pexels-anni-roenkae-4175070

7/31 – Toxic Leadership and Followership Typologies: A Partial Replication Study with Scale Refinement

R. Mark Bell

R. Mark Bell 

Toxic leadership and followership remain emerging sectors of inquiry within the overall field of leadership study. As a result of the relative paucity of research on both toxic leadership and followership, there remains a continued need for scale refinement in order to ensure valid and reliable scales are available to measure the constructs. The present study is a contribution in this vein specifically focused on refinement of Schmidt’s (2014) Toxic Leadership Scale (TLS) and Kelley’s (1992) Followership Styles Questionnaire (FSQ). A review of the relevant toxic leadership and followership literature is provided with a focus on the initial development of the TLS and the FSQ followed by a review of the use and validation of each instrument in other published research. The study then shifts to a partial replication of Bell’s (2017) study where both the TLS and FSQ were used. As such, Bell’s initial study is described as Phase I, and the partial replication work and results are described as Phase II. The same hypotheses are retested in Phase II but with a larger sample that only varied from Phase I in level of education. The differences in the hypotheses’ results from Phase I and II are described. Additionally, Phase II includes the scale refinement work on the TLS and FSQ including reliability analysis and factor analysis that produced new factor findings on both instruments. A final discussion of these results is provided including the presentation of the two instruments in revised fashion being noted as the TLS-1 and the FSQ-3. Several research recommendations are subsequently addressed. 

Literature Review

Various facets of the toxic leadership and followership literature are presented, but this review is not intended to be comprehensive in nature. A more thorough review of both theories is presented in Bell’s (2017) research. The foundational aspects of toxic leadership theory are presented with a special focus on the development of Schmidt’s (2014) TLS instrument and its use in other studies. Similarly, the basic tenets of followership are presented with specific consideration of the development of Kelley’s (1992) FSQ and its application in other research. 

Toxic Leadership Theory 

The leadership literature is replete with research and practitioner publications focused on positive leadership. By comparison, the study of negative leadership is not well represented in the literature, and this is a perplexing oddity especially considering the apparent widespread nature of bad leadership. The pervasive nature of toxicity is reported by Kusy and Holloway (2009) who claimed “a whopping 94 percent [of individuals surveyed] have worked with someone toxic in their career” (p. 5). Kusy and Holloway are referring specifically to the widespread nature of toxicity in the workplace. However, Bell (2017) found “that 78% of followers have worked with a toxic leader” (p. 89). Bell’s finding is specific to the widespread nature of toxic leadership in the workplace. As such, it seems clear many toxic individuals exist in the workplace and many of them achieve leadership positions. 

One of the first authors to use the term toxic to describe a certain form of bad leadership was Whicker (1996) who ultimately defined a toxic leader as one who is “maladjusted, malcontent, and often malevolent, even malicious” (p. 11). With the term toxic introduced into the literature, certain authors began to explore this form of bad leadership at a conceptual level. Reed (2004) noted how describing toxic leaders was a challenge but pointed to various concepts such as pettiness, abusiveness, maliciousness, self-aggrandizement, and a general uncaring attitude toward others. Lipman-Blumen (2005) described toxic leaders as those who demonstrated destructive behaviors coupled with certain dysfunctional personal traits. Lipman-Blumen further noted how those behaviors and qualities should be understood as toxic when they began having serious and long lasting negative impacts on the organization and its individual members. Beyond descriptions and concepts, a movement toward establishing a theoretical understanding of toxic leadership began to develop over time. Pelletier (2009) took Lipman-Blumen’s descriptions and conducted qualitative and quantitative research in order to develop the Perceptions of Toxic Leadership Scale. Working with a similar intent toward theory development yet distinct from Pelletier’s work, Schmidt (2008) also conducted qualitative and quantitative research and developed the initial Toxic Leadership Scale. Schmidt’s work on developing this instrument is described further in the subsequent section as the TLS is an integral aspect of the present work. 

Toxic leadership scale. Schmidt’s (2008) stated research goal was to understand the specific and unique toxic leader behaviors that have negative impacts on followers in somewhat predictable patterns. Based on his qualitative work with certain military units, Schmidt found toxic leaders are best described as “narcissistic, self-promoters who engage in an unpredictable pattern of abusive and authoritarian supervision” (p. 57). Schmidt claimed toxic leadership involves five unique dimensions–abusive supervision, authoritarian leadership, narcissism, unpredictability, and self-promotion. Bell (2017) provides a concise description of each dimension as demonstrated in Table 1.

Table 1: Schmidt’s (2008) Toxic Leadership Dimension

DimensionDescription
Abusive Supervision“sustained display of hostile verbal and nonverbal behaviors, excluding physical contact” (Tepper, 2000, p. 178).
Authoritarian Leadership“leader’s behavior that asserts authority and control over subordinates and demands unquestionable obedience from subordinates” (Cheng, Chou, Wu, Huang, & Farh, 2004, p. 91).
Narcissism“a personality trait encompassing grandiosity, arrogance, self-absorption, entitlement, fragile self-esteem, and hostility” (Rosenthal & Pittinsky, 2006, p. 617).
Unpredictability“negative behavior has negative effects, unpredictable negative behavior might exasperate the negative results” (Schmidt, 2008, p. 30).
Self-Promotion“leaders act in ways that promote their own interests above and beyond the interest of the units they are leading, usually with the intention of maintaining a positive image to upper levels of the leadership hierarchy” (Schmidt, 2008, p. 28).
Note. Schmidt derived the abusive supervision, authoritarian leadership, and narcissism dimensions from existing literature and derived the unpredictability and self-promotion dimensions from his qualitative research.
Source. Table originally presented in Bell (2017) as Table 2: Schmidt’s (2008) Toxic Leadership Dimensions (p. 38).

