence) items were asked first, followed by those regarding regulatory focus and
authentic leadership. In so doing, direct reports’ ratings of authentic leadership are unlikely to have impacted their ratings of PS value congruence.
Moral potency. Supervisors responded to an 12-item measure of moral potency
developed by Hannah and Avolio (2010). Three items were used to measure moral ownership (e.g., I will… “assume responsibility and take action when I see an unethical act”). Four items were used to measure moral courage (e.g., I will…“confront my peers if they commit an ethical act”). Five items were used to measure moral efficacy (e.g., I am confident that I can… “determine what needs to be done when I face ethical dilemmas”).
I conducted a second-order confirmatory factor analysis (CFA) to examine the expected higher order moral potency structure. The CFA results demonstrated that the
second-order model provided a good fit to the data (χ2 = 68.93, df = 51, RMSEA = .06, SRMR
= .07, GFI = .88, NFI = .88). Based on the second-order CFA results, it should be appropriate to aggregate the scores of the three facets into a total moral potency score. The Cronbach’s alpha coefficient for this scale was .84.
Authentic leadership. Direct reports rated their immediate supervisors’ level of authentic leadership behaviors using the 16-item the Authentic Leadership Questionnaire (ALQ; Walumbwa et al., 2008). The ALQ assesses four dimensions of authentic leadership and several studies have indicated a high-order factor for this scale (e.g., Hannah, Walumbwa, et al., 2011; Walumbwa et al., 2010). Three items were used to measure balanced processing (e.g., “My supervisor solicits views that challenge his or her deeply held positions”). Four
items were respectively used to measure self-awareness (e.g., “My supervisor is eager to receive feedback to improve interactions with me”) and internalized moral perspective (e.g.,
“My supervisor makes decisions based on his/her core beliefs”). Five items were used to measure relational transparency (e.g., “My supervisor is willing to admit mistakes when they are made”).
Because the ALQ measure is relatively new (Hannah, Walumbwa, et al., 2011;
Walumbwa et al., 2010), I conducted a second-order CFA to determine whether these facets were nested a second-order factor. Results revealed that the second-order model fit the data well (χ2 = 394.40, df = 100, RMSEA = .09, SRMR = .06, GFI = .88, NFI = .95). These results provided support for aggregating the four facets into a total score of authentic leadership in my study. The Cronbach’s alpha coefficient for this scale was .93.
Further, I examined whether authentic leadership could be conceptualized and
aggregated into the unit level. To justify the appropriateness of data aggregation, I calculated the inter-rater agreement (rwg) and the intraclass correlations (ICCs) for unit-level authentic leadership (Bliese, 2000; James, Demaree, & Wolf, 1984). The results showed that the average of rwg values was .92, with individual rwg ranging from .66 to .98, suggesting a high degree of inter-rater agreement for authentic leadership within the work units. Moreover, we calculated the ICCs: ICC(1) equaled .39 and ICC(2) equaled .74, whereas the F value for ANOVA was significant in terms of between-unit variances for authentic leadership (F[75, 285] = 3.93, p < .001). These results indicate that it was appropriate to conceptualize authentic leadership as a unit-level variable.
Leadership strength. Consistent with Chan’s (1998) dispersion model and a previous
example of leadership strength measure (Hannah, Walumbwa, et al., 2011), leadership strength was calculated using the coefficient of variation (Allison, 1978), which corrects for the lack of independence between measures of central tendency and measures of dispersion. I
first divided the standard deviation for each unit’s authentic leadership measure by the unit mean. This value then standardized and its sign was reversed so that higher values represented higher levels of leadership strength.
Person-supervisor value congruence. Consistent with my theorizing of PS value congruence as an individual perception, I adapted Cable and DeRue’s (2002) three-item subjective fit measure to assess direct reports’ subjective perceptions of PS value congruence.
Specifically, I substituted the word “supervisor” for “organization” in the original items.
Sample items are “My personal values match my supervisor’s values and ideals” and “The things that I value are similar to the things my supervisor values.” The Cronbach’s alpha coefficient for this scale was .86.
Regulatory focus. I used the 18-item measure (nine items per subscale) developed by Lockwood, Jordan, and Kunda (2002) to assess individuals’ chronic regulatory focus strategy.
Direct reports were asked to indicate the extent to which these statements are true for them in their daily lives. Sample items for promotion focus are “I frequently imagine how I will achieve my hopes and aspirations” and “I often think about the person I would ideally like to be in the future.” Sample items for prevention focus are “In general, I am focused on
preventing negative events in my life” and “I frequently think about how I can prevent failure in my life.” The Cronbach’s alpha coefficients were .77 and .80 for promotion focus and prevention focus, respectively.
Employee voice. I used the ten-item scale (five items per subscale) developed and validated by Liang et al. (2012) to measure supervisor ratings of their direct reports’
promotive and prohibitive voice. The items were preceded with the instruction, “Please refer to your employee, ____, when answering the following items.” Sample items for promotive voice are “Proactively develop and make suggestions for issues that may influence the unit”
and “Make constructive suggestions to improve the unit’s operation.” Sample items for
prohibitive voice are “Advise other colleagues against undesirable behaviors that would hamper job performance” and “Dare to point out problems when they appear in the unit, even if that would hamper relationships with other colleagues.” The Cronbach’s alpha coefficients were .87 and .89 for promotive voice and prohibitive voice, respectively.
Control variables. I obtained the direct reports’ responses of demographic variables such as gender, educational level, age, rank ,and work-unit tenure to control for potential confounding effects on voice behavior (Detert & Burris, 2007; Hsiung, 2011). I also controlled for leader-member exchange (LMX), which might potentially influence the
leader-follower relationship in the workplace as well as employee outcomes (Hsiung, 2011). I measured LMX using Graen and Uhl-Bien’s (1995) seven-item scale (LMX-7). A sample item is “How well does your supervisor understand your job problems and needs?” The Cronbach’s alpha coefficient for this scale was .80.