Homework answers / question archive / Did the authors construct scales in a clear, concise manner? Were the authors’ explanations clear regarding the reliability and validity of their measures? What information did the authors provide Military Psychology ISSN: 0899-5605 (Print) 1532-7876 (Online) Journal homepage: https://www

Did the authors construct scales in a clear, concise manner? Were the authors’ explanations clear regarding the reliability and validity of their measures? What information did the authors provide Military Psychology ISSN: 0899-5605 (Print) 1532-7876 (Online) Journal homepage: https://www

Math

Share With

• Did the authors construct scales in a clear, concise manner?
• Were the authors’ explanations clear regarding the reliability and validity of their measures?
• What information did the authors provide Military Psychology ISSN: 0899-5605 (Print) 1532-7876 (Online) Journal homepage: https://www.tandfonline.com/loi/hmlp20 Trust in military leader: Scale development and validation ?smail Ye?ilba? & ?ahin Çetin To cite this article: ?smail Ye?ilba? & ?ahin Çetin (2019) Trust in military leader: Scale development and validation, Military Psychology, 31:2, 147-159, DOI: 10.1080/08995605.2019.1578150 To link to this article: https://doi.org/10.1080/08995605.2019.1578150 Published online: 21 Mar 2019. Submit your article to this journal Article views: 679 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=hmlp20 MILITARY PSYCHOLOGY 2019, VOL. 31, NO. 2, 147–159 https://doi.org/10.1080/08995605.2019.1578150 Trust in military leader: Scale development and validation ?smail Ye?ilba?a and ?ahin Çetinb a Turkish Land Forces Command HQ, Ankara, Turkey; bEducation Faculty, Maltepe University, Istanbul, Turkey ABSTRACT ARTICLE HISTORY The purpose of this study was to develop an original Trust in Military Leader Scale that can be used in Turkish military context. To this end, we first reviewed the literature to develop a conceptual framework. Next, an item pool was formed and afterwards, an initial scale was formed and evaluated. Semistructured interviews were carried out with 124 officers and noncommissioned officers for item generation. The pilot study involved 2,731 and the final study involved 2,308 participants, all of whom were military personnel. Data obtained through quantitative and qualitative methods were analyzed using such techniques as content analysis, descriptive analysis, consistency analysis, correlational analysis, and exploratory and confirmatory factor analysis. Exploratory factor analysis yielded an 18item scale with 4 factors: competence, benevolence, integrity, and fairness. In the next phase, the scale was analyzed with respect to content, criterion, and construct validity. To test predictive validity, the scale was used to find how much trust in military leader predicted organizational commitment and organizational citizenship behavior. Finally, a valid and reliable scale that can be used in measuring subordinates’ trust in the military leader was developed. Received 30 January 2018 Accepted 22 January 2019 Introduction Trust has been defined as “a psychological state comprising intention to accept vulnerability based upon positive expectations of the intentions or behavior of another” (Rousseau, Sitkin, Burt, & Camerer, 1998, p. 395) “irrespective of the ability [of the trustor] to monitor or control that person” (Mayer, Davis, & Schoorman, 1995, p. 712). The individual who trusts another person expects or believes that he/she can rely on that person’s words, behaviors, and good intentions (Bligh, 2017). Importance of trust in leadership has long been recognized by scholars and researchers in numerous fields and across multiple disciplines. For example, Bennis and Nanus (1985) suggested that leadership effectiveness depends on followers’ trust in the leader. Several leadership theories like transformational leadership (Podsakoff, Mackenzie, Moorman, & Fetter, 1990), charismatic leadership (Conger, Kanungo, & Menon, 2000), and leader–member exchange theory (Schriesheim, Castro, & Cogliser, 1999) emphasize trust as a key process. Braun, Peus, Weisweiler, and Frey (2013) showed that trust in supervisor mediated the relationship between individual perceptions of supervisors’ transformational leadership and job satisfaction. In a meta- KEYWORDS Trust; military leadership; scale development and validation analytic review of relevant research, Dirks and Skarlicki (2007) found significant relationships between trust in leadership and several job-related outcomes including job performance, organizational citizenship behavior, turnover intentions, job satisfaction, organizational commitment, and commitment to the leader’s decisions. Most definitions of trust share three basic characteristics: predictability (i.e., expectation that the trustee will act in a predictable manner), vulnerability (to the actions of the trustee) and risk, and uncertainty (i.e., not knowing for sure whether the trustor will do as agreed; Lester & Vogelgesang, 2012). Interdependence, risk, vulnerability, and uncertainty inherent in trust makes it even more critical and relevant in the military context (Costa, Roe, & Taillieu, 2001; Rousseau et al., 1998) because military operations often involve high levels of risk, interdependence, and uncertainty. Interpersonal trust or its lack thereof may have a crucial influence on the successful accomplishment of a mission and personal safety of the individuals involved. Studies conducted in operational contexts show that soldiers who do not trust their leaders could question or deliberately disobey orders or directives, or comply with them only after minimizing risk (Sweeney, Dirks, Sundberg, & Lester, 2011). CONTACT ?ahin Çetin sahincetin@maltepe.edu.tr Marmara E?itim Köyü 34857 Maltepe, Istanbul, Turkey. Authors’ Note: The views stated in this publication are those of the authors and do not reflect the official opinion of the Turkish Armed Forces. Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/hmlp. This article has been republished with minor changes. These changes do not impact the academic content of the article. © 2019 Society for Military Psychology, Division 19 of the American Psychological Association 148 ?. YE?ILBA? AND ?. ÇETIN Measuring subordinates’ trust in the military leader and understanding its antecedents is important. Although trust in teams and leadership has been wellstudied in the literature, research on identifying antecedents and developing a scale of trust in the military leader is limited. Therefore, we aimed to develop and validate an independent measure of trust in the military leader in the Turkish military context. We sought to make a unique contribution to the literature by (a) explicitly focusing on trust in military leaders, and (b) conducting this research in Turkey, a context that is different from the one upon which most extant literature is based. Leader trust scale development research Cook and Wall (1980) distinguished three approaches to the measurement of trust based on the degree to which it is direct or indirect. The most indirect method is marked by a reliance on some other behaviors (e.g., involving subordinates in decision making) to infer trust. The second approach is to create a situation in which the development of trust is essential to the performance of a task and study the level of performance as an index of trust. The third approach is the direct measurement of trust as an affective reaction by means of a scale. Most of the studies on trust, including the present one, involve direct measurement of trust by self-report scales. In all three perspectives, trust is conceptualized and measured as a perception of the follower rather than a property of the relationship or the leader. The first line of research on antecedents of trust seems to have started with the pioneering study of Jennings (1971) who used interviews of managers to identify four conditions (determinants) of trust: loyalty, accessibility (i.e., mental openness and receptiveness to ideas), availability (i.e., being present when needed), and predictability. Also using clinical interviews, Gabarro (1978) extended Jennings’ list by specifying nine bases of trust: integrity (honesty and moral character), motives (intentions and agenda), consistency of behavior (similar to what Jennings termed predictability), openness (expressing ideas freely), discreetness (keeping confidences), functional/specific competence (task-related knowledge and skills), interpersonal competence (social skills), business sense (common sense and wisdom about business), and judgment (ability to make good decisions). These two studies spurred similar research. For instance, Butler and Cantrell (1984) conducted an experimental study to investigate the relative importance of five determinants of dyadic trust suggested by Jennings (1971) and Gabarro (1978) from the perspectives of subordinates (trust in superiors) and superiors (trust in subordinates). They found that for both directions, the determinants were ranked in the same order of importance: competence, integrity, consistency, loyalty, and openness. Later, Butler (1991) extended and validated the list of trust conditions, but his study focused on determinants of interpersonal rather than managerial trust. Using previous research and interviews with managers he identified 10 determinants of interpersonal trust: availability, competence, consistency, fairness, integrity, loyalty, openness, overall trust, promise fulfillment, and receptivity. His Conditions of Trust Inventory (CTI) consisted of 44 items and with “overall trust” loading together with “integrity” on one factor, nine factors explained 73.5% of the variance. Somewhat different from these studies is the study carried out by Cook and Wall (1980) who defined trust along two dimensions: (a) faith in the intentions of others, and (b) confidence in the ability of others. They hypothesized that each dimension could refer to trust in either (a) peers or (b) management; hence a fourfold classification. Yet, the confirmatory factor analysis (CFA) yielded two dimensions: a “management” subscale and a “peers” subscale. Podsakoff et al. (1990) did not make a distinction between trust in the subordinate or the leader and they conceptualized trust as faith in and loyalty to the leader. They used six items to measure these two dimensions with each factor consisting of three items. Two of the items were derived from the scale developed by Cook and Wall (1980) and the rest were developed by the authors. Yet, results of the CFA indicated that all items loaded on one factor and they thus conceptualized trust as a unidimensional construct. Focusing on managers’ beliefs about the trustworthiness of peers, McAllister (1995) also conceptualized trust along two dimensions: affect-based and cognition-based trust. His measure consisted of 11 items (six items for cognitive trust and five for affective trust) and he found empirical support for his two-factor model. Another measure of trust in the leader was developed by Dirks (2000). Using a sample of 355 individuals from 30 National Collegiate Athletic Association (NCAA) teams, he examined the relationship between trust, leadership, and team performance with an instrument adapted from McAllister’s (1995) study. His scale consisted of 9 items that loaded onto a single factor (eigenvalue = 7.23) and explained 80% of the variance. Coefficient alpha for the scale was high (α = .96). A review of the relevant literature reveals that few studies have been carried out focusing on scale development in the Turkish context. The first study was conducted by ?slamo?lu, Birsel, and Börü (2007) and a Trust in Management Scale consisting of 40 items and 10 subscales (supporting, integrity and fairness, team leadership, creating a positive climate, self-confidence, MILITARY PSYCHOLOGY avoiding stress in the workplace, sharing information, building confidence, competence, empowerment, and consideration) was developed. The scale had high reliability (α = .96) and accounted for 60.5% of total variance. The second study in the Turkish context was carried out by Erdem and Aytemur (2009) with 102 middle-level managers from around Turkey. The scale, which had an alpha coefficient of .94 consisted of 19 items and three factors (competence, patronizing and fairness) and accounted for 68% of the variance. Another study investigated trust in three different foci (supervisors, peers, and subordinates) across the Turkish and Chinese contexts (Wasti, Tan, & Erdil, 2011). Using semi-structured interviews, researchers found ability, benevolence, and integrity to predict trust in both cultures. Benevolence was especially important and trust relationships usually went beyond the professional domain. Taken together, research on the antecedents of trust in the leader seems to feature a number of key factors. Three factors are especially common across many studies and different cultural contexts (Burke, Sims, Lazzara, & Salas, 2007; Gill, Boies, Finegan, & McNally, 2005; Mayer et al., 1995; Wasti et al., 2011): competence, benevolence, and integrity. Although