Schmidt (2014) conducted further research using the TLS and was able to establish validity and reliability with a shortened version of the original instrument. The short version of the TLS included 15 questions measuring the same original five dimensions. Schmidt confirmed a five-factor structure for the short version where each of the five dimensions was measured via three questions on the survey. The 15-question short version of the TLS was used in the original Bell (2017) study and again in the present replication work. The following section presents the use and validation of the TLS in other research.

TLS validation in other research. Schmidt’s (2014) TLS has been used in a limited number of subsequent studies (Gallus et al., 2013; Dobbs, 2014; Bell, 2017) where the researcher reported the alpha reliabilities. Gallus et al. used only eight items from the original scale and reported a reliability (a = .92) on those items as one factor. Both Dobbs (2014) and Bell (2017) used the complete 15-item TLS and reported acceptable reliabilities on each of the five dimensions. Hair, Black, Babin, and Anderson (2009) state alpha reliabilities of .70 are generally deemed acceptable with .60 being at the lower end of acceptability. Each study reports alpha reliabilities at or above the .70 level noted by Hair et al. The Cronbach alpha reliabilities reported by Schmidt (2014), Bell (2017), Dobbs (2014), and the present replication work are provided in Table 2. 

Table 2: Reported Reliabilities for Schmidt’s (2014) Toxic Leadership Survey

DimensionSchmidt (2014)Dobbs (2014)Bell 
(2017)
Bell 
(2019)
Abusive Supervisiona = .79a = .70a = .80a = .82 *
Authoritarian Leadershipa = .84a = .92a = .83a = .80
Narcissisma = .81a = .88a = .88a = .88
Unpredictabilitya = .85a = .81a = .82a = .84
Self-Promotiona = .85a = .89a = .70a = .75*
Note. The alpha reliability for self-promotion was improved from .74 to .75 when one 
question was removed, and the alpha reliability for abusive supervision was improved 
from .77 to .82 with the removal of one question.

Followership Typologies 

Although a growing representation of the leadership literature is focused on followership, the vast majority of all research work and practitioner publications in the field favor leadership over followership. In a leader-centric arena, the contributions of followers are oftentimes overlooked. However, according to Koonce, Bligh, Carsten, and Hurwitz (2016), “research suggests that followers contribute an average of 80% to the success of an organization” (p. xv). The relative lack of focus on followers is a surprising reality since there are vastly more followers in organizations than there are leaders, and leaders are almost always leading some and following others simultaneously (Koonce et al., 2016). Much of the formative research and writing on followership has been focused on the development of various classifications of follower types. Some of the pioneering research works in the field of followership were Kelley’s (1988, 1992) publications that presented five follower types including the alienated, the conformist, the passive or sheep, the pragmatics, and the effective or star followers. Kellerman (2008) also concluded followers, based on their engagement level, tend to fall into categories along a spectrum including the “isolate, bystander, participant, activist, and diehard” (p. 85). Chaleff (2009) outlined certain metrics related to the follower’s supporting and challenging of the leader and categorized followers as implementers, partners, resources, or individualists. Kelley’s typologies are the basis for the FSQ instrument which is a focal point in the present work.

Kelley’s (1992) followership typologies were based on the two specific behavior dimensions involving the follower’s level of active engagement and the follower’s level of independent, critical thinking. Kelley described each dimension by contrasting it with the opposite behavior so that active engagement is understood as the opposite of passive involvement, and independent, critical thinking is understood as the opposite of dependent, uncritical thinking. Table 3, as adapted from Bell (2017), provides a description of each follower type and notes the corresponding levels of active engagement and independent, critical thinking associated with each type. 

Table 3: Kelley’s (1992) Follower Styles Typologies

Follower TypeFollower Description
Alienated Followers:
not actively engaged but are independent, critical thinkers
Devil’s Advocate, Skeptical, Cynical, Disgruntled, Adversarial; Has been hurt and is ‘turned off’; Does not trust leader; Leader has mismanaged the follower  
Conformist Followers: the yes-people, who are actively engaged but are not independent, critical thinkersFollows instructions quickly/gladly regardless; Very trusting of and committed to the leader; Subservient attitude toward leader; Does not originate ideas; Aggressive desire to please leader
Passive Followers:  the sheep, who are neither actively engaged nor are they independent, critical thinkersRelies on leader’s judgment and thinking; Acts only when given direction to act; Follows with mediocre effort without asking why; Puts in time but little thought or effort
Pragmatic Followers: middle-of-the-road or lukewarm in their active engagement and independent, critical thinkingAttuned to the shifting winds; Knows how to work the system; Masters the middle ground; Plays political games; Bargains for self-interest; Willing to manipulate to advance personal interests
Effective Followers: star followers, who are both actively engaged and demonstrate high levels of independence in their thinkingExercise independent thinking apart from the leader; Willing to stand up to leaders; Have courage; Remain positive; Take risks; Solve problems independently
Note. Drawn from Kelley (1992): see pages 99-133 for more detail on each type and their corresponding descriptions and causal attributions.
Source. Information originally presented in Bell (2017) as part of Table 3: Kelley’s (1992) Follower Style Typologies (p. 55).

