Sage Journals: Discover world-class research

Abstract

Measures of political skill have been shown to be significant predictors of job performance across a variety of occupations and have consistently been related to positive work-related behaviors and outcomes. A reliability generalization study was conducted on the Political Skill Inventory (PSI), currently the most frequently utilized measuring instrument for assessing the construct of political skill, to determine the weighted mean internal consistency reliability estimate of the PSI and its four dimensions across samples while also examining the effect of six potential sources of measurement error that may impact the internal consistency reliability of the PSI and its four dimensions. Across the samples that reported a reliability estimate for the Ferris et al. 18-item PSI measure, the weighted mean reliability coefficient alpha of the PSI was .89 (k = 77, N = 15,987) and that for the six-item measure of the PSI was .81 (k = 11, N = 2,123). Potential sources of measurement error variance that could impact the typical score reliability of the PSI and three of the four PSI subscale dimensions were identified and are discussed.

Keywords

Political Skill Inventory reliability generalization political skill predictor

An increasing amount of research regarding the correlates and consequences of organizational politics has been conducted in recent years as internal politics have proven to be quite pervasive in the private as well as the public sector jobs (Zettler & Lang, 2015). In recent years, research on organizational politicking has focused on the construct of political skill and how it can predict job performance in organizations (Ferris, Treadway, Brouer, & Munyon, 2012). Political skill can be briefly defined as a comprehensive pattern of “social competencies, with cognitive, affective, and behavioral manifestations” (Ferris et al., 2007, p. 291). Examples of political skill manifestations include networking with others, influencing others, and perceiving close attention to other’s non-verbal behaviors (Ferris et al., 2005). Thus, political skill can help perform essential job functions related to organizational goals and researchers have become increasingly interested in measuring and validating the political skill construct (Ferris et al., 2012). The Political Skill Inventory (PSI) is the most frequently used measure for assessing political skill as it is currently defined as a construct (Ferris et al., 2012; Zettler & Lang, 2015). As the PSI is the most frequently used measure of political skill as a construct, a reliability generalization of the scale was conducted to examine the variability of reliability coefficients in studies that used the PSI.

For the past decade, the political skill construct has received promising empirical support, generating increased attention by both researchers and practitioners, thus advancing the understanding of organizational behavior. More specifically, the political skill construct has received an increasing amount of attention from organizations in recent years because of the generally strong relationship measures political skills have with job performance across a variety of industries (e.g., Blickle, Ferris, et al., 2011; Ferris et al., 2007; Zettler & Lang, 2015; etc.). The main explanation for the general success of political skill in predicting job performance is largely based on the reasoning that individual employees with high levels of political skill are able to quickly identify and understand the subtle and hidden intentions and needs of their organization, co-workers, supervisors, and customers and also possess the competence to strategically modify their own behavior to influence their workplace environment to their advantage (Zettler & Lang, 2015). Previous research has also suggested that those with political skill improve their own performance by getting colleagues to assist them, assertively controlling negotiations and sales, and enhancing social networks that help them perform better (i.e., back-up staff and technical equipment; Zettler & Lang, 2015). In support of these findings, many studies (e.g., Blickle, Ferris et al., 2011; Ferris et al., 2007) including a meta-analysis (Being, Davison, Minor, Novicevic, & Frink, 2011) have all found a significant positive correlation between political skill and job performance across a number of different job types (Bing, Davison, Minor, Novicevic, & Frink, 2011). The primary objective of the present research is to investigate the typical magnitude and variabilities of reliability coefficient estimates reported across existing studies that administered the PSI—one of the most widely used measures of the political skill construct.

Political skill has been broadly defined as “the ability to effectively understand others at work, and to use such knowledge to influence others to act in ways that enhance one’s personal or organizational objectives” (Ahearn, Ferris, Hochwarter, Douglas, & Ammeter, 2004, p. 311). As a result, political skill involves the combination of social astuteness with the ability to adjust behavior to different situations to achieve personal or organizational goals using a manner that is perceived as sincere, inspires trust and motivation, and influences the responses of others (Ferris et al., 2005). More specifically, effective communication, friendships, network building, alliances, and coalition building are critical for individuals to navigate the politics of organizations to effectively achieve desired outcomes and represent an important aspect of political skill in the organizational politics literature (Ahearn et al., 2004; Ferris et al., 1999; Pfeffer, 1992). The other major aspect of political skill represented in the organizational politics literature is genuineness and the appearance of trustworthiness (Ferris et al., 2005). More specifically, the ability to maintain the appearance of sincerity allows politically skilled individuals to achieve a high level of interpersonal influence with individuals at work because they are able to inspire trust that motivates others to assist them with achieving and providing resources for their organizational or personal goals (Ahearn et al., 2004). Thus, the four primary dimensions of the political skill construct reflected in the organizational politics literature are social astuteness, interpersonal influence, networking ability, and apparent sincerity. The social astuteness dimension refers to the ability to accurately perceive as well as comprehend social situations and interpersonal interactions while the apparent sincerity dimension refers to the individual ability to be perceived by others as being honest, authentic, sincere, genuine, and possessing no ulterior motives (Ferris et al., 2005; Pfeffer, 1992). The interpersonal influence dimension involves the ability to appropriately adapt and adjust behavior to each situation to achieve a desired response or outcome, and the networking ability dimension refers to the capability or capacity to easily develop friendships and create strong beneficial alliances and coalitions to create and take advantage of opportunities (Ferris et al., 2005; Pfeffer, 1992).

The first measure that was developed to assess the political skill construct was the unidimensional, six-item PSI scale (Ferris et al., 1999). Although the unidimensional six-item PSI (Ferris et al., 1999) had acceptable psychometric properties, it was created for the primary purposes of preliminary construct exploration (Ferris et al., 2005). As a result, the six-item scale attempted to measure the political skill construct by examining two primary dimensions that were theoretically related to the construct: social astuteness and interpersonal influence (Ahearn et al., 2004; Ferris et al., 1999; Ferris et al., 2005). However, the original six-item PSI measure (Ferris et al., 1999) addressed the social astuteness and interpersonal influence dimensions of the political skill construct without separating the two dimensions into independent factors. The original six-item PSI did not include items for the purposes of assessing the networking ability and apparent sincerity dimensions of the political skill construct reflected in the organizational politics literature (Ahearn et al., 2004; Ferris et al., 2005).

Although the original unidimensional six-item version of the PSI (Ferris et al., 1999) provided initial construct evidence supporting the existence of the underlined latent political skill construct, the call for a more comprehensive, content-valid measure that adequately assessed each of the four primary dimensions of political skill theorized by organizational politicking literature led to the development of the full 18-item PSI (Ferris et al., 2005). To maximize the content validity of the 18-item PSI (Ferris et al., 2005), a pool of 40 items was generated with each item intended to generally asses the political skill construct and to reflect the specific dimensions associated with political skill (Ferris et al., 1999). The items included in the 18-item PSI scale (Ferris et al., 2005) were selected from the 40-item pool based on the results from item analyses in which only the items (a) with the highest item-to-total correlations, (b) not significantly correlated with the Social Desirability Scale total score, and (c) that did not cross-load highly onto more than one factor were retained. A factor analysis of the 18 items of the PSI captured the four-factor solution, with each of the four factors representing a theoretical dimension of the political skill construct: networking ability (six items), interpersonal influence (four items), social astuteness (five items), and apparent sincerity (three items). Of the six items included in the Ferris et al. (1999) PSI, only three were retained as part of the 18-item version of the measure: “I am able to make most people feel comfortable and at ease around me” (interpersonal influence), “it is easy for me to develop good rapport with most people” (interpersonal influence), and “I understand people very well” (social astuteness; Ferris et al., 2005).

Convergent and discriminant validity evidence for the two measures can be found in the literature. For example, Ferris et al. (2005) found evidence that total score of both the six-item and 18-item PSI measures were strongly correlated with self-monitoring, consciousness, trait anxiety, and social desirability which have each been previously hypothesized to be related to the political skill construct. However, Ferris et al. (2005) also found that the 18-item PSI was more related to the influence tactics of upward appeal and coalition in addition to being less related to the influence tactic of assertiveness than the six-item PSI, which indicates the 18-item PSI may more comprehensively capture the political skill construct. The small relationship between the six-item PSI with the influence tactics of appeal and coalition along with the significant relationship between the six-item PSI with assertiveness may likely demonstrate potential improvements of the 18-item PSI over the original six-item measure, as it has been hypothesized that those high in political skill will be more likely to engage in the more subtle and less direct influence tactics of appeal and coalition while avoiding more overt influence tactics such as assertiveness, which involves demanding ordering and checking up on others to exercise influence (Ferris et al., 2005; Kipins, Schmidt, & Wilkinson, 1980).

The full 18-item PSI (Ferris et al., 2005) is currently the most frequently implemented measure of the political skill construct in research being administered to over 76 samples in multiple languages with the six-item unidimensional version of the PSI (Ferris et al., 1999) being the next most frequently administered political skill measure. Thus, as the six-item PSI and 18-item PSI scales are the most frequently implemented and empirically validated measures of the political skill construct found in existing research, both versions of the scale were included in the current study to determine the reliability estimates of each. However, due to the substantial body of evidence indicating adequate psychometric properties and the inclusion of all four dimensions reflecting the political skill construct in the 18-item PSI Ferris et al. (2005) scale, the reliability coefficients obtained from the 18-item PSI represent the primary focus of this generalization study. Thus, the reliability coefficients obtained from the 18-item Ferris et al. (2005) PSI are reported separately in the results from the reliability coefficients obtained from the unidimensional six-item Ferris et al. (1999) PSI. Among the samples that were included in this reliability generalization study that administered the 18-item PSI, reliability coefficients were often reported separately for each of the four dimensions in addition to the reliability coefficient for the entire PSI scale. Reliability estimates reported for the four dimensions were only obtained from samples that administered the 18-item version of the PSI as the six-item PSI measure is unidimensional and does not distinguish between any of the four PSI dimensions.