trust is thought to be more critical in the military context, a review of the relevant literature shows that there have been few empirical studies of trust in the leader and only one of these was aimed at developing a scale of trust in the military leader. The first study in this context was carried out by Deluga (1995) who used Butler’s Conditions of Trust Inventory (CTI) to collect data from 64 supervisor–subordinate dyads (97.5% of supervisors and 85.6% of subordinates were military) working at a military base in the Northeastern United States. Supervisors’ trust-facilitating behaviors (10 conditions of trust) were found to be positively associated with subordinates’ extra-role behaviors. In a study with company leaders, staff members and soldiers from Israel Defense Forces companies; Shamir, Brainin, Zakay, and Popper (2000) investigated the relationship between soldiers’ collective efficacy beliefs (conceptualized as perceived combat readiness) and some of its correlates (including confidence in the company commander). They used four independent items to measure soldiers’ confidence in their company commanders. The alpha coefficient for these items was .85. Perceived combat readiness of the unit was found to be significantly related to soldiers’ confidence in the leader. The only study aiming to develop an independent measure of trust in leaders (along with teams) at small unit level was carried out in Canada as part of a Defence Research and Development Project. The Leader Trust Scale, which had four factors (benevolence, integrity, predictability, and 149 competence) and 40 items in total, was later analyzed and revised in two further studies (Adams, Bruyn, & ChungYan, 2004; Adams & Sartori, 2006). However, some substantial differences (like the addition of some completely new items) between the two versions of the same scale across two studies raised concerns about the continuity of scale development work and a third study was undertaken by Blais and Thompson (2009) to further analyze and refine the scale. The third study led to a third version that retained 28 items (seven items per each subscale), and although the revised scale had good internal consistency reliability estimates, some weaknesses were reported with the benevolence subscale. Other studies of leader trust in military leader have either used scales developed in business organizations (e.g., Deluga, 1995) or independent items to be aggregated to a composite measure of trust (Shamir et al., 2000). Taking the reported weaknesses of the Leader Trust Scale into consideration, we aimed to develop and validate a new, independent scale of trust in the military leader. The present study was also motivated by the fact that the Leader Trust Scale (Adams et al., 2004; Adams & Sartori, 2006; Blais & Thompson, 2009) was carried out in Canada. Canadian military context is different from the Turkish context in important ways. Unlike Canada, which is high in individualism but low in power distance and uncertainty avoidance, Turkey represents a high power distance culture with high levels of uncertainty avoidance and collectivism (Hofstede, 1980). Such cultural differences can be expected to play into subordinates’ perceptions of trust in the military leader. By developing a new scale of trust in the military leader in a non-Western military culture context we hoped to contribute to the understanding of trust in the military leader as well as fill an important gap in the literature. Method This study was carried out in three consecutive phases each of which will be described in detail in the following sections. In the first phase, an item pool was formed through the collection of qualitative data. The second phase involved the development of the scale. The scale was later analyzed and tested (Devellis, 2003; Hinkin, 1995). Throughout this process, four different samples were used. Study 1: Item pool generation Method We first reviewed the relevant literature and conducted semi-structured interviews with military personnel to form an initial item pool. 150 ?. YE?ILBA? AND ?. ÇETIN Sample The sample consisted of 124 Turkish Armed Forces (TAF) officers/noncommissioned officers (NCOs) who were reached through a snowball sampling strategy. We conducted semistructured face-to-face interviews with most of the participants (110/124) and reached others (14/124) through e-mail. These participants, whom we reached through e-mail, were asked to provide written answers to the four questions in the written interview form and e-mail their answers back to the authors. Most of the participants in Study 1 (98.4%, n = 122) were male and aged between 26–35 (71%, n = 88). Officers comprised 63.7% (n = 79) of the sample and in terms of current duty, 48.4% (n = 60) of the participants were stationed in Ankara. Procedure In oral/written semistructured interviews, participants were asked four open-ended questions regarding the characteristics of a trustworthy leader: (a) What attitudes/behaviors of your leader might lead you to trust him/her? (b) What attitudes/behaviors of your leader might lead you to lose trust in him/her? (c) Briefly describe an episode you have experienced regarding trust/distrust in your leader. (d) List five characteristics that you think a leader should possess so that you will trust him/her. Comments expressed in written/oral semistructured interviews were analyzed inductively and coded by the authors to discover substantive categories. At this point, we allowed categories to “emerge” from the data and “make” their way into the developing model. Combining the analysis with a review of the relevant literature, we later sought to identify a list of leader characteristics necessary for trust. Results Our initial analysis of the responses to the three questions (i.e., 1, 2, and 4) led to the extraction of 1,068 expressions. These expressions were further analyzed, grouped according to similarity in meaning, and later categorized under 178 characteristics. Next, to discover most frequently expressed characteristics, we removed the ones that were expressed three times and less, thus shrinking the list from 178 to 56 characteristics. The most frequently expressed characteristics were (in descending order): fair (f = 97, 9.1%), competent (f = 68, 6.4%), honest, consistent, sincere, paternalist, considerate, caring, respectful and supporting. We next focused on the third question in the interview form (i.e., Briefly describe an episode you have experienced regarding trust/distrust in your leader) and analyzed the 86 episodes participants found important