Followership styles questionnaire. Kelley (1992) developed the FSQ based on the previously mentioned behavior dimensions of active engagement and independent, critical thinking. Kelley described active engagement as the degree of positive energy offered by the follower, and Kelley described independent, critical thinking as the level by which followers will think on their own apart from depending on the leader’s thinking. Kelley claimed to have arrived at these two behavior dimensions and their descriptions based on workshops he conducted in several organizations including working directly with over 1,000 people. Kelley specifically noted how he surveyed over 700 people in order to develop the foundational constructs for the questionnaire. The FSQ includes 20 questions with 10 questions specifically measuring active engagement and 10 questions measuring the independent, critical thinking dimension. Kelley did not report any alpha reliabilities for the FSQ.

FSQ validation in other research. The FSQ has been used in a limited number of research studies (Blanchard, Welbourne,  Gilmore, & Bullock, 2009; Gatti, Cortese, Tartari, & Ghislieri, 2014; Bell, 2017) where the researcher reported alpha reliabilities. Blanchard et al. (2009) had to ignore certain items in order to preserve the two factor structure. Likewise, Gatti et al. (2014) had to drop a number of items in order to establish a two factor structure. As with the previous researchers, Bell (2017) dropped several items in order to improve the reliabilities and maintain the two factor structure. With certain items dropped, each study reports alpha reliabilities at or above the .70 level previously noted by Hair et al. as acceptable. The Cronbach alpha reliabilities reported by Blanchard et al. (2009), Bell (2017), Gatti et al. (2014), and the present replication work are provided in Table 4. 

Table 4: Reported Reliabilities for Kelley’s (1992) Followership Styles Questionnaire

DimensionBlanchard et al. (2009)Gatti et al. (2014)Bell 
(2017)
Bell 
(2019)
Active Engagementa = .86a = .94a = .86a = .90 
Independent, Critical Thinkinga = .74a = .79a = .76a = .75
Enthusiastic Commitmentn/an/an/aa = .79*
Note. The alpha reliability for enthusiastic commitment is not applicable in the first three 
studies as this is a new factor established in the present work.

Phase I

An important element of this study is the partial replication of Bell’s (2017) research. In the present work, Phase I refers to Bell’s original research that is described briefly in this section. As such, this section describes the research design, method, hypotheses, results, and analysis provided in that first study. It is important to understand the basic construct of the first study since a partial replication was completed.

Design and Method 

Bell’s (2017) research was a quantitative, cross-sectional study utilizing a hierarchical multiple regression analysis. Bell employed three survey instruments in the study including the TLS, FSQ, and the LMX-7. Graen and Uhl-Bien (1995) endorse the LMX-7 claiming it is most useful to measure the leader/follower working relationship as a single dimension in conjunction with leader-member exchange theory. Bell packaged the three instruments as one online survey and collected data from non-traditional, adult students enrolled in online graduate programs at his university. The instrument required affirmation of a followership experience with a toxic leader, and basic demographics including gender, ethnicity, age, etc. were collected (Bell, 2017). The purpose of Bell’s study was to investigate the potential effect of the five dimensions of toxic leadership on the two dimensions of followership both directly and as moderated by LMX status.

Hypotheses Testing 

Based on the research design and purpose, Bell (2017) tested twenty distinct hypotheses. The effect of the five toxic leadership dimensions on active engagement were tested directly and then separately tested as moderated by LMX resulting in ten hypotheses (Bell, 2017). Similarly, the effect of the five toxic leadership dimensions on independent, critical thinking were tested directly and then separately tested as moderated by LMX resulting in ten more hypotheses for a total of twenty hypotheses (Bell, 2017). Appendix A demonstrates all twenty hypotheses. Examples from each of Bell’s hypothesis categories are as follows:

H1: The toxic leadership dimension of abusive supervision, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire

H1a: An interaction between the toxic leadership dimension of abusive supervision, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H10: The toxic leadership dimension of self-promotion, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire

H10a: An interaction between the toxic leadership dimension of self-promotion, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. (pp. 14-17).

Results and Analysis

Bell (2017) sent surveys to 556 non-traditional, adult students enrolled in online graduate programs at his university. 260 participants returned surveys, and 203 (78 %) of those returned were allowed to participate in the study as they indicated having had experience working with a toxic leader (Bell, 2017). Having nearly eight of ten participants respond affirmatively regarding working experience with a toxic leader is a substantive finding and one that demonstrates the pervasiveness of the issue. Interestingly, Bell’s regression analysis only found evidence to support one of the twenty hypotheses: the self-promotion dimension of toxic leadership did directly influence the follower’s active engagement. As has already been noted, Bell also conducted reliability analyses on the TLS and FSQ. All five dimensions of the TLS were found reliable, and the two dimensions of the FSQ were found reliable when certain elements were dropped.   

Phase II

As previously noted, two broad but important research objectives existed for the present study. The first objective was to complete a partial replication of Bell’s (2017) study using a similar yet distinct sample in order to examine the similarities and variances in results. The second objective involved scale refinement work in order to further validate and refine the TLS and FSQ. Both of these research objectives were accomplished and are presented here as Phase II.

Design and Method 

All the basic design and methodology aspects of Bell’s (2017) research were duplicated in the present work with the exception of the respondent’s education level and the overall sample size. The present work again employed the TLS, FSQ, and the LMX-7 in an identical manner as one online survey. The instrument again required affirmation of a followership experience with a toxic leader and collected the same basic demographics including gender, ethnicity, age, etc. Data were collected from non-traditional, adult students enrolled in online undergraduate programs at the same university. Two differences exist with the data collection in the present work as compared to the original study. First, data were collected from undergraduate students in the present study whereas data were collected from graduate students in the initial study. Second, the sample size in Bell (2017) was 203 usable respondents whereas the present study had 330 usable responses. A function of the replication work was to retest the same twenty hypotheses to determine whether the results were the same as in the original study and to determine results in consideration of one’s level of education since level of education was the only difference between the two sample groups. The results from the hypotheses testing are presented in the following section. Additionally, an important aspect of the current study was to conduct scale refinement work on both the TLS and FSQ. The results of those efforts also follow in subsequent sections.