Reliability

Reliability refers to “the results obtained with an evaluation instrument and not to the instrument itself” (Gronlund & Linn, 1990, p. 78). The psychometric literature is clear that reliability is based on scores gathered from a measurement scale and will vary from sample to sample (Thompson & Vacha-Haase, 2000; Vacha-Haase, Henson, & Caruso, 2002; Vacha-Haase, Kogan, & Thompson, 2000). Thus, it is informative to investigate how the reliability of a measure varies according to sample characteristics, administration procedures, and so on.

The reliability of the test for a population can be defined as the ratio of true score variance to observed score variance, or as the squared correlation between true and observed scores (Lord, 1959; Nunnally, 1978). In actual practice, however, true scores can never be determined; therefore, the reliability is typically estimated by coefficients of internal consistency, test–retest, alternate forms, and other reliability methods mentioned in the psychometric literature (Dimitrov, 2002; Vacha-Haase, 1998; Vacha et al., 2000). For this reliability generalization study, however, given the limited number of studies that reported other types of reliability, the internal consistency reliability coefficient of Cronbach’s alpha alone was investigated in this study. As Cronbach’s alpha will generally increase as the intercorrelations among test items increase and the intercorrelations of test items tend to increase when all of the test items measure the same underlying construct, higher values of alpha are generally thought to indicate that a particular set of test items measure a single construct. Although several researchers (e.g., Cronbach, 1951; Nunnally, 1978, etc.) have shown that alpha values can be considerably high even when the set of items measure multiple unrelated latent constructs, our use of alpha in this reliability generalization study of the PSI seems appropriate given researchers have generally agreed that alpha is most appropriately utilized when the items in a test measure a single construct (Nunnally, 1978).

Reliability Generalization

Reliability generalization (RG) reflects a meta-analytic approach to quantify reliability scores across samples and examine the variability of the reliability estimates (Thompson & Vacha-Haase, 2000; Vacha-Haase, 1998; Viswesvaran & Ones, 2000). Reliability coefficients are collected across studies to calculate central tendency and variability. Sample characteristics can be tabulated in an attempt to account for the variability in reliability estimates (e.g., Thompson & Vacha-Haase, 2000; Yin & Fan, 2000; etc.). In addition, the current reliability generalization study utilizes general linear modeling to provide indications of when the use of the PSI is more or less favorable in terms of expected measurement error. More specifically, sample gender distribution, the number of items included in the measure, sample population, language, geographic location, and the number of response options are all examined in the current study as potential factors that might influence the score reliability obtained from the PSI. Thus, the findings of the current study could likely serve as an additional source of information and reference for future practitioners and researchers in that it may shed light on potential contextual moderators that may lower (or raise) the reliability of the PSI measure depending on different sample and situational circumstances.

Factors That Influence Reliability

Number of Items

As mentioned previously, as the 18-item PSI (Ferris et al., 2005) and the six-item PSI (Ferris et al., 1999) are the most frequently implemented and empirically validated measures of the political skill construct, both versions of the PSI were included in the current study to determine the reliability estimates of each as well as potential contextual variables that may impact the overall scale reliability of each. Although implementing the shortened six-item version of the PSI may be practically appealing to many organizations and researchers due to the reduced time and cost involved during administration, many researchers have contended that the 18-item PSI more accurately reflects the theoretical definition of the political construct by explicitly including items for each of the four subdimensions. In terms of reliability, research has shown consistently that there is a tendency for the internal consistency of a measure to increase when the number of items increases as alpha is a function of the average interterm correlation and the number of items in a measure (Hellman, Fuqua, & Worley, 2006).

Research Question 1: Will the average internal consistency reliability estimate of the full 18-item PSI be significantly higher than that of the shortened six-item version of the PSI?

Number of Response Options

Given the high prevalence of Likert-type scales in organizational research, determining the optimal number of response options for participants to maximize internal consistency is an important consideration during the construction of such scales (Mattel & Jacoby, 1971, 1972). Early work on response format by Bendig (1954) provided evidence that the internal consistency reliability of a test may not be affected much by the number of alternatives provided in Likert-type scales. Providing many response options may not increase reliability if raters consistently utilize only a small proportion of them. In some instances, research has found that reducing the number of response options to an extreme degree can result in lowered estimates of reliability (Mattel & Jacoby, 1971, 1972). For example, Schmidt and Hunter (1998) found that the average correlation of items was significantly reduced when the response options were reduced from multiple to dichotomous response options. However, as all of the PSI measures included in the current reliability generalization study utilize response options from either 1 to 5 or 1 to 7 Likert-type scale, it seems unlikely that there would be a significant difference in the average internal consistency reliability estimates between the two.

Research Question 2: Will the average internal consistency reliability of the PSI scales that utilize 5 response options significantly differ from the PSI scales that utilize 7 response options?

Language

Increasing number of organizations are using personality assessment for selection purposes in international contexts (Bartram, 2008; Oswald, 2008). When measuring a construct for occupational assessment (e.g., as part of a job selection procedure), organizations will often use various adaptations of the same instrument to assess people from various countries and will then make direct comparisons between these people (Bartram, 2008; Oswald, 2008). This raises a need to compare the results of people who have completed an instrument in different countries or using different language versions (Bartram, 2008). The goals of translating a test are to preserve (a) translinguistic and (b) transcultural meaning of test items—so that candidates read and respond to equivalent measures (Daouk, Rust, & McDowall, 2005). However, this is not always possible as some items may sound stronger or weaker in a target language, and as a result, wording sometimes must be changed to reflect the same strength of the item in both language versions of a test (Bartram, 2008; Daouk et al., 2005). Thus, the measurement error of the test may differ across language versions of a measure due to differences in translation of items.

Research Question 3: Will the average internal consistency reliability of the PSI administered in English differ from translated non-English versions of the PSI?

Geographic Location

In addition to language, cultural differences between the various geographic locations a test is administered in may lead to systematic differences in measurement error between samples from different geographic locations (Bartram, 2008; Daouk et al., 2005; Mushquash & Bova, 2007; Oswald, 2008). Specifically, decrements in reliability and validity may arise when a measurement instrument that has been developed in a particular culture is applied to other cultural groups for which it was not originally intended (Mushquash & Bova, 2007; Oswald, 2008). Thus, cultural differences between geographic locations regarding the targeted latent construct of political skill and its four subdimensions may lead to systematic differences in the measurement error of the PSI between samples from different geographic locations.

Research Question 4: Will average internal consistency reliability estimates of the PSI significantly differ between different geographic locations?

Sample Population

By knowing in which types of populations a measure generally has demonstrated low reliability, researchers and practitioners can make more informed decisions when evaluating test scores. Indeed, previous RG studies have found varying reliability coefficients for the same measure and subscales administered to different types of sample populations such as students and employees (e.g., Dimitrov, 2002; Hellman et al., 2006; Vacha-Haase, 1998; Vacha-Haase et al., 2000, etc.). As the PSI was designed to measure employee political skill in the workplace setting, determining if it has an acceptable reliability across samples consisting of all employees is an important contribution of the current RG study. In addition, the current study investigates if the PSI is more or less reliable when it is administered to employees in occupational settings compared with when it is administered to full- or part-time students in academic settings.

Research Question 5: Will the average internal consistency reliability of PSI scores of all-employee samples administered in occupational settings significantly differ from the PSI scores of full- and part-time student samples administered in academic settings?

Sample Gender Distribution

Previous research has demonstrated the demographic composition of a sample can affect the internal consistency of a measure. For example, one investigation found that more homogeneous groups that were administered the Coppersmith Self-Esteem Inventory yielded higher reliability coefficients than heterogeneous samples (Lane, White, & Henson, 2002). One demographic variable that may be especially important to consider when administering an organizational measure to a sample is gender as previous research has indicated that differences in measurement error between countries and languages are smaller on average than differences in measurement error between males and females on international selection tests (Bartram, 2008; Oswald, 2008). For example, Oswald (2008) found there were significant gender differences in scoring and reliability between men and women participants in all but one of the 32 scales contained in the Occupational Personality Questionnaire Inventory (OPQ32i) that describe an individual’s preferred and typical behavior at work (Oswald, 2008). As a result of potential differences in measurement error based on gender, it is important to determine if there are systematic differences in internal consistency reliability estimates of PSI scores based on the gender composition of a sample (Bartram, 2008; Oswald, 2008).

Research Question 6: Will the average internal consistency reliability of PSI scores significantly differ based on the gender composition of the sample?

Method

Following the methods suggested by Vacha-Haase (1998), reliability generalization was completed to (a) calculate the typical reliability of scores of a test across studies, (b) examine the amount of variability in reliability coefficients for the given test, and (c) identify the sources of variability in the reliability coefficients across studies. The keywords “Political Skill Inventory” were entered in a search using the “tests and measurements” search option in the PsycINFO database. From this initial search, we located 98 original journal articles that reportedly included the PSI as a measure in their study. For this study, of the 98 articles, we included only those that met the following criteria: (a) the study administered the PSI scale, (b) a reliability coefficient was reported in the study, (c) the reliability coefficient reported was from their own local sample (reliability coefficients cited from previous studies were excluded), and (d) the study was available in English. Of all of the studies in the original search list, only one was written in a language other than English (Marinho-Araujo, 2014), and as no translated version could be found, it was left out of our analyses. Two more studies were left out of the analysis because they used a sample of school-aged children 6 to 12 years of age and a modified version of the PSI (Lavi & Slone, 2011, 2012). The remaining 95 studies which met these four qualifications were included in the current reliability generalization study. In the 95 studies, there were reliability coefficients reported for 101 separate samples.

In addition, separate reliability coefficients were often reported separately for each of the four dimensions in addition to the reliability coefficient for the entire PSI scale. For example, Zettler and Lang (2015) reported reliability coefficients for the entire PSI, network ability, interpersonal influence, social astuteness, and apparent sincerity, and found that aside from the political skill dimensions of interpersonal influence and social astuteness, all internal consistency estimates were acceptable in their sample (α > .70). For this reliability generalization study of the PSI, whenever reliability coefficients were reported for the entire PSI and its dimensions, all five of these reliability coefficients were recorded. However, each dimension and total score were separately analyzed, and as such, there is no violation of the independence assumption in the computation of the means reported in this article.