and described briefly (in writing or orally) regarding trust/distrust in their leader. At this point following a deductive strategy and using the categories from relevant literature, we first coded data and later sorted it into four themes: benevolence (38%, n = 33), fairness (28%, n = 24), integrity (28%, n = 24), and competence (6%, n = 5). For instance, some of the codes under benevolence were “supporting subordinates, watching subordinates’ back, caring for subordinates, establishing direct communication with subordinates and taking responsibility.” We later used the 56 characteristics derived through thematic analysis, the four themes from memorable episodes and the items from the relevant manager/leader trust literature (Adams & Sartori, 2006; Dirks, 2000; Erdem & Aytemur, 2009; ?slamo?lu et al., 2007; McAllister, 1995; Podsakoff et al., 1990; Shamir et al., 2000) to generate the initial item pool for the Trust in Military Leader Scale (TMLS). Items from these scales were compared and similar items blended to generate new items. Drawing upon earlier work (Burke et al., 2007; Colquitt, Scott, & Lepine, 2007; Mayer et al., 1995), we used the three factors that were common across most trust scales in the literature: competence, benevolence, integrity. The item pool consisting of 130 items (43 items each for competence and benevolence and 44 items for integrity) included all of the characteristics and codes from the interviews. Study 2: Scale development Method In the second phase of the study, the initial scale consisting of 130 items was assessed by experts for content validity and later tested in a pilot study. Content validity is the degree to which items represent a proper sample of the theoretical content of a construct. For content validity, items need to be reflecting what they are intended to measure (Nunnally & Bernstein, 1994). Sample In the first part of this phase of the study, items of the initial scale was assessed by 10 army officers receiving postgraduate education on leadership at a military institute in Ankara. The average age of participants was 30.4 years and length of service as an officer was 10.4 years. Next, 11 scholars (8 male, 3 female) were asked to assess the items of the initial scale. Their fields of study were management (55%, n = 6), psychology (27%, n = 3), law (9%, n = 1) and economics (9%, n = 1) respectively. Most of the participants had the title of professor (36%, n = 4) and two listed “trust in manager” among their research interests. MILITARY PSYCHOLOGY Following experts’ assessment of the initial item pool, we tested the scale in a pilot study. After securing official permission from the Turkish Land Forces Command (TLFC) to carry out the study, the scale was e-mailed to all TLFC officers and NCOs personally (134,725 individuals as of January 2014) through the intranet. A total of 2,861 questionnaires were returned. The 95 of these questionnaires were canceled due to improper responding (too many missing answers, same answers for all questions etc.) and 35 questionnaires were canceled because respondents reported that they had been with their present leaders for less than six months. Consequently, the sample of the pilot study consisted of 2,731 participants who were predominantly male (99.3%, n = 2,713). Majority of the participants (71%, n = 1,939) were NCOs. The average age was 36.73 years and the average length of service was 16.08 years. The majority (57.1%, n = 1,560) had been with the present leader for at least one year or more and on average participants had been with their present leaders for 1.68 years. Procedure To assess the items of the initial scale, the first group of participants (army officers receiving postgraduate education) were asked to study each item individually and specify whether it seemed to measure the intended construct, whether it was appropriate for the military context/intended respondents and whether it was clear and well-expressed. Following participants’ individual evaluations, we held a panel meeting with participants to discuss items one by one and decide, as authors, whether to retain, modify or discard each item. Following this phase, the second group of participants (scholars) assessed the scale individually. Lawshe’s content validity ratio was used for analyzing items. Both groups of participants were provided with information involving a brief overview of the study as well as the meaning of the construct and its subscales before evaluation. Following the assessment of the items by the two groups of participants, the scale with 54 items was tested in a pilot study. Items were scored on a 5-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree). Statistics Package for Social Sciences (SPSS 20.0) was used for statistical analysis of the data. Results For the first part of Study 2, participants’ assessment of the 130 items led to the removal of 52 items on the grounds that they were “not adequately reflecting what 151 they were intending to measure” or “not contextually appropriate for the military context/intended respondents” leaving 78 items (26 items for each subscale). Next, the panel of subject matter experts (scholars) participating in the second part of Study 2 were asked to assess the 78-item scale individually and specify whether each item was “essential,” “useful, but not essential,” or “not necessary.” Participants were also asked to provide additional remarks concerning the items if any. Items that were deemed “essential” by a critical number of panel members were included in the final instrument and items failing to achieve this critical level discarded. Lawshe (1975, p. 568) contends that for a panel of 11 subject matter experts, critical content validity ratio (CVR) is .59. Thus 24 items with a lower (CVR < .59) ratio value were removed (seven items from the competence subscale, nine items from the benevolence subscale, and eight items from the integrity subscale) leaving a 54-item scale. Additional remarks by the experts were also analyzed and some minor modifications (e.g., rewording some items) made by authors to make sure items were clear and relevant to the context/construct. Consequently, the final version of the TMLS which had 54 items and three subscales (competence 21 items, benevolence 17 items, and integrity 18 items) was ready for the pilot study. Findings of the pilot study (N = 2,731) showed that the mean for the overall scale was 3.66, the standard deviation was 1.19, and the alpha coefficient was .99. The item with the highest mean was “My leader has a high level of personal