Hypotheses Testing and Results

All twenty hypotheses used in Bell’s (2017) work and in the current study are provided in Appendix A. In the present work, as with the initial study, the effect of the five toxic leadership dimensions on active engagement were tested directly and then separately tested as moderated by LMX resulting in ten hypotheses. Likewise, the effect of the five toxic leadership dimensions on independent, critical thinking were tested directly and then separately tested as moderated by LMX resulting in ten more hypotheses for a total of twenty hypotheses. 

In the current study, surveys were sent to 1,094 non-traditional, adult students enrolled in online undergraduate programs. 437 participants returned surveys, and 330 (75.5 %) of those returned were allowed to participate in the study as they indicated having had experience working with a toxic leader. The finding that 75.5 % of respondents claimed to have worked with a toxic leader in their career is nearly the same as the 78 % found in the original study and again indicates the significance of the toxic leader problem. Table 5 provides the means, standard deviations, and correlations among the independent and dependent variables. Hair et al. (2009) note how multicollinearity can be a problem across the independent variables when significant correlations (.70 or above) are found. Several of the toxic leadership variables do correlate at the .70 level or above. Additionally, the followership variables also correlate at the .70 level. This multicollinearity issue corroborates the need for scale refinement work on these two measures.

Table 5: Means, Standard Deviations, and Correlations Among Independent and Dependent Variables

VariableMSD12345678
1. Self-promotion3.911.00
2. Abusive Supervision3.631.25.61**
3. Unpredictability3.811.10.62**.76**
4. Narcissism4.121.05.67**.73**
5. Auth. Leadership3.771.04.62**.62**.65**.72**
6. LMX2.88.97-.41**-.45**-.46**-.39**-.36**
7. Active Engagement4.70.93.05.04.07.13**.11*.31**
8. Ind., Critical Thinking4.26.91.18**.18**.21**.26**.22**.17**.70**
*< .05.
**< .01.

Hierarchical multiple regression analysis was conducted to test the ten hypotheses involving the direct and moderated effects of the toxic leader dimensions on the follower’s active engagement. The results were as follows. H3 was supported as a significant (p = .004) direct effect was found between the toxic leader dimension of narcissism and the follower’s active engagement. H1, H2, H4, and H5 were not supported as no significant direct effect was detected between the other four toxic leader dimensions and the follower’s active engagement. H3a was supported as a significant (p = .02) effect was found between the toxic leader dimension of narcissism, as moderated by LMX level, and the follower’s active engagement. H1a, H2a, H4a, and H5a were not supported as no significant direct effect was detected between the other four toxic leader dimensions, as moderated by LMX level, and the follower’s active engagement. The R2 change demonstrated in the model did indicate an increase in the explanatory power from model 1 to model 2 (1% to 18%) and from model 2 to model 3 (18% to 24%). These levels of explanatory power are still relatively modest. 

Hierarchical multiple regression analysis was also conducted to test the ten hypotheses involving the direct and moderated effects of the toxic leader dimensions on the follower’s independent, critical thinking. The results were as follows. H8 was supported as a significant (= .02) direct effect was found between the toxic leader dimension of narcissism and the follower’s independent, critical thinking. H6, H7, H9, and H10 were not supported as no significant direct effect was detected between the other four toxic leader dimensions and the follower’s independent, critical thinking. Hypotheses 6a through 10a were not supported as no significant effect was detected between the five toxic leader dimensions, as moderated by LMX level, and the follower’s independent, critical thinking. The R2 change demonstrated in the model does indicate an increase in the explanatory power from model 1 to model 2 (0% to 18%) and from model 2 to model 3 (18% to 19%). As with the active engagement regression model, these levels of explanatory power are still relatively modest. 

TLS Scale Refinement 

Several analytical procedures were conducted on the TLS instrument. Initially, the reliability of each dimension was ascertained using Cronbach’s alpha. The self-promotion dimension’s reliability was improved from (= .74) to (= .75) after removing one item. The abusive supervision dimension’s reliability was also improved from (= .77) to (= .82) after removing one item. The other three dimensions could not be improved by removing items. The reliabilities on the other dimensions were found as authoritarian leadership (= .80), narcissism (= .88), and unpredictability (= .84). Hair et al. (2009) claim alpha reliabilities of .70 or higher are considered as acceptable. As such, each dimension of the TLS scale is demonstrably reliable. 

Factor analysis was conducted on the TLS once the reliability had been established. As was previously noted, many of the variables within the TLS instrument are highly correlated. The two items dropped in the reliability analysis were not included in the factor analysis. As suspected, all the items loaded on one factor. The one factor was strong with an eigenvalue of 7.48 and explained 57.54% of the variance. All the factor loadings exceeded the .60 level Hair et al. (2009) note as acceptable. Table 6 demonstrates the factor loadings for the single factor. Appendix B presents the modified version of Schmidt’s (2014) instrument with the 13 items that factor into the unidimensional toxic leadership construct. All the original five dimensions are still represented, but toxic leadership is apparently a unidimensional construct of five individual, yet highly correlated, dimensions. This approach mirrors Graen and Uhl-Bien’s (1995) approach to understanding leader-member exchange. Graen and Uhl-Bien, in advocating for the use of the LMX-7, claimed LMX is a single dimension (measured by the LMX-7) formed from several highly correlated dimensions. The modified toxic leadership instrument is called the TLS-1 (= .94) as it seems to measure the one toxic leadership dimension via the 13 questions provided. 

