Supervisors' interpersonal styles: An integrative perspective and a measure based on self-determination theory

This research addresses recent calls for an alternative integrative framework to apprehend leaders' behaviors and examines the validity ofaquestionnaire anchored in this theoretical approach. Building upon Self-Determination Theory, we examined a tripartite approach of supervisors' behaviors (supportive, thwarting, and indifferent toward subordinates' psychological needs). The psychometric properties of this Tri-partite Measure of Interpersonal Behaviors-Supervisor (TMIB-S) were tested through three studies. Results from bifactor exploratory structural equation modeling supported a solution including one global factor and three specific factors reflecting need supportive, thwarting, and indifferent behaviors. This solution was fully invariant across distinct samples of French-and English-speaking employees. Results also supported the criterion-related and discriminant validity of the


INTRODUCTION
People do not leave their jobs.They leave their managers.This was suggested by a Gallup survey revealing that roughly half of 7272 surveyed American adults had left their job primarily to "get away from their manager" (Harter & Adkins, 2015).Managers are known to play a pivotal role as they contribute to propel organizations toward their desired future state while creating a work environment where employees can thrive and experience well-being (Inceoglu et al., 2018).This complex balancing act that managers have to perform on a daily basis emphasizes why their leadership styles (i.e., "sets of behaviors that leaders employ to influence the behaviors of subordinates"; Skakon et al., 2010, p. 109) constitute a high-stake issue for both organizations and employees.
Three core leadership styles were originally introduced by Lewin et al. (1939), namely, democratic, authoritarian, and laissez-faire, setting the stage for over 80 years of research that has described a variety of leadership behaviors that still share conceptual similarities with Lewin et al.'s (1939) seminal proposal.These behaviors typically fall under three main categories: (1) leadership behaviors relying on guidance, choice, and participative decision-making (autonomy and structure; e.g., leader-member exchange, LMX; Graen & Uhl-Bien, 1995) thus matching the democratic style; (2) leadership behaviors stressing control and order, leaving no room for initiative or choice (structure but no autonomy; e.g., abusive supervision; Tepper, 2000), thus matching the authoritarian style; and (3) behaviors characterizing leaders who provide no guidance or direction and give employees freedom (autonomy but no structure; e.g., passive-avoidant leadership; Avolio et al., 1999), thus matching the laissez-faire style.Because of this conceptual overlap, wherein many distinct theories encapsulate similar ideas in a fragmented manner, leadership researchers keep calling for a more integrative framework to apprehend leaders' behaviors (e.g., Avolio, 2007;Piccolo et al., 2012).
The present research offers to address this call.Rather than adding one more highly specific type of behavior to the already long list of behaviors considered in leadership research (Anderson & Sun, 2017;Piccolo et al., 2012), we take a step back to propose an overarching representation of leadership behaviors anchored in Self-Determination Theory (SDT; Ryan & Deci, 2000, 2017).This theory conceptualizes leadership as a way to promote subordinates' selfdetermined motivation and well-being.Indeed, most of the existing leadership research has treated leadership as a way to influence subordinates' behaviors and performance (see Inceoglu et al., 2018) and in doing so has failed to document the differentiated effects of distinct supervisory behaviors for motivation and well-being.Therefore, we build upon recent advances in SDT, uncovered in the sport context (Bhavsar et al., 2019), to test the validity of an alternative integrative conceptual approach and measurement (one grounded in a motivational perspective) of work supervisors' interpersonal behaviors.This perspective might not replace classical leadership theories (e.g., Avolio & Bass, 1991) when organizational outcomes are considered.
Yet, being anchored in the currently dominant theoretical framework on employees' motivation and well-being, our perspective has the advantage of providing clearer guidance regarding the motivational and health implications of leaders' behaviors for subordinates.This better understanding is made possible not only by the theoretical framework we rely upon (i.e.,SDT) but also by the methodological approach we pursue.Indeed, prior leadership research has failed to adequately address the multidimensionality of supervisory behaviors, which has led to erroneously similar levels of predictive validity of distinct supervisory behaviors across several criteria (Judge & Piccolo, 2004).In contrast, recent research based on SDT has resorted to a rigorous examination of the multidimensionality of interpersonal behaviors (Bhavsar et al., 2019;T oth-Kir aly et al., 2021), allowing to better grasp their unique implications.In the present research, we adopt this advanced methodological approach (i.e., bifactor exploratory structural equation modeling) to achieve a fine-grained representation of employees' perceptions of their supervisors' behaviors.In sum, our research seeks to (1) offer an alternative integrative theoretical approach of leaders' behaviors based on recent findings from SDT in the sport domain, (2) test the validity of a questionnaire (adapted from the sport area) anchored in this theoretical approach, (3) analyze the multidimensionality of supervisory behaviors, and (4) examine the motivational and well-being consequences associated with these distinct types of leadership behaviors.