Following the protocol outlined by previous reliability generalization studies (e.g., Vacha-Haase, 1998, Vacha-Haase et al., 2000), six features were included as predictor variables that may predict a significant amount of variability in the reliabilities reported across the studies. The predictor variables included in this study were the following: (a) the length of the PSI scale (full 18-item PSI scale coded as 0 and the Ferris et al., 1999, six-item unidimensional PSI scale coded as 1), (b) language in which PSI was administered (English coded as 1 and others as 0), (c) response format (seven-option response scale coded as 1 and five-option response scale as 0), (d) the gender mix of the sample (0%-25% males coded as 0, 25%-50% as 1, 50%-75% as 2, and 75%-100% as 3), (e) if the study included student or employee samples (all-student samples coded 1 and employee samples coded 0), and (f) geographic location of the sample (samples in the United States coded as 0, samples in Germany coded as 1, and samples in China, Taiwan, and Japan coded as 2).

Upon further examination, there were 101 samples in the 95 articles that reported reliability coefficients of the total PSI or at least one of the four subdimensions. Of the 101 samples included in the analysis, 77 samples reported overall scale reliability coefficient for the full 18-item PSI scale, which was analyzed as the overall scale reliability estimate of the PSI. The remaining samples either reported reliability coefficients of the shortened six-item Ferris et al. PSI measure or only reported reliability estimates for one or more of the PSI dimensions from the 18-item PSI without reporting the overall scale reliability and were not included in analyses of the overall scale reliability. Of the 77 samples that reported reliability coefficients of the full 18-item PSI, there was a total of 15,987 participants (with sample sizes varying from 25 to 558). In addition to reporting the weighted Cronbach’s alpha reliability coefficient of the full 18-item PSI, the weighted alpha reliability coefficients of each of the four dimensions of the PSI were also reported independently using the samples that reported separate reliability coefficients for the four dimensions. More specifically, of the samples that reported reliability coefficient estimates for one or more of the PSI dimensions, 27 samples reported a reliability coefficient for the networking ability dimension, 27 reported for the interpersonal influence dimension, 25 for the social astuteness dimension, and 25 for the apparent sincerity dimension. We summarize in the tables whenever there were over three reliability confidents reported for each of analysis. Thus, the tables in our results do not report the reliability coefficients for the four PSI dimensions unless there was a sufficient number of samples in each of the groups being compared. The source PSI reliability reports are designated with asterisks in our references.

Results

The weighted coefficient alpha internal consistency reliability estimate for the Ferris et al. (2005) 18-item PSI and each of the four PSI subdimensions are summarized in Table 1. Coefficient alpha internal consistency reliability estimates for the Ferris et al. (2005) 18-item and the Ferris et al. (1999) six-item PSI are summarized in Table 2, and the coefficient alpha internal consistency reliability estimates by response option type are summarized in Table 3. Estimates of coefficient alpha for samples administered English and non-English versions of the PSI are summarized in Table 4, and estimates of coefficient alpha by geographic location are summarized in Table 5. Table 6 compares estimates of coefficient alphas for all-student samples with employee samples. Finally, estimates of coefficient alpha by sample gender composition are summarized in Table 7. It should be noted that the four dimensions of the scale are only present in Tables 1, 3, and 4, as we did not include results of the analysis that were based on less than three reliability estimates.

Table 1.

Internal Consistency Reliability Estimate of the 18-Item PSI and the Four Dimensions.

Dimension	n	k	M _wt	SD _wt	M _unwt	SD _unwt	M _sqwt	SD _sqwt	M _squnwt	SD _squnwt	80% CI
Overall PSI	15,987	77	.89	.044	.89	.057	.943	.210	.942	.239	[.88, .90]
Networking ability	6,524	27	.85	.056	.85	.055	.922	.237	.922	.235	[.84, .86]
Interpersonal influence	6,311	27	.80	.076	.81	.075	.896	.276	.901	.274	[.78, .82]
Social astuteness	5,993	25	.77	.080	.77	.084	.875	.283	.880	.290	[.75, .79]
Apparent sincerity	5,993	25	.72	.088	.70	.097	.849	.297	.838	.311	[.70, .74]

Note. PSI = Political Skill Inventory; k = number of reliabilities included in the meta-analysis; wt = sample size weighted; unwt = unweighted or frequency weighted; sqwt = square root of the estimates weighted; squnwt = square root of the estimates, unweighted; CI = confidence interval computed as M_wt ± SD_wt / √k (1.28).

Table 2.

Internal Consistency Reliability Estimate of the 18-Item PSI and Six-Item PSI.

PSI scale	n	k	M _wt	SD _wt	M _unwt	SD _unwt	M _sqwt	SD _sqwt	M _squnwt	SD_squnwt	80% CI
18-item PSI	15,987	77	.89	.044	.89	.057	.943	.210	.942	.239	[.88, .90]
Six-item PSI	2,123	11	.81	.072	.81	.074	.899	.272	.902	.272	[.78, .84]

Table 3.

Internal Consistency Reliability Estimate of the PSI by Response Scale Type.

Dimension	Response option	n	k	M _wt	SD _wt	M _unwt	SD _unwt	M _sqwt	SD _sqwt	M _squnwt	SD_squnwt	80% CI
Overall PSI	All samples	15,987	77	.89	.044	.89	.057	.943	.210	.942	.239	[.88, .90]
	5-point response option	5,445	28	.89	.037	.89	.039	.943	.197	.943	.197	[.88, .90]
	7-point response option	10,351	48	.89	.065	.88	.065	.943	.255	.938	.255	[.88, .90]
Networking ability	All samples	6,524	27	.85	.056	.85	.055	.922	.235	.922	.237	[.84, .86]
	5-point response option	1,116	5	.84	.055	.83	.055	.917	.235	.911	.235	[.81, .87]
	7-point response option	5,408	22	.85	.056	.84	.056	.922	.237	.917	.237	[.83, .87]
Interpersonal influence	All samples	6,311	27	.80	.076	.81	.075	.896	.276	.901	.274	[.78, .82]
	5-point response option	951	5	.82	.059	.81	.068	.906	.243	.900	.261	[.79, .85]
	7-point response option	5,360	22	.80	.078	.81	.078	.894	.280	.900	.280	[.78, .82]
Social astuteness	All samples	5,993	25	.77	.080	.77	.084	.875	.283	.880	.290	[.75, .79]
	5-point response option	951	5	.72	.087	.72	.097	.849	.295	.849	.311	[.67, .77]
	7-point response option	5,042	20	.77	.076	.79	.077	.875	.276	.889	.277	[.75, .79]
Apparent sincerity	All samples	5,993	25	.72	.088	.70	.097	.849	.297	.838	.311	[.70, .74]
	5-point response option	951	5	.69	.084	.68	.093	.831	.290	.825	.305	[.64, .74]
	7-point response option	5,042	20	.73	.087	.71	.099	.854	.295	.843	.315	[.71, .75]

Table 4.

Internal Consistency Reliability Estimate of the PSI by Language Version.

Dimension	Language	n	k	Mwt	SDwt	Munwt	SDunwt	Msqwt	SDsqwt	Msqunwt	SDsqunwt	80% CI
Overall PSI	All samples	15,987	77	.89	.044	.89	.057	.943	.210	.942	.239	[.88, .90]
	English	8,824	43	.90	.028	.89	.027	.949	.167	.943	.164	[.89, .91]
	Non-English	7,163	34	.88	.057	.88	.081	.938	.239	.938	.285	[.87, .89]
Networking ability	All samples	6,524	27	.85	.056	.85	.055	.922	.235	.922	.237	[.84, .86
	English	3,906	17	.88	.051	.87	.055	.938	.226	.933	.235	[.86, .90]
	Non-English	2,618	10	.81	.040	.82	.038	.90	.200	.906	.195	[.79, .83]
Interpersonal influence	All samples	6,311	27	.80	.076	.81	.075	.896	.274	.901	.276	[.78, .82]
	English	3,633	17	.85	.047	.85	.049	.922	.217	.922	.221	[.84, .86]
	Non-English	2,678	10	.74	.062	.75	.073	.860	.249	.866	.270	[.71, .77]
Social astuteness	All samples	5,993	25	.77	.080	.77	.084	.875	.283	.880	.290	[.75, .79]
	English	3,315	15	.82	.046	.82	.051	.906	.214	.906	.226	[.80, .84]
	Non-English	2,678	10	.70	.060	.71	.079	.837	.245	.843	.281	[.68, .72]
Apparent sincerity	All samples	5,993	25	.72	.088	.70	.097	.849	.297	.838	.311	[.70, .74]
	English	3,315	15	.75	.101	.72	.116	.866	.318	.849	.341	[.72, .78]
	Non-English	2,678	10	.68	.048	.67	.049	.825	.219	.819	.221	[.66, .70]

Table 5.

Internal Consistency Reliability Estimate of the Overall PSI by Geographic Location.

Geographic location	n	k	M _wt	SD _wt	M _unwt	SD _unwt	M _sqwt	SD _sqwt	M _squnwt	SD_squnwt	80% CI
All samples	15,987	77	.89	.044	.89	.057	.943	.210	.942	.239	[.88, .90]
United States	8,460	42	.90	.026	.90	.026	.949	.161	.949	.161	[.89, .91]
Germany	4,305	22	.87	.033	.88	.039	.933	.182	.938	.197	[.86, .88]
China, Thailand, and Japan	2,108	8	.89	.049	.89	.061	.943	.221	.943	.245	[.87, .91]

Table 6.

Internal Consistency Reliability Estimate of Overall PSI by Sample Population.