integrity” (M = 3.66) and the item with the lowest mean was “My leader empathizes with me” (M = 3.25). Correlations between items were all higher than the critical value of .40. Cronbach’s alpha coefficient and the correlations between items showed that the scale had internal consistency (Berthon, Ewing, & Hah, 2005, pp. 157–159). Next, Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was conducted and the KMO value was found to be .979 implying a sufficient sample size (?slamo?lu & Aln?aç?k, 2013, p. 369). Bartlett’s test of sphericity was carried out and the chi-square value (70,308.386) was found to be significant. Next, the correlation matrix was analyzed and the correlation between some items was found to be very high (r > .90) implying the problem of multicollinearity (Çokluk, ?ekercio?lu, & Büyüköztürk, 2012, p. 35). Four items that had multicollinearity (r > .90) with more than one item (Items 8, 23, 25, and 33) were removed from the scale and principal components analysis was conducted for the remaining 50 items. Using varimax rotation, principal components analysis was repeated iteratively until the best factor structure was obtained. Items that loaded highly (> .40) on more than one factor or that loaded on more than one factor 152 ?. YE?ILBA? AND ?. ÇETIN with a very small (< .10) difference between factor loadings were removed one by one and varimax rotation was rerun. We referred to subject matter experts’ comments in the second phase of Study 2 as well as descriptive, correlational, and internal consistency analyses to make sure important content was not lost in this stage. Finally, as shown in Table 1, a parsimonious four-factor scale consisting of 18 items was obtained. As indicated in Table 1, the scale consisting of four factors and 18 items (six items for competence, six items for benevolence, three items for integrity, and three items for fairness) had a Cronbach’s alpha coefficient of .98 and accounted for 87.8% of total variance which is higher than the cut-off value of 60% (Hinkin, 1998, p. 112). The final phase of the study involved testing of the new scale. Study 3: Scale testing Method In this phase, we administered the TMLS along with three other scales to assess how well it predicted an outcome for another measure and how well it measured the intended construct. Like in the pilot study, four scales were e-mailed to all TLFC officers and NCOs personally through the intranet. A total of 2,377 questionnaires were returned and, after canceling 69 questionnaires for a variety of reasons (e.g., too many missing answers, same answers to all questions etc.), 2,308 questionnaires were evaluated. The average age of the participants was 36.72 years and the average years of service was 15.84 years. Like in the second study, participants were predominantly male (99.2%, n = 2,289) and the majority (66.7%) were NCOs. Procedure All four scales used in this phase were scored on a 5-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree). We used SPSS 20.0 and LISREL 8.8 software to analyze data statistically. We analyzed the correlation between the TMLS and Trust in the Manager Scale to assess concurrent validity. Predictive validity was assessed through hierarchical regression analysis in which trust in the military leader was the independent variable and OC and OCB were dependent variables. Drawing upon Hinkin’s (1995, p. 976) suggestion, we used CFA to assess construct validity of the scale. Sample In the final phase of the present study, the TMLS was administered along with three other scales: Trust in the Manager Scale (Erdem & Aytemur, 2009), Organizational Commitment (OC) Scale (Ergin, 1999) and Organizational Citizenship Behavior (OCB) Scale (?ahin & Gürbüz, 2012). Results Using data from this phase of the study we first ran descriptive analyses for the TMLS and next analyzed the correlation between the TMLS and its subscales. Results can be seen in Table 2. Table 1. Factor structure of the trust in military leader scale. Item no. Factors Factor 1. Competence 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 KMO 0.98 My My My My My My leader leader leader leader leader leader has adequate job related experience. has a high level of job related knowledge. is highly skilled. is good at dealing with difficult situations. is a good orator. is highly skilled in planning work and assigning responsibilities. Cronbach’s α 0.96 Variance 28.513% Factor loadings Eigenvalue = 5.132 Factor 2: Benevolence My leader makes me feel I am valuable. My leader is concerned about my well being. My leader watches my back. My leader empathizes with me. My leader cares for what I think on an issue. My leader protects my rights and interests. Factor 3: Integrity My leader has a high level of personal integrity. My leader is honest. My leader is truthful. Factor 4: Fairness 0.97 27.139% Eigenvalue = 4.885 My leader treats subordinates equally. My leader applies rules equally to everyone. My leader is fair in distributing rewards. Total variance explained = 87.8% Eigenvalue = 2.213 0.94 19.847% Eigenvalue = 3.572 0.94 0.798 0.783 0.724 0.718 0.690 0.669 0.734 0.727 0.714 0.693 0.680 0.656 0.838 0.759 0.713 12.296% 0.629 0.624 0.592 Cronbach’s alpha (scale) = 0.98 Note. The scale was originally developed in Turkish, but a panel of five experienced and competent teachers of English worked together to assure translation validity through a process of three steps: items were translated into English individually, the panel of teachers came together and reached a consensus on the best definition of each item and finally translated items were assessed and confirmed by the panel. MILITARY PSYCHOLOGY 153 Table 2. Descriptive statistics and correlations for the trust in military leader scale. Descriptive statistics Subscales 1. Competence 2. Benevolence 3. Integrity 4. Fairness Overall scale M 3.48 3.38 4.05 3.41 3.53 SD 1.21 1.29 1.08 1.33 1.16 Correlations Cronbachα 0.96 0.97 0.96 0.95 0.98 2 0.859** 3 0.761** 0.791** 4 0.816** 0.887** 0.791** Overall Scale 0.943** 0.964** 0.866** 0.929** Note. N = 2,308. ** p < .01. As shown in Table 2, the overall scale had a high Cronbach’s alpha coefficient (α = .98). Score average for the overall scale was found to be 3.53 and the standard deviation was 1.16. The subscale with the highest score was integrity (M = 4.05) and the item with the highest average score was item number 13 (i.e., “My leader has a high level of personal integrity”). The subscale with the lowest score average was benevolence (M = 3.38). Subscales had also high reliability coefficients with Cronbach’s