Table 6: Factor Loadings for TLS Instrument

VariableToxic Leadership (a = .94)
Self-promotion.655
Self-promotion.748
Abusive Supervision.780
Abusive Supervision.762
Unpredictability.796
Unpredictability.788
Unpredictability.737
Narcissism.808
Narcissism.827
Narcissism.782
Authoritarian Leadership.688
Authoritarian Leadership.788
Authoritarian Leadership.681
Note. Not included are Schmidt’s (2014) self-promotion question 1 and abusive 
supervision question 1 as these were dropped in the reliability analysis. 
See Bell (2017) for the full version of Schmidt’s (2014) original TLS.

FSQ Scale Refinement 

The same basic procedures were used in the FSQ scale refinement work as were used with the TLS. The reliability of each dimension on the FSQ was determined as a first step. The active engagement dimension proved reliable (= .87) with keeping all 10 items. Similarly, with keeping all 10 items, the independent, critical thinking dimension proved reliable (= .78). Both of these reliabilities exceed Hair et al.’s (2009) .70 threshold of acceptability. 

Factor analysis on the FSQ required a number of steps before individual factors could be isolated and validated. The factor analysis did initially include all 20 items since none were dropped in the reliability analysis. The first iteration revealed the items on the scale loaded on three unique factors. Hair et al. (2009) recommends dropping items that load at .40 or above on multiple factors. As such, three items were dropped after the first iteration, and the factor analysis was conducted on the remaining 17 items. The second run of factor analysis still revealed the three-factor structure but also netted another item that loaded at or above the .40 level. That item was removed, and factor analysis was conducted on the remaining 16 items. One more item was removed after the final factor analysis because it did not meet Hair et al.’s .60 level threshold of acceptability for retaining items. The final analysis revealed the three-factor structure based on the remaining 15 items. Those three factors had significant eigenvalues. Factor 1’s eigenvalue was 5.87 and explains 36.67% of the variance. Factor 2’s eigenvalue was 2.02 and explains 12.67% of the variance. Factor 3’s eigenvalue was 1.74 and explains 10.87% of the variance. These 3 factors cumulatively explain 60.21% of the total variance. 

Since the original FSQ was based on the two followership dimensions of active engagement (AE) and independent, critical thinking (ICT), some additional consideration was necessarily given to how to allocate descriptions of the three distinct dimensions. Indeed, each question that loaded on a particular factor was reviewed in order to help determine which dimension the factor best represented. Factor 3 was the simplest to ascertain as all four items were originally part of the ICT dimension, and the four questions are easily discerned as indicative of that dimension because they are reflective of an internal cognitive effort. For example, one of those items asks: “Do you make a habit of internally questioning the wisdom of the leader’s decision rather than just doing what you are told?” That question seems to convey an internal cognitive effort focus (e.g. internally questioning). Thus, factor 3 is considered as representative of the ICT dimension.

The other two factors both had a mixture of items from both the AE and ICT dimensions of the original measure. The majority of the factor 1 items were originally part of the AE dimension, and those questions were easily ascertained as indicative of the AE dimension because they represent an outward effort of activity. The other questions on factor 1 were reviewed closely because they were originally part of the ICT dimension. Upon close review, these items were determined to share the appearance of an outward effort of activity like those of the AE dimension rather than being reflective of an internal cognitive effort such as the other ICT items. For example, one of those items asks “Do you try to solve tough problems (technical or organizational) rather than look to the leader to do it for you?” That question seems to convey outward activity focus (e. g. solving problems on one’s own initiative). As such, factor 1 is considered as representative of the AE dimension. Both Blanchard et al. (2009) and Gatti et al. (2014) made similar decisions relative to these items.

Factor 2 also had a mixture of items from both the AE and ICT dimensions of the original measure. These four questions were reviewed closely in an attempt to discern what followership style dimension they represented. Blanchard et al. (2009) and Gatti et al. (2014) both reviewed those items as part of a third factor and labeled them as related to the follower’s attitude. Ultimately these four questions appear to be measuring the level of energy or enthusiasm an individual demonstrates via dedication and commitment. In short, these items appear to measure a follower’s level of enthusiastic commitment as contrasted with the follower’s level of apathetic detachment. Describing the dimension by contrasting it with its alternative (enthusiastic commitment vs. apathetic detachment) follows the method Kelley (1992) used to describe the original two dimensions (active engagement vs. passive involvement and independent, critical thinking vs. dependent, uncritical thinking). An example items asks: “Are you highly committed to and energized by your work and organization, giving them your best ideas and performance?” That question seems to clearly convey interest in one’s energy or enthusiasm toward dedication and commitment. A more specific label than attitude is warranted. Therefore, factor 2 is considered as representative of the enthusiastic commitment (EC) followership dimension.  

With the establishment of the three factor structure of the FSQ, Cronbach’s alpha reliability analysis was completed on all three dimensions. The reliabilities found were AE (= .90) , EC (= .79) , and ICT (= .75). As such, all three dimensions met Hair et al.’s (2009) .70 level of acceptability. Table 7 demonstrates the factor loadings and alpha reliabilities for the FSQ instrument. Since the scale refinement work on the FSQ netted a new dimension and a reduction of items, a modified instrument is offered. Appendix C presents the modified version of Kelley’s (1992) instrument. The modified instrument is called the FSQ-3 as it seems to measure followership styles in three dimensions.