Sample type	n	k	M _wt	SD _wt	M _unwt	SD _unwt	M _sqwt	SD _sqwt	M _squnwt	SD_squnwt	80% CI
All samples	15,987	77	.89	.044	.89	.057	.943	.210	.942	.239	[.88, .90]
Non-student samples	13,411	67	.89	.048	.89	.061	.943	.219	.941	.247	[.88, .90]
All student samples	2,576	10	.89	.018	.89	.018	.943	.135	.943	.135	[.88, .90]

Table 7.

Internal Consistency Reliability Estimate of the PSI by Sample Gender Composition.

Gender composition	n	k	M _wt	SD _wt	M _unwt	SD _unwt	M _sqwt	SD _sqwt	M _squnwt	SD_squnwt	80% CI
All samples	15,987	77	.89	.044	.89	.057	.943	.210	.942	.239	[.88, .90]
0%-25% male	1,243	6	.88	.022	.88	.024	.917	.148	.926	.155	[.87, .89]
25%-50% male	6,676	30	.89	.053	.88	.076	.940	.230	.935	.276	[.88, .90]
50%-75% male	5,166	23	.89	.033	.89	.040	.941	.182	.942	.200	[.88, .90]
75%-100% male	992	5	.87	.037	.88	.039	.910	.192	.917	.197	[.85, .89]

The sample size weighted mean and standard deviation was computed for each distribution analyzed. We also computed the unweighted mean and standard deviation for each distribution of reliability coefficients. As the unweighted mean and standard deviation do not weigh the reliability estimates by sample size of each study included in the analysis and each reliability coefficient is weighted the same, the sample size weighted mean gives the best estimate of the mean reliability while the unweighted mean helps ensure that the results of the analysis are not skewed by a few large samples (Viswesvaran, Ones, & Schmidt, 1996). In addition, the mean and standard deviation of the square root of the reliabilities were also calculated as the mean of the square root of the reliabilities differ slightly from the square root of the mean of the reliabilities. Thus, for each of the distributions included in our analysis, we reported (a) sample size weighted mean reliability estimates, (b) sample size weighted standard deviation estimates, (c) unweighted mean reliability estimates, (d) unweighted standard deviation estimates, (e) sample size weighted square root of the reliability mean, (f) sample size weighted standard deviation of the square root of reliabilities, (g) unweighted square root of the reliability mean, (h) unweighted standard deviation of the square root of the reliabilities, and (i) the sample size weighted (80%) confidence interval for the sample size weighted mean reliability estimate. Confidence intervals were calculated by multiplying the ratio of the weighted sample standard deviation (SD_wt) to the square root of the number of samples (k) by 1.28 and adding and subtracting the resulting product from the sample weighted mean (M_wt) to create the upper and lower bound reliability estimates.

The results from Table 1 indicate that the sample weighted coefficient alpha for the 18-item PSI scale across the samples included in the meta-analysis that reported an alpha coefficient for the entire 18-item scale was .89 (k = 77, N = 15,987). Each of the four PSI subdimensions had sample weighted coefficient alpha estimates greater than .70, with an average coefficient alpha of .78 across dimensions (k = 26, N = 6,205). Specifically, the networking ability dimension of the PSI had the highest sample weighted reliability estimate (α = .85, k = 27, N = 6,524) of the four PSI subdimensions while the apparent sincerity dimension had the lowest (α = .72, k = 25, N = 5,993). Although the sample weighted coefficient alpha across the four dimensions is respectable, results indicated that the average sample weighted reliability estimate of the specific PSI dimensions is significantly lower than the average reliability estimate of the full 18-item PSI, t(98) = 10.31, p < .001.

The results from Table 2 indicate that the weighted reliability coefficient alpha for the shortened six-item PSI scale was .81 (k = 11, N = 2,123) and was comparable with the 18-item PSI, which had a weighted reliability alpha of .89 (k = 77, N = 15,987). However, an independent t test with the overall sample weighted reliability coefficient as the dependent variable and scale version as the independent variable revealed that the 18-item PSI had a significantly higher overall reliability coefficient than the six-item PSI, t = 3.80, p < .001. In addition, results indicated that the lower bound 80% confidence interval for the 18-item PSI (α = .88) was greater than the upper bound 80% confidence interval for the six-item PSI (α = .84), indicating that the 18-item PSI likely has a higher average reliability estimate across samples than the six-item PSI.

Results from Table 3 indicate that the sample weighted reliability coefficient alpha for samples that administered the full 18-item PSI using a five-option response option was .89 (k = 28, N = 5,445) and did not differ from that for a seven-option response option which was also .89 (k = 48, N = 10,351). Thus, utilizing a five-option response scale instead of a seven-option response scale did not impact the estimated reliability of the 18-item PSI measure. Results from Table 3 also indicate the weighted reliability coefficients for each of the four PSI dimensions across samples that used a five-option response scale as well as samples that utilized a seven-option response scale. The weighted reliability coefficient across the four PSI subdimensions for samples that utilized a five-option response scale was .77 while the weighted reliability coefficient across the four dimensions among samples that utilized a seven-option response scale was .79, indicating that like the full 18-item PSI scale, the weighted reliability estimate of the four PSI dimensions did not substantially differ between samples that utilized a five-option response scale and samples that utilized a seven-option response scale.

The results from Table 4 indicate that the sample weighted reliability coefficient alpha for the full 18-item PSI administered in English was .90 (k = 43, N = 8,824) and was not significantly different than the weighted reliability coefficient alpha for the translated non-English versions of the full 18-item PSI which was .88 (k = 34, N = 7,163), t = 1.22, p = .227. In addition, Table 5 also indicates the sample weighted reliability coefficient estimates for the four PSI scale dimensions for English and non-English samples. The averaged reliability estimates across the four individual scales of the PSI administered in English was noticeably greater (α = .83) than the averaged reliability estimates across the scales administered in non-English languages (α = .73). English-administered versions of the 18-item PSI had a significantly higher coefficient alpha of the social astuteness dimension than non-English versions of the scale, t = 4.34, p < .001. English versions of the 18-item PSI also had significantly higher coefficient alphas for the networking ability dimension of the scale than non-English versions of the PSI, t = 2.71, p = .012. Similarly, English versions of the 18-item PSI also had significantly higher coefficient alphas for the interpersonal influence dimension than non-English versions of the PSI, t = 4.21, p < .001.

The results from Table 5 summarize the reliability coefficient estimates of the full 18-item PSI by geographic location. Results from a one-way ANOVA that was conducted using the geographic location as the independent variable and the weighted alpha reliability coefficient as the dependent variable indicated that the geographic location did not significantly impact the overall reliability of the PSI, F = .83, p = .438. More specifically, samples that were administered the PSI in the United States had a sample weighted coefficient alpha of .90 (k = 42, N = 8,460), samples that were administered the PSI in Germany had a weighted coefficient alpha of .87 (k = 22, N = 4,305), and samples that were administered the PSI in China, Taiwan, and Japan had a weighted coefficient alpha of .89 (k = 8, N = 2,108). Thus, results indicated that the weighted alpha coefficient estimates of the full 18-item PSI were quite acceptable in each of the three geographic locations.

Results from Table 6 indicate that the sample weighted coefficient alpha estimate of the full 18-item PSI for all-student samples that were administered the PSI in academic settings was .89 (k = 10, N = 2,576), same as that in nonacademic settings .89 (k = 67, N = 13,411). Thus, the results indicated that the sample-weighted average reliability estimate of the 18-item PSI did not differ between all-student samples that were administered the PSI in academic settings and employee samples that were administered the PSI in nonacademic settings.

Finally, Table 7 summarizes the overall reliability of the 18-item PSI by the sample gender composition. Results from a one-way ANOVA with sample gender composition as the independent variable and the weighted reliability coefficient of the overall 18-item PSI scale as the dependent variable indicated that the gender composition of the sample did not significantly affect the weighted reliability estimate of the 18-item PSI scale, F = .203, p = .894. The similar weighted reliability coefficients between samples with different gender compositions seem to indicate that the gender composition of the sample does not significantly affect the overall estimated reliability of the 18-item PSI scale.

A six-factor ANOVA was subsequently conducted with PSI scale length, language administered, geographic location, response format, sample population type, and sample gender proportion as predictor variables and the sample weighted Cronbach’s alpha coefficient of the overall PSI scale as the dependent variable to test for the combined effects for the six predictors on the total PSI’s reliability. Results indicated that the six predictor variables significantly impacted the weighted reliability of the overall PSI scale, F = 2.68, p = .010. More specifically, the length of the PSI measure (six-item vs. the 18-item version) was significantly associated with the sample weighted Cronbach’s alpha coefficient after controlling for the language of administration, response scale type, sample population type, sample gender demographics, and geographic location, F = 13.27, p < .001. However, the language of the PSI, the response scale type, sample population type, the gender composition of the sample, and the geographic location that the PSI was administered in were each not significantly related to the overall reliability of the PSI scale.

Subsequently, four 5-factor ANOVA were conducted using the language of administration, geographic location, response format, sample population type, and sample gender proportion as predictor variables and the sample weighted reliability estimate of each of the four PSI subdimensions as the dependent variable to test for the combined effects of the five predictors on each of the four PSI subdimensions. Scale length was not included as a predictor in each in the analysis of the four PSI dimensions because only 18-item versions of the PSI reported independent reliability coefficients for each of the dimensions. For the networking ability dimension of the PSI, results indicated that the language of administration, geographic location, response option format, sample population type, and the sample gender composition accounted for a marginally significant amount of variance in the reliability coefficients, F = 13.37, p = .054. The language the PSI was administered in accounted for a significant amount of the variance in the estimated reliability coefficients in networking ability, F = 13.36, p = .004. Specifically, those who received an English version of the PSI (M = .87, SD = .055) had significantly higher reliability ratings compared with those who received a non-English version of the PSI (M = .82, SD = .038), t(25) = 2.71, p = .012. However, the other four predictors were not significantly related to the reliability of the networking ability dimension of the PSI.