alpha values ranging between .95 and .98. Correlations between items were all higher than the cutoff value of .40 (Hinkin, 1998, p. 112). All subscales correlated significantly with each other and the overall scale as well. For the overall scale, the highest correlation was with the benevolence subscale (r = .964), and the lowest correlation was with integrity (r = .866). For subscales, the highest correlation was between benevolence and fairness (r = .887), and the lowest correlation was between competence and integrity (r = .761). Reliability coefficients for the other three scales used in this phase were also satisfactory (i.e., between .70 and 1.0). Cronbach’s alpha was .98 for the Trust in the Manager Scale; .78 for the OC Scale and .77 for the OCB Scale. Next, we carried out a CFA to assess the factor structure of the TMLS (Hinkin, 1998, p. 114). First-order CFA was conducted on LISREL 8.8 using maximum likelihood method because data obtained from 2,308 participants followed a normal distribution. The path analysis for the first-order CFA can be seen in Figure 1. For CFA, χ2, df, χ2/df, root mean square error of approximation (RMSEA), comparative fit index (CFI), and goodness-of-fit index (GFI) values were analyzed for assessing goodness of fit. We assessed indices for the null model and the single-factor model as well as first-order and secondorder four-factor models. Second-order four-factor model was involved in the analysis to test if the four factors of military leader trust subsumed under a higher-order factor. Results can be seen in Table 3. As shown in Table 3, first-order four-factor model had better fit than alternative models because RMSEA (.056), GFI (.95), and CFI (1) values for the first-order four-factor model were good. However, (χ2/df) value (8.33) for the first-order four-factor model was above the critical value of (2< χ2/df ≤ 3) (Brown, 2006, p. 113). We believe that this may be due to the very large sample size because chi-square is sensitive to Figure 1. First order confirmatory factor analysis (n = 2308). Note. (Yetenek: Competence, Yarsev: Benevolence, Karbut: Integrity, Adalet: Fairness). 154 ?. YE?ILBA? AND ?. ÇETIN Table 3. Confirmatory Factor Analysis Goodness of Fit indices for alternative models. Model comparison test Model A. First order four-factor model B. Second order four-factor model C. Null model D. First order three-factor model model E. Single factor model χ 1066.30 1125.46 1637.41 2417.79 10,565.25 2 SD 128 130 129 130 134 χ2/df 8.33 8.66 12.69 17.68 78.84 RMSEA 0.056 0.058 0.071 0.087 0.180 GFI 0.95 0.95 0.93 0.90 0.66 CFI 1 1 0.99 0.99 0.96 Comparisonn — B&A C&A D&A E&A Δχ2 — 59.16 571.11 1351.49 9498.95 ΔSD> — 2 1 2 6 Note. RMSEA = root mea square error of approximation; GFI = goodness-of-fit index; CFI = comparison fit index. N = 2,308. sample size (Tabachnick & Fidell, 2007, p. 695). Guided by Bergh’s (2015) suggestion that adjusting the sample size may be one strategy to deal with this problem, we reconducted the analysis with a smaller sample (n = 193) consisting of those participants who had been with the current leader for longer than 3 years. Results of the CFA for the subsample can be seen in Table 4. As seen in Table 4, with the new sample of 193 participants, the (χ2/df) value (1.79) and the CFI (.99) values for the four-factor model were good and RMSEA (.064) and GFI (.89) values were acceptable (Brown, 2006, p. 113). We next compared the four-factor model with the single-factor and the three-factor models. Drawing upon prior work (Adams & Sartori, 2006; Burke et al., 2007; Mayer et al., 1995) we obtained the three-factor model by combining fairness and integrity into one factor. As can be seen in Table 4, the four-factor model gave a better fit than alternative models because (χ2/df) values (3.03 > 3 for the three-factor model and 7.33 > 3 for the single-factor model) and GFI values (.89 for the threefactor model and .64 for the single-factor model) for alternative models were not within acceptable limits. Next, we analyzed the correlation between the TMLS and Trust in Manager Scale (Erdem & Aytemur, 2009). Results of the Pearson correlation analysis can be seen in Table 5. As shown in Table 5, significant correlations between the overall scales (r = .97) and all subscales of the TMLS and Trust in Manager Scales (with correlation coefficients ranging between r = .79 and r = .94) show that the TMLS has convergent validity. Yet, the high correlation value between the two scales (r = .97) is an important issue and will be discussed in more detail in Discussion. We later analyzed how well the TMLS predicted another measure. Research has shown that trust in the leader is a significant predictor of a variety of job-related attitudes and behaviors (e.g., Burke et al., 2007; Dirks & Ferrin, 2002; Zhu, Newman, Miao, & Hooke, 2013). Hence we investigated whether trust in military leader predicted organizational commitment (OC) and organizational citizenship behavior (OCB). The relationship between trust in leader and OC (e.g. Dirks & Ferrin, 2002; Zhu et al., 2013) and trust in leader and OCB (Burke et al., 2007; Colquitt et al., 2007; Yang & Mossholder, 2010; Zhu et al., 2013) is well documented in relevant literature. We, therefore, conducted a regression analysis to test whether trust in the military leader predicted OC and OCB. Results can be seen in Table 6. As seen in Table 6 we first entered into our analysis such control variables as gender, degree, years of service with the present leader, age, years of service and status. Demographic variables explained 6.4% of the variance in OC and 2.2% of the variance in OCB. In the second step, Table 4. GFI indices for the Four-Factor Model and alternative models using a subsample. Model comparison test Model A. Four-factor model B. Three factor model C. Single factor model χ2 229.19 396.42 982.98 df 128 131 134 χ2/df 1.79 3.03 7.33 RMSEA 0.064 0.10 0.18 GFI 0.89 0.81 0.64 CFI 0.99 0.98 0.96 Δχ2 — B and A C and A ΔSD — 167.23 753.79 Note. RMSEA = root mea square error of approximation; GFI = goodness-of-fit index; CFI = comparison fit index. N = 193. Table 5. Correlation between Trust in Military Leader and Trust in Manager Scales. Trust in Manager Scale Trust in Military Leader Scale Competence Benevolence Integrity Fairness Trust in Military Leader Note. N = 2,308. **p < .01. Competence Patronizing Fairness Trust in Manager Scale 0.940** 0.851** 0.777** 0.825** 0.923** 0.829** 0.931** 0.811** 0.871** 0.928** 0.841** 0.901** 0.793** 0.942** 0.932** 0.914** 0.937** 0.832** 0.916** 0.972** Δχ2 — 