Table 7: Factor Loadings for FSQ Instrument

VariableFactor 1 (AE)
= .90 
Factor 2 (EC) 
= .79 
Factor 3 (ICT) 
= .75 
Active Engagement.723
Active Engagement.763
Active Engagement.823
Active Engagement.725
Active Engagement.781
Active Engagement.805
Active Engagement.768
Enthusiastic Commitment.776
Enthusiastic Commitment.807
Enthusiastic Commitment.782
Enthusiastic Commitment.655
Independent, Critical Thinking.655
Independent, Critical Thinking.778
Independent, Critical Thinking.797
Independent, Critical Thinking.733
Note. Not included are questions 5, 6, 7, 14, and 15, from Kelley’s (1992) FSQ. See Bell (2017) 
for the full version of Kelley’s (1992) original FSQ.

Summary 

Phase II accomplished two important research objectives. First, a partial replication of Bell’s (2017) study was completed using the same basic design and methodology. The only exceptions were the respondent’s education level (graduate vs. undergraduate students) and a larger sample size (203 vs. 330). The replication work did net some interesting results that varied from the original study. In the original research, only the toxic leadership dimension of self-promotion was found to have an effect on the follower’s active engagement levels. In the replication study, only the toxic leadership dimension of narcissism was found to have an effect, and the narcissism dimension affected both the follower’s active engagement level and their level of independent, critical thinking. Second, scale refinement work was conducted on both the TLS instrument and the FSQ instrument. The TLS was found to have highly correlated variables, and factor analysis revealed the TLS instrument seems to measure only one factor–toxic leadership. Therefore, toxic leadership is proposed as being a unidimensional construct formed by five highly correlated dimensions. A new instrument, the TLS-1 (= .94), is suggested for use to measure toxic leadership as one dimension. The FSQ instrument was found to include three factors instead of the two dimensions originally proposed with the scale. A review of the individual items included in the FSQ in consideration of their factor loadings netted a new followership dimension called enthusiastic commitment. This new dimension needs to be added to the original active engagement and independent, critical thinking dimensions. A reliability analysis on the three factor structure indicates the instrument is reliable. As such, a new instrument, the FSQ-3, is suggested for use to measure the three followership dimensions.  

Discussion

The results of the partial replication are discussed specifically related to the level of education control variable. A discussion is also provided concerning the results of the scale refinement work including the modified instruments. Finally, several future research needs are noted in order to continue to refine and validate the TLS-1 and the FSQ-3 instruments.

Level of Education 

Several of the hypotheses’ results varied between Phase I and Phase II, and the education level of the sample group was a noteworthy difference. Both studies collected data from non-traditional adult learners who were enrolled in online programs at the same university. Phase I’s group were all enrolled in graduate programs so an earned bachelor’s degree necessarily existed in their past whereas Phase II’s group were enrolled in an undergraduate program so it is expected the vast majority, if not all, of those students did not have an earned bachelor’s degree. As such, the level of education was a de facto control variable between the two studies. One can speculate concerning the reasons why the level of education may be a factor in how followers respond to toxic leaders. Increased job mobility, more varied career experiences, increased self-esteem generated by greater education, and/or less motivation by lower level security needs are all items potentially influencing how more/less educated individuals respond to toxic leaders. Future research is warranted in this area in order to determine what variables may help certain followers remain actively engaged when dealing with a toxic leader. However, it should be noted the results of Phase I and Phase II could potentially be different based on the scale refinement work previously discussed. Some additional post-hoc analysis should be conducted on the original data sets in order to help clarify and refine those results.  

New Instruments

One of the study’s important results stems from the refinement work conducted with both original instruments as two newly refined instruments are presented. Some historical precedent for the changes in these instruments seems to exist in the limited research where the scales have been used. For example, Gallus et al. (2013) found a single factor structure could be established by deleting certain items from the TLS. Dobbs (2014) derived a single variable called “Overall Toxic Leadership” (p. 38) by averaging the responses to all 15 questions on the TLS. Schmidt (2014) refers to other unpublished research where the TLS factors “loaded onto a single second-order construct” (p. 3) understood simply as toxic leadership. Similarly, Blanchard et al. (2009) initially found a three-factor structure with the FSQ, but only used the AE and ICT factors for their analysis. Blanchard et al. labeled this third factor as “attitude and affect” (p. 120). Gatti et al. (2014) also found three factors and chose to drop certain items to only focus on the AE and ICT variables. Gatti et al. described this third factor as the follower’s attitude. The results of the present work along with the results by the other mentioned research indicates a shift is needed related to these scales. The TLS-1 (= .94) is suggested for use in measuring the unidimensional construct that is toxic leadership. The FSQ-3 is suggested for use in measuring the three followership dimensions of active engagement (= .90), enthusiastic commitment (= .79), and independent, critical thinking (= .75). 

Future Research Needs 

Several future research needs should be considered. In a general sense, the majority of the research needed at present related to both toxic leadership and followership styles involves more scale refinement. As under-represented aspects of the overall leadership literature, the future study of both toxic leadership and followership would be greatly enhanced by establishing some primary instruments to measure the constructs. While the present work is a step in that direction, much more work needs to be done. More replication of both Phase I and Phase II are needed along with more varied scale refinement efforts. More replication is useful to determine if the same factor structures will continue to hold each time the instruments are used. Additional scale refinement on both the TLS-1 and the FSQ-3 is critical especially in the area of establishing discriminant validity. Establishing discriminant validity between toxic leadership and some positive leadership style, such as servant leadership, would be useful. To do so most effectively, a unidimensional servant leadership instrument such as the Essential Servant Leadership Behaviors scale developed by Winston and Fields (2015) should be used. Likewise, establishing discriminant validity between the followership styles and some negative followership traits would also be quite useful. To accomplish this, an instrument such as Henderson’s (2015) Toxic Followership Types scale could be used. Many other research needs exist relative to both toxic leadership and effective followership. However, it seems apparent no greater need exists at present than to fully refine instrumentation for both toxic leadership and effective followership so the instruments could be available for use to accomplish the many other existing research needs.