For the interpersonal influence dimension of the PSI, results indicated that the five predictors accounted for a significant amount of variance in the reliability coefficients, F = 3.91, p = .029. The language the PSI was administered in was significantly associated with reliability estimates, F = 21.15, p = .001. Similar to the networking ability subdimension, samples that were administered English versions of the PSI (M = .85, SD = .049) had a significantly higher estimated reliability coefficient for the interpersonal influence subdimension compared with samples that were administered non-English versions of the PSI (M = .75, SD = .079), t(25) = 4.21 p < .001. However, the response scale type, sample population type, gender composition of the sample, and geographic location were each not significantly related to the reliability of the interpersonal influence dimension of the PSI.

For the social astuteness dimension of the PSI, the five predictors accounted for a significant amount of variance in the reliability coefficients, F = 4.48, p = .026. The language the PSI was administered in accounted for a significant amount of variance in the reliability estimates, F = 22.14, p = .002. Similar to the networking ability and interpersonal influence dimensions of the PSI, samples that were administered English versions of the PSI (M = .77, SD = .051) had a significantly higher estimated reliability coefficient for the social astuteness subdimension compared with samples that were administered non-English versions of the PSI (M = .61, SD = .079), t(23) = 4.34, p < .001.

Finally, for the apparent sincerity dimension of the PSI, the five predictors did not account for a significant amount of variation in the reliability estimates, F = .79, p = .617. Thus, unlike the other three dimensions, the language the PSI was administered in did not account for a significant amount of variance in the estimated reliability coefficients for the apparent sincerity subdimension of the PSI, F = .31, p = .595. Our results seemed to indicate that the language the PSI is administered in may significantly affect the estimated reliability of the networking ability, social astuteness, and interpersonal influence dimensions of the PSI but not the apparent sincerity subdimension. See Table 4 to compare the sample-weighted reliability estimates for each of the PSI dimensions administered to English samples and non-English samples.

Discussion

The primary objective of the present research was to utilize RG techniques to investigate the magnitude and variabilities of reliability estimates that were reported across existing studies using of the PSI developed by Ferris et al. (2005). Across the samples that used and reported reliability estimates for the full Ferris et al. (2005) PSI measure, the sample-weighted alpha coefficient of the 18-item PSI was .89 (SD = .044). Given that previous research has found that most psychometric tests have average alpha coefficients that fall within the range of .75 to .83 (Nunnally, 1978) and many professionals apply the general principle that requires a reliability of .70 before they utilize a measure, the average alpha of the full 18-item PSI obtained in the current RG study provides substantial evidence that supports the measurement integrity of the scale. In addition, each of the four subdimensions of the PSI demonstrated moderate to high sample weighted alpha reliability coefficients (α > .70). Thus, the reported reliability coefficients of the full Ferris et al. (2005) 18-item measure and each of the four PSI subscale dimensions seems to justify its implementation in organizational and research settings, although what is considered to be an acceptable mean reliability coefficient can be a somewhat subjective inference that is often ultimately determined by the context of a study (Hunter & Schmidt, 1990).

Contributions to Theory and Implications for Practice Regarding the Reliability of the PSI

Although the estimated reliability of the 18-item PSI and its dimensions seemed to be acceptable on average, results of the current study also indicated that the estimated reliability of the full 18-item PSI measure (Ferris et al., 2005) was significantly greater than the average weighted reliability estimate for the six-item unidimensional PSI scale (Ferris et al., 1999) as well as each of the four specific PSI subdimensions. Among the samples that were administered the shortened six-item version of the PSI (Ferris et al., 1999), the average weighted estimated alpha reliability coefficient was .81 (SD = .072) which was significantly lower than that of the 18-item PSI (M_wt = .89, SD = .044). In addition, and due to the six-item unidimensional Ferris et al. (1999) PSI’s lack of content validity evidence and failure to include item statements representing manifestations of the networking ability and apparent sincerity dimensions of the political skill construct, future researchers and practitioners should generally seek to administer the Ferris et al. (2005) 18-item version of the PSI in research samples, especially when gathering data to help guide organizational decisions.

Results of the current study also indicated that the estimated reliability of the full 18-item PSI measure (Ferris et al., 2005) was greater than the average weighted reliability estimate for each of the four PSI subdimensions (apparent sincerity, networking ability, interpersonal influence, and social astuteness). The length of the instrument used for measurement could be a possible explanation for the difference in reliability between the full PSI and specific PSI dimensions as the majority of the studies included in the analysis included a summing the four dimensions of into a single composite score. The relationship between reliability estimates and the number of items can be described as convex, with the reliability increasing rapidly as the number of items increases up to a certain point, but increases being relatively small after a certain amount of items. The measures included in the current analysis were long enough that the presence of additional items and the Spearman–Brown formula did not seem to make an appreciable difference in the overall reliability estimated for the four PSI subdimensions. An additional potential explanation for higher alpha coefficients for ratings of the overall PSI is related to the broad construct of political skill compared with the four PSI subdimensions that more narrowly defined facets of political skill. Indeed, previous research has found evidence that broader constructs are generally more reliably rated than more narrowly defined traits, supporting the possibility the overarching construct of political skill may be more easily identified across various types of behavioral manifestations than the four more narrowly defined dimensions of political skill (Ones & Viswesvaran, 1996; Viswesvaran & Ones, 2000). As a result, it seems possible that the findings of the current study reflect similar findings of previous research that the underlying general political skill construct may more reliability predict a wide range of organizational outcomes than any one of the four PSI subdimensions assessed independently. However, future research is needed to examine how overall scores of the full Ferris et al. (2005) 18-item PSI measure and scores of specific PSI subdimensions assessed independently are each related to various types of general and specific organizational outcomes.

Potential fluctuation of reliability coefficients was evident in the full 18-item PSI measure along three of the four dimensions. For the full Ferris et al. (2005) 18-item PSI measure, results indicated that sample type (e.g., all-student samples or employees), response style type (e.g., five-option or seven-option self-reported Likert-type scale), the gender makeup of the sample, the geographic location of the sample, and whether or not the PSI was administered in English or another language did not significantly affect the overall reliability of the PSI scale. However, as previously mentioned, results did indicate that samples that used the full 18-item PSI scale had significantly higher overall reliability coefficients than samples that used the shortened six-item version of the PSI scale.

Given the influence of score reliabilities on effect size (Hellman et al., 2006; Wilkinson & Task Force on Statistical Inference, 1999), the use of shorter PSI versions seems to clearly be a decision that should be made after careful consideration. Although shortened unidimensional scales of the PSI may seem to have acceptable psychometric properties, such scales were developed for the purposes of preliminary construct exploration of political skill, have suboptimal internal consistency, and often do not adequately measure each of the four political skill dimensions as well as the Ferris et al. (2005) full 18-item PSI measure (Ferris, Rogers, Blass, & Hochwarter, 2009). Thus, the Ferris et al. (2005) 18-item PSI should be preferred over the shortened Ferris et al. (1999) six-item unidimensional PSI scale.

Reliability of the Four PSI Dimensions

As far as the four dimensions of the PSI (networking ability, interpersonal influence, social astuteness, and apparent sincerity), results indicated that the language the measure is administered in may have an effect on the reliability coefficients of the networking ability, interpersonal influence, and social astuteness dimensions of the PSI. Specifically, our results indicated that the sample-weighted reliability coefficients of the networking ability, interpersonal influence, and social astuteness subdimensions of the PSI were significantly higher in samples that were administered English versions of the PSI compared with samples that were administered the PSI in a non-English language. It may be due to lack of standardization in the precise wording in the way the PSI is translated from English to other languages that affected the estimates. Previous research has found that translating certain items into different languages can pose a unique challenge to standardization as the verbatim translation of some words may have drastically different meanings in two languages and may result in the re-wording of some items in the translated version of the measure which may potentially affect item interpretation across languages (Bartram, 2008; Oswald, 2008). However, the average Cronbach’s alpha reliability estimate of the translated social astuteness subscale seemed to be adequate and the average overall reliability of translated non-English versions of the full PSI scale is not significantly different compared with the overall reliability of English versions of the PSI. However, future practitioners including only select dimensions of the PSI without the full 18-item Ferris et al. (2005) PSI, especially when using translated non-English versions of the subscales, should be concerned about the reliability of their assessments. The present study also found that the full 18-item PSI scale (Ferris et al., 2005) seems to have generally high internal consistency across student and employee samples, English and non-English samples, predominantly male and predominantly female samples, and when using both five- and seven-option response scales.

Limitations and Directions for Future Research

Although the average reliability estimates of the PSI measure and each dimension of the PSI were calculated and certain moderating variables that could explain the variability were studied, there were certain limitations regarding the lack of reliability and sample information that may have hindered the scope of this reliability generalization study. As previously mentioned, most of the samples included in this study did not provide estimates of reliability for each of the PSI dimensions (social astuteness, networking ability, interpersonal influence, and apparent sincerity) in addition to the overall PSI scale. The majority of the 101 samples included in the analysis only indicated the Cronbach’s alpha of the full PSI measure without reporting the reliability coefficients for the four PSI subdimensions, which limited the number of samples that were included in the analysis of the reliability of the four PSI dimensions. For example, there were only 27 samples in the articles we included that reported a reliability for networking ability, 27 that reported a reliability coefficient for interpersonal influence, 25 that reported a reliability coefficient for social astuteness, and 25 samples that reported a reliability coefficient for apparent sincerity.

In addition, although 101 samples were included in the final analysis of this reliability generalization of the PSI scale, many of the samples were missing information relating to gender and race that limited the use of gender and race as potential predictor variables in the analysis. Although gender was included as a predictor variable in the analysis by coding the percentage of males in the sample quartiles, previous reliability generalization studies have measured gender by coding all male or all female samples as 0 and all “mixed” samples with male and female participants as 1 (e.g., Vacha-Haase, 1998; Vacha-Haase et al., 2000). As none of the 101 samples in the current reliability generalization indicated they had exclusively all male or all female samples, samples were broken into four groups by their percentage of males (0%-25% male, 25%-50% male, 50%-75% male, and 75%-100% male) which did not allow for absolute distinctions to be made following the protocol of previous reliability generalization research regarding gender which have compared all male, all female, and mixed male and female samples (Vacha-Haase, 1998; Vacha-Haase et al., 2000).