3 6 MILITARY PSYCHOLOGY Table 6. Hierarchical regression between Trust in Leader, OC, and OCB. Organizational Commitment Model 1 2 Predictor Gender Degree Years of service with the leader Age Years of service Status Gender Degree Years of service with the leader Age Years of service Status Trust in Military Leader R2 N β .038 −.043 −.028 ΔR 2 .064*** OCB β .016 −.006 −.014 ΔR2 .022** .235** .185* −.225** −.172* −.226*** −.112*** .031 .133*** .011 .065*** −.047* −.009 −.020 −.009 .213** −.194** −.150*** .374*** .170* −.150* −.059* .262*** .197*** 2,308 .087*** 2,308 Note. OC = Organizational Commitment Scale; OCB = Organizational Citizenship Behavior Scale. *p < .05. ** p < .01. *** p < .001. controlling for the demographic variables we entered trust in the military leader into our analysis and found that it explained an additional 13.3% of the variance in OC and 6.5% of the variance in OCB. Total variance explained by the model as a whole was 19.7% for OC and 8.7% for OCB. Trust was a significant positive predictor of both OC and OCB. Thus, meeting the two criteria of concurrent and predictive validity, it can be argued that the TMLS has criterion validity. Before moving on to the discussion of our findings, we explored the correlations between all of the scales and subscales for the component measures of trust in the military leader, trust in manager, organizational commitment, and organizational citizenship behavior so that future meta-analyses can make use of these data. Significant positive correlations were found between all scales and subscales. Results have been presented in Table 7 in the appendix. Discussion The purpose of this study was to develop a reliable and valid measure of trust in the military leader. It involved three phases and a reliable and valid TMLS consisting of four factors and 18 items (six items for competence, six items for benevolence, three items for integrity, and three items for fairness) was developed. CFA also indicated that the first order four-factor model was a viable model. The scale had a Cronbach’s alpha value of .98 and explained 87.8% of total variance. Integrity, which is essentially about honesty, truthfulness and moral conduct had higher mean rating (M = 4.05) than other scales of trust in a military leader 155 (i.e., competence, fairness and benevolence). Competence involves a leader’s job-related knowledge and experience, skillfulness in dealing with difficulties, communicative skills as well as effective planning and organizing ability and as such it is more about the leader’s performance than his/her character. Fairness and benevolence are more about a leader’s interactions with his/her subordinates. It should be noted that fairness was emphasized by interviewees in the first study as the most important quality a leader must have for earning subordinates’ trust. A military leader is perceived to be fair if he/she treats all subordinates equally and is fair in distributing rewards and applying rules. Benevolence is related with a leader’s consideration of and care for his/her subordinates. Benevolent leaders support subordinates, care for them, value their input, and are concerned about their rights, interests, and well-being. In the assessment of the scale, we carried out various analyses to validate the scale. Findings of the second and third studies are generally consistent. The high internal consistency of the scale is an issue and it can be related with the high level of homogeneity between participants, as well as lack of reversely scored items that serve the useful purpose of disrupting undesirable response sets (Nunnally, 1967; Pedhazur & Schmelkin, 1991). Furthermore, it is important to note that similar scale development studies also found high reliability coefficients. For instance, Adams and Sartori’s (2006) Leader Trust Scale had a Cronbach’s alpha value of .97. Other leader/manager trust scales developed in nonmilitary contexts too had Cronbach’s alpha values well above .95 (e.g., Börü, ?slamo?lu, & Birsel, 2007, found an alpha value of .96; Erdem & Aytemur, 2009, found an alpha value of .94; McAllister, 1995, found an alpha value of .95). CFA showed that the four-factor model had better fit than alternative models. A comparison of similar scales developed in Turkish nonmilitary settings indicates that trust consists of similar factors in both contexts and that a great deal of overlap exists between all scales. For instance, the scale developed by Börü et al. (2007) has four factors: ability, benevolence, truthfulness (similar to the integrity factor of the present scale), and openness. Likewise, the scale developed by Erdem and Aytemur (2009) had three factors; competence, patronizing (similar to the benevolence factor of the present scale), and fairness. Wasti et al. (2011) too defined three components of trust in the manager across Turkish and Chinese cultures: benevolence, integrity, and ability. Thus it can be argued that antecedents of trust are fairly consistent across different types of organizations and trust is related with a relatively universal set of leader characteristics or behaviors. 156 ?. YE?ILBA? AND ?. ÇETIN A review of relevant literature in English indicates that trust in the military leader is associated with similar characteristics or behaviors in Canada too. Using a 7-point Likert scale of leader trust, Adams and Sartori (2006) found competence, integrity, benevolence, and predictability to be antecedents of trust in the military leader. Despite slight differences in terms of the mean ratings of antecedents, there seems to be a great deal of overlap between their findings and our findings. These findings too suggest that trust is more of a universal than a context-dependent concept showing little variation across cultures. Assessment of the correlational validity was carried out through an analysis of the convergent and predictive validity of the scale. A very high correlation (r = .97) was found between the overall scales of TMLS and Trust in Manager Scale (Erdem & Aytemur, 2009) and their subscales. We believe that this high correlation which is above the critical value of .80 and the marginal difference between average scores (.07; ?encan, 2005, p. 768) can be accounted for by two factors. Firstly, two scales tapping two similar constructs were administered simultaneously causing respondents to make very similar evaluations. Although leadership and management are two distinct and different constructs, most scholars seem to agree that there is a high degree of overlap between the two roles and that managers and