Conclusion

The purpose of the study was to complete a partial replication of Bell (2017) and to conduct scale refinement work with the TLS and the FSQ. The limited but relevant literature related to both toxic leadership and followership was reviewed with a special emphasis on the development and previous usage of both instruments. The original study, Phase I, was described in terms of design, methods, and results. Phase II was described in like manner in order to demonstrate how the study was replicated. Only one variable, level of education, differentiated the sample groups. In Phase II, the toxic leader dimension of narcissism was found to impact the follower’s AE and ICT. The second part of Phase II detailed the various iterations of reliability analysis and factor analysis revealing toxic leadership as a unidimensional construct and revealing a third followership dimension called enthusiastic commitment. The two modified instruments, the TLS-1 and the FSQ-3, are suggested for future use. The TLS-1 measures the one dimensional toxic leader construct whereas the FSQ-3 measures the three dimensions of followership. Several future research needs were discussed with special emphasis on continued refinement and testing of the TLS-1 and the FSQ-3. Researchers interested in either toxic leadership or followership have been hindered by the lack of consistently acceptable instruments to measure those concepts. Continued use and testing of these modified instruments may help lead to some commonly accepted measures for both toxic leadership and followership.

References

Bell, R. M. (2017). The dysfunction junction: The impact of toxic leadership on follower effectiveness (UMI No. 10260880). Available from ProQuest Dissertations and Theses Global database. 

Blanchard, A., Welbourne, J., Gilmore, D., & Bullock, A. (2009). Followership styles and employee attachment to the organization. Psychologist-Manager Journal (Taylor & Francis Ltd), 12(2), 111-131.

Chaleff, I. (2009). The courageous follower: Standing up to & for our leaders (3nd ed.). San Francisco: Berrett-Koehler.

Cheng, B. Chou, L., Wu, T., Huang, M., & Farh, J. (2004). Paternalistic leadership and subordinate responses: Establishing a leadership model in Chinese organizations. Asian Journal of Social Psychology, 7, 89-117.

Dobbs, J. M. (2014). The relationship between perceived toxic leadership styles, leader effectiveness, and organizational cynicism (Doctoral dissertation). Available from ProQuest Dissertations and Theses Global database. (UMI No. 1459425092) 

Gallus, J. A., van Driel, M., Walsh, B. M., Gouge, M. C., & Antolic, E. (2013). Intolerable cruelty: A multilevel examination of the impact of toxic leadership on U.S. military units and service members. Military Psychology (American Psychological Association), 25(6), 588-601. doi:10.1037/mil0000022

Gatti, P., Cortese, C. G., Tartari, M., & Ghislieri, C. (2014). Followers’ active engagement: Between personal and organizational dimensions. BPA – Applied Psychology Bulletin (Bollettino Di Psicologia Applicata), (270), 2-11.

Graen, G. B., & Uhl-Bien, M. (1995). Relationship based approach to leadership: Development of leader–member exchange (LMX) theory of leadership over 25 years: Applying a multi-domain perspective. The Leadership Quarterly, 6, 219-247.

Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2009). Multivariate data analysis (7th ed.). Upper Saddle River, NJ: Prentice Hall.

Henderson, C. (2015). Toxic followership: A measure of followers associated with destructive leadership (Order No. 10000125). Available from ProQuest Dissertations & Theses Global. (1759139102).

Kellerman, B. (2008). Followership: How followers are creating change and changing leaders. Boston, MA: Harvard Business Press.

Kelley, R. E. (1988). In praise of followers. Harvard Business Review, 66(6), 142-148.

Kelley, R. E. (1992). The power of followership: How to create leaders people want to follow and followers who lead themselves. New York, NY: Doubleday.

Koonce, R., Bligh, M. C., Carsten, M. K., & Hurwitz, M.(Eds.). (2016). Followership in action: Cases and commentaries. United Kingdom: Emerald Publishing.

Kusy, M. E., & Holloway, E. L. (2009). Toxic workplace: Managing toxic personalities and their systems of power. San Francisco, CA: Jossey-Bass.

Lipman-Blumen, J. (2005). The allure of toxic leaders: Why we follow destructive bosses and corrupt politicians—And how we can survive them. Oxford, NY: Oxford University Press.

Padilla, A., Hogan, R., & Kaiser, R. B. (2007). The toxic triangle: Destructive leaders, susceptible followers, and conducive environments. Leadership Quarterly, 18(3), 176-194. doi:10.1016/j.leaqua.2007.03.001

Pelletier, K. L. (2009). The effects of favored status and identification with victim on perceptions of and reactions to leader toxicity (Doctoral dissertation). Available from ProQuest Dissertations and Theses Global database. (UMI No. 304861401)

Reed, G. E. (2004). Toxic leadership. Military Review, 84(4), 67.