In addition to the absence of all male and all female samples, racial and demographic information in the samples were not included in this reliability generalization of the PSI scale as potential moderator variables as racial and ethnic demographic information was provided in only 30 (28.85%) of the 101 possible samples that were included in our analysis. Future research on the PSI scale should seek to examine how the effects of gender and race influence the overall reliability of the PSI scale, especially as previous studies have found that subdimensions of political skill (e.g., networking ability) may have different inferences and interpreted differently depending on racial and ethnic group membership (Blass, Brouer, Perrewé, & Ferris, 2007). However, the results of the current study indicated that there were no significant overall differences in the weighted reliability of the full 18-item PSI scale based on language or geographic location, providing evidence that the measurement error of the PSI may likely not significantly differ based on ethnic group membership.

Although none of the potential sources of measurement error relating to sample differences individually significantly impacted the overall reliability of the PSI measure, the results of the current reliability generalization study of the PSI did seem to generally indicate that more homogeneous or demographically similar samples consisting of participants that were taken from relatively homogeneous populations generally tended to have a slightly lower degree of variability regarding the PSI’s interitem reliability compared with more heterogeneous samples that consist of individual participants taken from a more demographically diverse population. As a result, a general implication for future practitioners and researchers implementing the PSI is to employ data collection strategies that maximize the homogeneity of the sample population whenever possible such as sampling employee participants working at the same organization or with the same occupation type to mitigate any potential differences in measurement error in the sample that would generally be more likely to occur from utilizing data collection strategies that yield more heterogeneous samples. Future studies examining potential differences in the overall reliability estimates of the PSI measure between samples based on occupation type, organizational homogeneity, and other potential sources of systematic differences in measurement error between samples are needed to clarify this relationship.

Various moderators effecting the political skill job performance relationship have been identified by researchers, indicating that high levels of political skill are only associated with highest degree of performance without the presence of certain conditions (Zettler & Lang, 2015). For example, high levels of political skill are not associated with the highest levels of performance when individual conscientiousness is high, there are high levels of procedural and distributive justice in an organization, or organizational politics are high (Andrews, Kacmar, & Harris, 2009; Blickle, Weber, & Oerder, 2008). In addition, previous moderation analyses have found that the highest levels of political skill do not always yield the highest level of performance, thus the meta-theoretical principle of the TMGT effect “too much of a good thing is bad” may apply to the political skill construct (Andrews et al., 2009). Just as the TMGT effect applies to other nonmonotonic relations, it also explains why high political skill is associated with lower levels of performance than medium levels of political skill (Pierce & Aguinis 2013; Zettler & Lang, 2015). For example, extremely high levels of political skill may be associated with increased time networking rather than working, and expending effort for self-promoting purposes to achieve individual goals at the expense of the goals of the organization (Zettler & Lang, 2015). Given the PSI has a curvilinear relationship with job performance, it also seems possible that the reliability of the PSI may also fluctuate between those with different scores on the measure, indicating heteroskedasticity. Thus, it may be that the reliability of the PSI may be more or less reliable for those that report having high political skill compared with those that report having little political skill. Thus, future research could also seek to clarify if the reliability of the PSI is heteroskedastic and fluctuates across different scores on the measure.

Conclusion

In conclusion, the results of the present reliability generalization study provide novel insights to the organizational politicking literature by indicating that the (Ferris et al., 2005) full 18-item PSI scale has an acceptable weighted Cronbach’s alpha reliability coefficient across all of the samples that were identified in existing literature, a finding that adds to the body of research examining the measurement integrity of the PSI. The examination of the reliability of the PSI and its dimensions are vital to the construction of generalizable theories of political skill in future research. In practice and in research, the accurate implementation and administrative use of political skill assessments depend on the reliability of political skill measures. Thus, the results of the current meta-analysis offer valuable and novel insights of the psychometric characteristics of the PSI that can be utilized by future researchers and practitioners.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) received no financial support for the research,authorship,and/or publication of this article.

Author Biographies

Ryan K. Jacobson,is currently an Industrial-Organizational psychology PhD candidate conducting dissertation research at Florida International University,having attained his MS in Industrial/Organizational psychology in 2016. His primary research interests focus on how organizational policies influence reporting behavior of sexual harassment in the workplace,using an intersectional perspective to examine perceptions of candidate competence and hireability as a function of the race and gender of the candidate,examining the utility of political skill measures as predictors of job performance,and examining how specific organizational outcomes are related to employee bystander perceptions of counterproductive behavior in the workplace.

Chockalingam Viswesvaran,is the director of the Industrial/Organizational psychology program and a tenured professor in the psychology department at Florida International University. Viswesvaran is a well-known and frequently cited researcher in the field of Industrial/Organizational psychology and serves on the editorial boards of Journal of Applied Psychology,Educational and Psychological Measurement,Journal of Organizational Behavior,Journal of Personnel Psychology,Journal of Business and Psychology and is the Editor of the International Journal of Selection and Assessment. Dr. Viswesvaran is also a fellow of the Society for Industrial and Organizational Psychology (SIOP) as well as a Fellow of Division 5 (measurement) and Division 14 (Industrial- organizational psychology) of the American Psychological Association.

References

Ahearn

K. K.

Ferris

G. R.

Hochwarter

W. A.

Douglas

Ammeter

A. P.

(2004). Leader political skill and team performance. Journal of Management, 30, 309-327.

Andrews

M. C.

Kacmar

K. M.

Harris

K. J.

(2009). Got political skill? The impact of justice on the importance of political skill for job performance. Journal of Applied Psychology, 94, 1427-1437. doi:10.1037/a0017154

*Baer

(2012). Putting creativity to work: The implementation of creative ideas in organizations. Academy of Management Journal, 55, 1102-1119.

*Banister

C. M.

Meriac

J. P.

(2015). Political skill and work attitudes: A comparison of multiple social effectiveness constructs. The Journal of Psychology: Interdisciplinary and Applied, 149, 775-795.

Bartram

(2008). Global norms: Towards some guidelines for aggregating personality norms across countries. International Journal of Testing, 8, 315-333.

Bendig

A. W.

(1954). Reliability and the number of rating scale categories. Journal of Applied Psychology, 38, 38-40.

Bing

M. N.

Davison

H. K.

Minor

Novicevic

Frink

D. D.

(2011). The prediction of task and contextual performance by political skill: A meta-analysis and moderator test. Journal of Vocational Behavior, 79, 563-577.

*Blass

F. R.

Brouer

R. L.

Perrewé

P. L.

Ferris

G. R.

(2007). Politics understanding and networking ability as a function of mentoring: The roles of gender and race. Journal of Leadership & Organizational Studies, 14, 93-105.

*Blickle

Diekmann

Schneider

P. B.

Kalthöfer

Summers

J. K.

(2012). When modesty wins: Impression management through modesty, political skill, and career success—A two-study investigation. European Journal of Work and Organizational Psychology, 21, 899-922. doi:10.1080/1359432X.2011.603900

10.

*Blickle

Ferris

G. R.

Munyon

T. P.

Momm

Zettler

Schneider

P. B.

Buckley

M. R.

(2011). A multi-source, multi-study investigation of job performance prediction by political skill. Applied Psychology: An International Review, 60, 449-474.

11.

*Blickle

John

Ferris

G. R.

Momm

Liu

Haag

Oerder

(2012). Fit of political skill to the work context: A two-study investigation. Applied Psychology: An International Review, 61, 295-322.

12.

*Blickle

Kane-Frieder

Oerder

Wihler

von Below

Schütte

Ferris

G. R.

(2013). Leader behaviors as mediators of the leader characteristics-follower satisfaction relationship. Group & Organization Management, 38, 601-629. doi:10.1177/1059601113503183

13.

*Blickle

Kramer

Schneider

P. B.

Meurs

J. A.

Ferris

G. R.

Mierke

. . . Momm

T. D.

(2011). Role of political skill in job performance prediction beyond general mental ability and personality in cross-sectional and predictive studies. Journal of Applied Social Psychology, 41, 488-514. doi:10.1111/j.1559-1816.2010.00723

14.

*Blickle

Kramer

Zettler

Momm

Summers

J. K.

Munyon

T. P.

Ferris

G. R.

(2009). Job demands as a moderator of the political skill-job performance relationship. The Career Development International, 14, 333-350.

15.

*Blickle

Meurs

J. A.

Wihler

Ewen

Plies

Günther

(2013). The interactive effects of conscientiousness, openness to experience, and political skill on job performance in complex jobs: The importance of context. Journal of Organizational Behavior, 34, 1145-1164. doi:10.1002/job.1843

16.

*Blickle

Momm

T. S.

Kramer

Mierke

Liu

Ferris

G. R.

(2009). Construct and criterion-related validation of a measure of emotional reasoning skills: A two-study investigation. International Journal of Selection and Assessment, 17, 101-118.

17.

*Blickle

Oerder

Summers

(2010). The impact of political skill on career success of employees’ representatives. Journal of Vocational Behavior, 77, 383-390. doi:10.1016/j.jvb.2010.05.007

18.

*Blickle

von Below

Johannen

(2011). Self-ratings of political skill in job application: A within- and between-subjects field experiment. International Journal of Selection and Assessment, 19, 98-104.

19.

*Blickle

Weber

Oerder

(2008). Personality, political skill, and job performance. Journal of Vocational Behavior, 72, 377-387.

20.

*Blickle

Wendel

Ferris

G. R.

(2010). Political skill as moderator of personality—Job performance relationships in socioanalytic theory: Test of the getting ahead motive in automobile sales. Journal of Vocational Behavior, 76, 326-335.

21.

*Brouer

R. L.

Badaway

R. L.

Gallagher

V. C.

Haber

J. A.

(2015). Political skill dimensionality and impression management choice and effective use. Journal of Business and Psychology, 30, 217-233.