leaders are not necessarily very different people. Both roles are indispensable and organizational success depends upon managers fulfilling leadership roles when needed or vice versa (Yukl, 2013). Secondly, we believe that collecting data from the same source (i.e., subordinates’ rating of leaders) may have caused common method bias. Further analysis of the scales shows that despite some similarities, the two scales have some important differences as well. Trust in Manager Scale is the product of a Turkish Scientific and Technological Research Council Project and has three subscales (seven items for competence, seven items for patronizing, and five items for fairness). TMLS involves two subscales (benevolence and integrity) and some important leader behaviors/characteristics (such as powerful oration and personal integrity) that are not represented in the Trust in Manager Scale. Considering the fact that these qualities/behaviors have traditionally been emphasized in military education in Turkey, we believe that the TMLS represents an advancement in military leader trust research. Results of predictive validity analysis indicate that controlling for demographic variables trust in the military leader is a significant predictor of both OC and OCB. Previous research also found similar results. For instance, Zhu et al. (2013) found affective trust, which is more emotional in nature rather than cognitive, to be a significant mediator of the relationship between transformational leadership and affective commitment (β = .32, p < .01), as well as the one between transformational leadership and OCB (β = .28, p < .01). The mediating role of trust in the relationship between transformational leadership and OCB was also confirmed by Podsakoff et al. (1990). Yang and Mossholder (2010) too found affective trust in supervisor (β = .43, p < .01) to be a significant predictor of affective organizational commitment as it accounted for about 31% of the variance in organizational commitment. Leader trust-unit commitment relationship was also confirmed by Tremblay (2010) who held his study with a Canadian sample of military personnel (N = 1,443). In their metaanalysis Dirks and Ferrin (2002) found significant relationships between trust in leadership and each of the types of OCB (e.g., for altruism, r = .19; for civic virtue, r = .11; for conscientiousness, r = .22; for courtesy, r = .22; and for sportsmanship r = .20) and organizational commitment (r = .49). Our findings support previous literature. We believe that the present study makes valuable contributions to the relevant national and international literature of leader trust. Development of a reliable and valid measure of trust in the military leader through a rigorous study combining qualitative and quantitative research designs and using such a large sample of military personnel makes this study valuable and unique. TMLS with English and Turkish versions can provide valuable insights for leader development programs in military organizations. The high internal consistency of the scale is an issue that should be addressed in future research. Therefore we suggest that the scale be tested and refined by researchers in studies in nonmilitary as well as military contexts. We also believe that the instrument can be used in hierarchical and bureaucratic nonmilitary organizations and suggest that it be tested in studies to be conducted in such organizations. Disclosure statement No potential conflict of interest was reported by the authors. References Adams, B. D., Bruyn, L. E., & Chung-Yan, G. (2004). Creating a measure of trust in small military teams (DRDC No: CR 2004-077). Toronto, ON: Canada Defence Research and Development. Adams, B. D., & Sartori, J. A. (2006). Validating the trust in teams and trust in leaders scales (DRDC No: CR 2006-008). Toronto, ON: Canada Defence Research and Development. Bennis, W., & Nanus, B. (1985). Leaders: The strategies for taking charge. New York, NY: Harper & Row. Bergh, D. (2015). Chi-squared test of fit and sample size: A comparison between a random sample approach and a chi-square value adjustment method. Journal of Applied MILITARY PSYCHOLOGY Measurement, 16(2), 204–217. doi:10.1007/978-3-66247490-7_15 Berthon, P., Ewing, M., & Hah, L. L. (2005). Captivating company: Dimensions of attractiveness in employer branding. International Journal of Advertising, 24(2), 151–172. doi:10.1080/02650487.2005.1107912 Blais, A. R., & Thompson, M. (2009). The trust in teams and trust in leaders scale: A review of their psychometric properties and ?tem selection (DRDC No: TM 2009-161). Toronto, ON: Canada Defence Research and Development. Bligh, M. C. (2017). Leadership and trust. In J. Marques & D. Satinder (Eds.), Leadership today: Practices for personal and professional performance (pp. 21–42). Switzerland: Springer. Börü, D., ?slamo?lu, G., & Birsel, M. (2007). Güven: Bir anket geli?tirme çal??mas?. Marmara Üniversitesi Sosyal Bilimler Enstitüsü Öneri Dergisi, 7(27), 49–59. Braun, S., Peus, C., Weisweiler, S., & Frey, D. (2013). Transformational leadership, job satisfaction, and team performance: A multilevel mediation model of trust. The Leadership Quarterly, 24(1), 270–283. doi:10.1016/j. leaqua.2012.11.006 Brown, T. A. (2006). Confirmatory factor analysis for applied research. New York, NY: Guilford Press. Burke, C. S., Sims, D. E., Lazzara, E. H., & Salas, E. (2007). Trust in leadership: A multi-level review and integration. The Leadership Quarterly, 18(6), 606–632. doi:10.1016/j. leaqua.2007.09.006 Butler, J. K. (1991). Toward understanding and measuring conditions of trust: Evolution of a conditions of trust inventory. Journal of Management, 17(3), 643–663. doi:10.1177/014920639101700307 Butler, J. K., & Cantrell, R. S. (1984). A behavioral decision theory approach to modeling dyadic trust in superiors and subordinates. Psychological Reports, 55(1), 19–28. doi:10.2466/pr0.1984.55.1.19 Çokluk, Ö., ?ekercio?lu, G., & Büyüköztürk, ?. (2012). Sosyal bilimler için çok de?i?kenli istatistik SPSS ve LISREL uygulamalar? (2. Bask?). Ankara, Turkey: Pegem Akademi. Colquitt, J. A., Scott, B. A., & Lepine, J. A. (2007). Trust, trustworthiness, and trust propensity: A meta-analytic test of their unique relationships with risk taking and job performance. The Journal of Applied Psychology, 92(4), 909–927. doi:10.1037/0021-9010.92.4.909 Conger, J., Kanungo, B., & Menon, S. (2000). Charismatic leadership and follower effects. Journal of Organizational Behavior, 21, 747–767. doi:10.1002/1099-1379(200011)21:7about establishing reliability and validity?