Rosenthal, S. A., & Pittinsky, T. L. (2006). Narcissistic leadership. Leadership Quarterly, 17(6), 617-633. doi:10.1016/j.leaqua.2006.10.005

Schmidt, A. A. (2008). Development and validation of the toxic leadership scale (Master’s thesis). Available from ProQuest Dissertations and Theses Global database. (UMI No. 193655997)

Schmidt, A. A. (2014). An examination of toxic leadership, job outcomes, and the impact of military deployment (Doctoral dissertation). Available from ProQuest Dissertations and Theses Global database. (UMI No.1558874321)

Tepper, B. J. (2000). Consequences of abusive supervision. The Academy of Management Journal, 2, 178.

Whicker, M. L. (1996). Toxic leaders: When organizations go bad. Westport, CT: Quorum Books.

Winston, B. & Fields, D. (2015). Seeking and measuring the essential behaviors of servant leadership. Leadership and Organizational Development Journal 36(4) 413-434.

Appendix A

Hypotheses from Bell (2017) and Present Study

H1: The toxic leadership dimension of abusive supervision, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire

H1a: An interaction between the toxic leadership dimension of abusive supervision, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H2: The toxic leadership dimension of authoritarian leadership, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire

H2a: An interaction between the toxic leadership dimension of authoritarian leadership, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H3: The toxic leadership dimension of narcissism, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire.  

H3a: An interaction between the toxic leadership dimension of narcissism, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H4: The toxic leadership dimension of unpredictability, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire

H4a: An interaction between the toxic leadership dimension of unpredictability, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H5: The toxic leadership dimension of self-promotion, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire

H5a: An interaction between the toxic leadership dimension of self-promotion, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s AE, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H6: The toxic leadership dimension of abusive supervision, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire

H6a: An interaction between the toxic leadership dimension of abusive supervision, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H7: The toxic leadership dimension of authoritarian leadership, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire

H7a: An interaction between the toxic leadership dimension of authoritarian leadership, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H8: The toxic leadership dimension of narcissism, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire

H8a: An interaction between the toxic leadership dimension of narcissism, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H9: The toxic leadership dimension of unpredictability, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire.  

H9a: An interaction between the toxic leadership dimension of unpredictability, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire, among out-group members than in-group members. 

H10: The toxic leadership dimension of self-promotion, as measured by the Toxic Leadership Scale, predicts a negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire

H10a: An interaction between the toxic leadership dimension of self-promotion, as measured by the Toxic Leadership Scale, and the LMX in-group/out-group status, as measured by the LMX-7, predicts a greater negative effect on a follower’s independent critical thinking, as measured by the Followership Styles Questionnaire, among out-group members than in-group members.

Appendix B

TLS-1

Modified Version of Schmidt’s (2014) Toxic Leadership Scale (= .94)

All items are rated on a 5-point Likert scale response format, with answers ranging between 1= “Strongly Disagree” to 5 = “Strongly Agree.”

All items begin with the phrase “My current supervisor…”

  1. Will only offer assistance to people who can help him/her get ahead
  2. Accepts credit for successes that do not belong to him/her
  3. Publicly belittles subordinates
  4. Reminds subordinates of their past mistakes and failures
  5. Allows his/her current mood to define the climate of the workplace
  6. Expresses anger at subordinates for unknown reasons
  7. Varies in his/her degree of approachability
  8. Has a sense of personal entitlement
  9. Thinks that he/she is more capable than others
  10. Believes that he/she is an extraordinary person
  11. Controls how subordinates complete their tasks
  12. Does not permit subordinates to approach goals in new ways
  13. Determines all decisions in the unit whether they are important or not

Appendix C

FSQ-3

Modified Version of Kelley’s (1992) Followership Styles Questionnaire

All items are rated on a 6-point Likert scale response format, with answers ranging between 0 = “Rarely” to 3 = “Occasionally” to 6 = “Almost Always”

Active Engagement (= 90)

  1. Can the leader give you a difficult assignment without the benefit of much supervision, knowing you will meet your deadline with highest-quality work and that you will “fill the cracks” if need be?
  2. Do you take the initiative to seek out and successfully complete assignments that go above and beyond your job?
  3. When you are not the leader of a group project, do you still contribute at a high level, often doing more than your share?
  4. Do you help out other co-workers, making them look good, even when you don’t get any credit?
  5. Do you independently think up and champion new ideas that will contribute significantly to the leader’s or the organization’s goals?
  6. Do you try to solve the tough problems (technical or organizational) rather than look to the leader to do it for you?
  7. Do you actively and honestly own up to your strengths and weaknesses rather than put off evaluation?

Enthusiastic Commitment (= .79)

  1. Are your personal work goals aligned with the organization’s priority goals?
  2. Are you highly committed to and energized by your work and organization, giving them your best ideas and performance?
  3. Does your enthusiasm also spread to and energize your co-workers?
  4. Does your work help you fulfill some societal goal or personal dream that is important to you?

Independent, Critical Thinking (= .75)

  1. Do you make a habit of internally questioning the wisdom of the leader’s decision rather than just doing what you are told?
  2. When the leader asks you to do something that runs contrary to your professional or personal preferences, do you say “no” rather than “yes?”
  3. Do you act on your own ethical standards rather than the leader’s or the group’s standards?
  4. Do you assert your views on important issues, even though it might mean conflict with your group or reprisals from the leader?

R. Mark Bell is Assistant Professor of Management at Wayland Baptist University and serves as a faculty member in WBU’s Doctor of Management program. Mark holds an M.B.A. and a Ph.D. in Organizational Leadership. Mark’s research interests include effective followership and ineffective leadership. 

The author can be contacted via email: bellm@wbu.edu

Leave a Comment