22.

*Brouer

R. L.

Douglas

Treadway

D. C.

Ferris

G. R.

(2013). Leader political skill, relationship quality, and leadership effectiveness: A two-study model test and constructive replication. Journal of Leadership & Organizational Studies, 20, 185-198.

23.

*Brouer

R. L.

Duke

Treadway

D. C.

Ferris

G. R.

(2009). The moderating effect of political skill on the demographic dissimilarity—Leader-member exchange quality relationship. Leadership Quarterly, 20, 61-69.

24.

*Chopin

S. M.

Danish

S. J.

Seers

Hook

J. N.

(2012). Effects of mentoring on the development of leadership self-efficacy and political skill. Journal of Leadership & Organizational Studies, 6(3), 17-32.

25.

Cronbach

L. J.

(1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297-334.

26.

*Cullen

K. L.

Fan

Liu

(2014). Employee popularity mediates the relationship between political skill and workplace interpersonal mistreatment. Journal of Management, 40, 1760-1778.

27.

*Dahling

J. J.

Whitaker

B. G.

Levy

P. E.

(2009). The development and validation of a new machiavellianism scale. Journal of Management, 35, 219-257.

28.

Daouk

Rust

McDowall

(2005). Testing across languages and cultures: Challenges for the development and administration of tests in the internet era. Selection and Development Review, 21(4), 11-11.

29.

Dimitrov

D. M.

(2002). Reliability: Arguments for multiple perspectives and potential problems with generalization across studies. Educational and Psychological Measurement, 62, 783-801.

30.

*Ewen

Wihler

Blickle

Oerder

Ellen

B. P.

Douglas

Ferris

G. R.

(2013). Further specification of the leader political skill–leadership effectiveness relationships: Transformational and transactional leader behavior as mediators. Leadership Quarterly, 24, 516-533.

31.

*Ewen

Wihler

Frieder

R. E.

Blickle

Hogan

Ferris

G. R.

(2014). Leader advancement motive, political skill, leader behavior, and effectiveness: A moderated mediation extension of socioanalytic theory. Human Performance, 27, 373-392.

32.

*Fan

Litchfield

R. C.

Islam

Weiner

Alexander

Liu

Kulviwat

(2013). Workplace social self-efficacy: Concept, measure, and initial validity evidence. Journal of Career Assessment, 21, 91-110.

33.

Ferris

G. R.

Berkson

H. M.

Kaplan

D. M.

Gilmore

D. C.

Buckley

M. R.

Hochwarter

W. A.

Witt

L. A.

(1999). Development and initial validation of the political skill inventory. Paper presented at the 59th annual national meeting of the Academy of Management, Chicago.

34.

*Ferris

G. R.

Rogers

L. M.

Blass

F. R.

Hochwarter

W. A.

(2009). Interaction of job-limiting pain and political skill on job satisfaction and organizational citizenship behavior. Journal of Managerial Psychology, 24, 584-608.

35.

Ferris

G. R.

Treadway

D. C.

Brouer

R. L.

Munyon

T. P.

(2012). Political skill in the organizational sciences. In Ferris

G. R.

Treadway

D. C.

(Eds.), Politics in organizations: Theory and research implications (pp. 487-528). New York, NY: Routledge.

36.

*Ferris

G. R.

Treadway

D. C.

Kolodinsky

R. W.

Hochwarter

W. A.

Kacmar

C. J.

Douglas

Frink

D. D.

(2005). Development and validation of the political skill inventory. Journal of Management, 31, 126-152.

37.

Ferris

G. R.

Treadway

D. C.

Perrewé

P. L.

Brouer

R. L.

Douglas

Lux

(2007). Political skill in organizations. Journal of Management, 33, 290-320.

38.

*Gallagher

V. C.

Laird

M. D.

(2008). The combined effect of political skill and political decision making on job satisfaction. Journal of Applied Social Psychology, 38, 2336-2360.doi:10.1111/j.1559-1816.2008.00394

39.

*Geertshuis

S. A.

Cooper-Thomas

H. D.

Price

(2013). Appraisals of self and situation as determinants of upward influencing tactics. Journal of Applied Social Psychology, 43, 1563-1577. doi:10.1111/jasp.12106

40.

*Gentry

W. A.

Gilmore

D. C.

Shuffler

M. L.

Leslie

J. B.

(2012). Political skill as an indicator of promotability among multiple rater sources. Journal of Organizational Behavior, 33, 89-104.

41.

*Gentry

W. A.

Leslie

J. B.

Gilmore

D. C.

Ellen

B. P.

Ferris

G. R.

Treadway

D. C.

(2013). Personality and political skill as distal and proximal predictors of leadership evaluations. The Career Development International, 18, 569-588.

42.

Gronlund

N. E.

Linn

R. L.

(1990). Measurement and evaluation in teaching (6th ed.). New York, NY: Macmillan.

43.

*Guidice

R. M.

Mero

N. P.

(2012). Hedging their bets: A longitudinal study of the trade-offs between task and contextual performance in a sales organization. Journal of Personal Selling & Sales Management, 32, 451-471.

44.

Hellman

C. M.

Fuqua

D. R.

Worley

(2006). A reliability generalization study on the Survey of Perceived Organizational Support: The effects of mean age and number of items on score reliability. Educational and Psychological Measurement, 66, 631-642.

45.

*Hochwarter

W. A.

Ferris

G. R.

Gavin

M. B.

Perrewé

P. L.

Hall

A. T.

Frink

D. D.

(2007). Political skill as neutralizer of felt accountability–job tension effects on job performance ratings: A longitudinal investigation. Organizational Behavior and Human Decision Processes, 102, 226-239.

46.

*Hochwarter

W. A.

Summers

J. K.

Thompson

K. W.

Perrewé

P. L.

Ferris

G. R.

(2010). Strain reactions to perceived entitlement behavior by others as a contextual stressor: Moderating role of political skill in three samples. Journal of Occupational Health Psychology, 15, 388-398.

47.

*Huang

Frideger

Pearce

J. L.

(2013). Political skill: Explaining the effects of nonnative accent on managerial hiring and entrepreneurial investment decisions. Journal of Applied Psychology, 98, 1005-1017.

48.

Hunter

J. E.

Schmidt

F. L.

(1990). Methods of meta-analysis: Correcting for error and bias in research findings. Newbury Park, CA: SAGE.

49.

*Jawahar

I. M.

Meurs

J. A.

Ferris

G. R.

Hochwarter

W. A.

(2008). Self-efficacy and political skill as comparative predictors of task and contextual performance: A two-study constructive replication. Human Performance, 21, 138-157.

50.

*Kimura

(2013). The moderating effects of political skill and leader–member exchange on the relationship between organizational politics and affective commitment. Journal of Business Ethics, 116, 587-599.

51.

Kipnis

Schmidt

S. M.

Wilkinson

(1980). Intraorgani-zational influence tactics: Explorations in getting one’s way. Journal of applied psychology, 65(4), 440.

52.

*Kisamore

J. L.

Jawahar

I. M.

Liguori

E. W.

Mharapara

T. L.

Stone

T. H.

(2010). Conflict and abusive workplace behaviors: The moderating effects of social competencies. The Career Development International, 15, 583-600.

53.

*Kolodinsky

R. W.

Hochwarter

W. A.

Ferris

G. R.

(2004). Nonlinearity in the relationship between political skill and work outcomes: Convergent evidence from three studies. Journal of Vocational Behavior, 65, 294-308.

54.

*Kolodinsky

R. W.

Treadway

D. C.

Ferris

G. R.

(2007). Political skill and influence effectiveness: Testing portions of an expanded Ferris and Judge (1991) model. Human Relations, 60, 1747-1777.

55.

*Laird

M. D.

Zboja

J. J.

Ferris

G. R.

(2012). Partial mediation of the political skill–reputation relationship. The Career Development International, 17, 557-582.

56.

*Laird

M. D.

Zboja

J. J.

Martinez

A. D.

Ferris

G. R.

(2013). Performance and political skill in personal reputation assessments. Journal of Managerial Psychology, 28, 661-676.

57.

Lane

G. G.

White

A. E.

Henson

R. K.

(2002). Expanding reliability generalization methods with KR-21 estimates: An RG study of the Coopersmith Self-Esteem Inventory. Educational and Psychological Measurement, 62, 685-711.

58.

Lavi

Slone

(2011). Resilience and political violence: A cross-cultural study of moderating effects among Jewish and Arab-Israeli youth. Youth Society, 43, 845-872.

59.

Lavi

Slone

(2012). Parental practices and political violence: The protective role of parental warmth and authority-control in Jewish and Arab Israeli children. American Journal of Orthopsychiatry, 4, 550-561.

60.

*Liu

Ferris

G. R.

Weitz

B. A.

Perrewé

P. L.

(2014). When ingratiation backfires: The role of political skill in the ingratiation–internship performance relationship. Academy of Management Learning & Education, 13, 569-586.

61.

*Liu

Ferris

G. R.

Zinko

Perrewé

P. L.

Weitz

(2007). Dispositional antecedents and outcomes of political skill in organizations: A four-study investigation with convergence. Journal of Vocational Behavior, 7, 146-165.

62.

Lord

F. M.

(1959). Tests of the same length do have the same standard error of measurement. Educational and Psychological Measurement, 19, 233-239.

63.

*Lvina

Johns

Treadway

D. C.

Blickle

Liu

Ferris

G. R.

(2012). Measure invariance of the Political Skill Inventory (PSI) across five cultures. International Journal of Cross Cultural Management, 12, 171-191.

64.

Marinho-Araujo

C. M.

(2014). Psicologia Escolar na educação superior: desafios e potencialidades. In Guzzo

R. S. L.

(Org.), Psicologia Escolar: desafios e bastidores na educação pública (pp. 219-239). Campinas: Alínea.

65.

Mattel

M. S.

Jacoby

(1971). Is there an optimal number of alternatives for Likert scale items? Study 1: Reliability and validity. Educational and Psychological Measurements, 31, 657-674.

66.

Mattel

M. S.

Jacoby

(1972). Is there an optimal number of alternatives for Likert scale items? Effects of testing time and scale properties. Journal of Applied Psychology, 56, 506-509.

67.

*Meisler

(2014). Exploring emotional intelligence, political skill, and job satisfaction. Employee Relations, 36, 280-293.

68.

*Meurs

J. A.

Perrewé

P. L.

Ferris

G. R.

(2011). Political skill as moderator of the trait sincerity-task performance relationship: A socioanalytic, narrow trait perspective. Human Performance, 24, 119-134.

69.

*Miller Burke

Attridge

. (2011). Pathways to career and leadership success: Part 1—A psychosocial profile of $100k professionals. Journal of Workplace Behavioral Health, 26, 175-206.

70.

*Momm

Blickle

Liu

(2010). Political skill and emotional cue learning. Personality and Individual Differences, 49, 396-401.

71.

*Momm

Blickle

Liu

Wihler

Kholin

Menges

J. I.

(2015). It pays to have an eye for emotions: Emotion recognition ability indirectly predicts annual income. Journal of Organizational Behavior, 36, 147-163.

72.

*Moss

J. A.

Barbuto

J. E.

Jr. (2010). Testing the relationship between interpersonal political skills, altruism, leadership success and effectiveness: A multilevel model. Journal of Behavioral and Applied Management, 11, 155-174.

73.

Mushquash

C. J.

Bova

(2007). Cross-cultural measurement and assessment issues. Journal on Developmental Disabilities, 12, 53-66.

74.

Nunnally

J. C.

(1978). Assessment of reliability. In Psychometric theory (2nd ed.). New York, NY: McGraw-Hill, pp. 244-245.

75.

*Oerder

Blickle

Summers

J. K.

(2014). How work context and age shape political skill. Journal of Managerial Psychology, 29, 582-599.

76.

Ones

D. S.

Viswesvaran

(1996). Bandwidth-fidelity dilemma in personality measurement for personnel selection. Journal of Organizational Behavior, 36, 147-163.

77.

Oswald

F. L.

(2008). Global personality norms: Multicultural, multinational, and managerial. International Journal of Testing, 8, 400-408.

78.

*Park

H. I.

O’Rourke

O’Brien

K. E.

(2014). Extending conservation of resources theory: The interaction between emotional labor and interpersonal influence. International Journal of Stress Management, 21, 384-405.

79.

Pfeffer

(1992). Managing with power: Politics and influence in organizations. Boston, MA: Harvard Business School Press.

80.

Pierce

J. R.

Aguinis

(2013). The too-much-of-a-good-thing effect in management. Journal of Management, 39, 313-338.

81.

*Randel

(2011). Need for power, collective identity, and political skill: An investigation in Taiwan. The Journal of Social Psychology, 151, 395-398.

82.

*Rosen

C. C.

Levy

P. E.

(2013). Stresses, swaps, and skill: An investigation of the psychological dynamics that relate work politics to employee performance. Human Performance, 26, 44-65.

83.

*Rosen

C. C.

Slater

D. J.

Chang

Johnson

R. E.

(2013). Let’s make a deal: Development and validation of the ex post I-deals scale. Journal of Management, 39, 709-742.

84.

*Roulin

Bangerter

Levashina

(2014). Interviewers’ perceptions of impression management in employment interviews. Journal of Managerial Psychology, 29, 141-163.

85.

Schmidt

F. L.

Hunter

J. E.

(1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological Bulletin, 124, 262-274.

86.

*Semadar

Robins

Ferris

G. R.

(2006). Comparing the validity of multiple social effectiveness constructs in the prediction of managerial job performance. Journal of Organizational Behavior, 27, 443-461.

87.

*Shaughnessy

B. A.

Treadway

D. C.

Breland

J. A.

Williams

L. V.

Brouer

R. L.

(2011). Influence and promotability: The importance of female political skill. Journal of Managerial Psychology, 26, 584-603.

88.

*Shi

Chen

(2012). Psychometric properties of a Chinese translation of the political skill inventory. Psychological Reports, 110, 233-246.

89.

*Shi

Chen

Zhou

(2011). Testing differential mediation effects of sub-dimensions of political skills in linking proactive personality to employee performance. Journal of Business and Psychology, 26, 359-369.

90.

*Silvester

Wyatt

Randall

(2014). Politician personality, machiavellianism, and political skill as predictors of performance ratings in political roles. Journal of Occupational and Organizational Psychology, 87, 258-279.

91.

*Snell

S. J.

Tonidandel

Braddy

P. W.

Fleenor

J. W.

(2014). The relative importance of political skill dimensions for predicting managerial effectiveness. European Journal of Work and Organizational Psychology, 23, 915-929.

92.

*Solga

Betz

Düsenberg

Ostermann

(2015). Political skill in job negotiations: A two-study constructive replication. International Journal of Conflict Management, 26, 2-24.

93.

*Sun

Pan

Chow

I. H. S.

(2014). The role of supervisor political skill in mentoring: Dual motivational perspectives. Journal of Organizational Behavior, 35, 213-233.

94.

Thompson

Vacha-Haase

(2000). Psychometrics is datametrics: The test is not reliable. Educational and Psychological Measurement, 60, 174-195.

95.

*Treadway

D. C.

Adams

Hanes

T. J.

Perrewé

P. L.

Magnusen

M. J.

Ferris

G. R.

(2014). The roles of recruiter political skill and performance resource leveraging in NCAA football recruitment effectiveness. Journal of Management, 40, 1607-1626.

96.

*Treadway

D. C.

Breland

J. W.

Williams

L. M.

Cho

Yang

Ferris

G. R.

(2013). Social influence and interpersonal power in organizations: Roles of performance and political skill in two studies. Journal of Management, 39, 1529-1553.

97.

*Treadway

D. C.

Ferris

G. R.

Duke

A. B.

Adams

G. L.

Thatcher

J. B.

(2007). The moderating role of subordinate political skill on supervisors’ impressions of subordinate ingratiation and ratings of subordinate interpersonal facilitation. Journal of Applied Psychology, 92, 848-855.

98.

Vacha-Haase

(1998). Reliability generalization: Exploring variance in measurement error affecting score reliability across studies. Educational and Psychological Measurement, 58, 6-20.

99.

Vacha-Haase

Henson

R. K.

Caruso

J. C.

(2002). Reliability generalization: Moving toward improved understanding and use of score reliability. Educational and Psychological Measurement, 62, 562-569.

100.

Vacha-Haase

Kogan

L. R.

Thompson

(2000). Sample compositions and variabilities in published studies versus those in test manuals: Validity of score reliability inductions. Educational and Psychological Measurement, 60, 509-522.

101.

*Vigoda-Gadot

Meisler

(2010). Emotions in management and the management of emotions: The impact of emotional intelligence and organizational politics on public sector employees. Public Administration Review, 70, 72-86.

102.

Viswesvaran

Ones

D. S.

(2000). Measurement error in “big five factors” personality assessment: Reliability generalization across studies and measures. Educational and Psychological Measurement, 60, 224-235.

103.

Viswesvaran

Ones

D. S.

Schmidt

F. L.

(1996). Comparative analysis of the reliability of job performance ratings. Journal of Applied Psychology, 81, 557-574.

104.

*Voss

C. M.

Brettel

(2013). Conditions of departmental power: A strategic contingency exploration of marketing’s customer-connecting role. Journal of Strategic Marketing, 21, 160-178.

105.

*Wei

Chiang

F. T.

(2012). Developing and utilizing network resources: Roles of political skill. Journal of Management Studies, 49, 381-402.

106.

*Whitaker

B. G.

Dahling

J. J.

(2013). The influence of autonomy and supervisor political skill on the use and consequences of peer intimidation in organizations. Human Performance, 26, 353-373.

107.

Wilkinson, L., & Task Force on Statistical Inference. (1999). Statistical methods in psychology journals: Guidelines and explanations. American Psychologist, 54, 594-604.

108.

*Wu

Kwan

H. K.

Wei

Liu

(2013). Ingratiation in the workplace: The role of subordinate and supervisor political skill. Journal of Management Studies, 50, 991-1017.

109.

*Wu

Yim

F. H.

Kwan

H. K.

Zhang

(2012). Coping with workplace ostracism: The roles of ingratiation and political skill in employee psychological distress. Journal of Management Studies, 49, 178-199.

110.

Yin

Fan

(2000). Assessing the reliability of Beck Depression Inventory Scores: Reliability generalization across studies. Educational and Psychological Measurement, 60, 201-223.

111.

*Zettler

Lang

J. W. B.

(2015). Employees’ political skill and job performance: An inverted U-shaped relation? Applied Psychology: An International Review, 64, 541-577.

112.

*Zhao

Peng

Sheard

(2013). Workplace ostracism and hospitality employees’ counterproductive work behaviors: The joint moderating effects of proactive personality and political skill. International Journal of Hospitality Management, 33, 219-227.

113.

*Zhou

Z. E.

Yang

L. Q.

Spector

P. E.

(2015). Political skill: A proactive inhibitor of workplace aggression exposure and an active buffer of the aggression-strain relationship. Journal of Occupational Health Psychology, 20, 405-419.

114.

*Zinko

(2013). A continued examination of the inverse relationship between political skill and strain reactions: Exploring reputation as a mediating factor. Journal of Applied Social Psychology, 43, 1750-1759.

A Reliability Generalization Study of the Political Skill Inventory

Abstract

Keywords

Reliability

Reliability Generalization

Factors That Influence Reliability

Number of Items

Number of Response Options

Language

Geographic Location

Sample Population

Sample Gender Distribution

Method

Results

Discussion

Contributions to Theory and Implications for Practice Regarding the Reliability of the PSI

Reliability of the Four PSI Dimensions

Limitations and Directions for Future Research

Conclusion

Footnotes

Declaration of Conflicting Interests

Funding

Author Biographies

References