SDT as an integrative framework
To answer scholars' call for an alternative integrative framework to apprehend leaders' behaviors (Anderson & Sun, 2017), we propose to shift attention toward the motivational perspective offered by SDT (Ryan & Deci, 2000, 2017).Unlike most leadership theories (Inceoglu et al., 2018), SDT has a main focus on subordinates' motivation and well-being and considers that any type of environmental characteristic is likely to impact on these critical indices of individual functioning.In this regard, SDT proposes the satisfaction of employees' basic psychological needs for autonomy (feeling ownership of one's actions), competence (feeling efficient in accomplishing personally important tasks), and relatedness (feeling secure and accepted in one's relationships) as the most critical drivers of motivation and well-being (Ryan & Deci, 2017).The existence and importance of these three needs for motivation and well-being have been empirically validated across cultures (for an overview, see Vansteenkiste et al., 2020).From this perspective, any environmental condition, including leadership behaviors, likely to nurture the satisfaction of these needs should help drive employees' motivation and support their well-being (Ryan & Deci, 2017).This core assumption of SDT has thus far been supported across various domains in research focusing on the need supportive interpersonal behaviors of sport coaches (Ntoumanis et al., 2018), healthcare workers (Ntoumanis et al., 2020), teachers (Cheon et al., 2019), and supervisors (Slemp et al., 2018).
A tripartite approach of supervisors' behaviors Initial propositions showed the importance, and qualitatively distinct nature, of interpersonal behaviors likely to satisfy these basic psychological needs versus those likely to thwart these needs (Myers et al., 2014;Rocchi, Pelletier, Cheung, et al., 2017;Rocchi, Pelletier, & Desmarais, 2017).More recently, Bhavsar et al. (2019) proposed, and validated, a more comprehensive tripartite conceptualization of coaches' interpersonal behaviors including need indifferent behaviors alongside need supportive and thwarting ones.Interestingly, this conceptualization echoes the three types of leadership behaviors (democratic, authoritarian, and laissez-faire) initially described by Lewin et al. (1939), thus suggesting that this conceptualization might also be relevant in the work area.Hence, we suggest that supervisors, just like coaches (Bhavsar et al., 2019), can display behaviors that can be perceived as supportive, thwarting, or indifferent toward the psychological needs of their employees.More precisely, need supportive supervisors promote the satisfaction of their subordinates' psychological needs through behaviors conveying understanding, encouragement, and appreciation.Need thwarting supervisors threaten their subordinates' psychological needs through behaviors involving pressure, nonconstructive criticism, and rejection.Finally, need indifferent supervisors are those who neglect, or ignore, their subordinates' psychological needs through behaviors reflecting disinterest, disorganization, and relational distance.
Interestingly, this conceptualization also echoes the "full-range leadership theory" (Avolio & Bass, 1991), which has dominated the leadership field up to this day by offering an integration of three distinct leadership styles (laissez-faire, transactional, and transformational).The popularity of this theory has been further increased by the creation of the Multifactor Leadership Questionnaire (MLQ; Avolio et al., 1999), which constitutes a clear asset for practitioners and researchers who want to simultaneously assess all three forms of leadership behaviors.Indeed, besides the MLQ, most other measures only focus on a narrow subset of leadership behaviors (e.g., Tepper, 2000).The MLQ thus appears to be a very practical measure.However, research anchored in this framework has shown that the behaviors assessed by the MLQ display similar levels of predictive validity across several criteria (e.g., subordinates' performance) and has failed to document which of the less desirable forms of leadership (i.e., transactional and laissez-faire) have the most detrimental effects for subordinates (Judge & Piccolo, 2004;Skogstad et al., 2007).More generally, research based on this theoretical approach has mostly ignored the motivational processes and health consequences associated with these distinct leadership behaviors (Inceoglu et al., 2018).Finally, a more practical limitation is that the MLQ has been copyrighted (via Mindgarden) and is thus not easily accessible to researchers and practitioners.These limitations have obvious and important implications for researchers and practitioners.In trying to overcome these limitations, our research, based on SDT, does not claim to replace existing leadership conceptualizations and measures but rather to complement these well-established approaches with a new perspective that could contribute to a better understanding of supervisors' interpersonal behaviors and their implications for employees' motivation and well-being.

Existing instruments to measure supervisors' behaviors within the SDT framework
In work-related SDT research, autonomy supportive behaviors (for a review see Slemp et al., 2018) and, in a more limited way, autonomy thwarting (or controlling) behaviors (e.g., Gillet, Fouquereau, Forest, et al., 2012;Richer & Vallerand, 1995) have attracted the most attention.A few instruments have been developed to measure these interpersonal styles (e.g., WCQ; Baard et al., 2004;PASS-E;Moreau & Mageau, 2012).However, no multidimensional tool has been developed to concurrently measure work supervisors' supportive and thwarting behaviors in relation to the other two needs (competence and relatedness), alongside behaviors that support or thwart the need for autonomy.Indeed, although autonomy supportive behaviors (and the construct of autonomy support more generally) can theoretically be seen as contributing to the satisfaction of all three needs (Ryan & Deci, 2000), most measures developed to assess these behaviors have mainly focused on the need for autonomy (Slemp et al., 2018).The broader construct of need supportive behaviors (Bhavsar et al., 2019;Rocchi, Pelletier, Cheung, et al., 2017;Rocchi, Pelletier, & Desmarais, 2017) helps to better delineate how supervisors may contribute to the satisfaction of all three needs.Furthermore, despite the connection between need indifferent behaviors and laissez-faire leadership (Avolio et al., 1999;Lewin et al., 1939), as well as between need thwarting behaviors and various types of authoritarian leadership behaviors (e.g., Lewin et al., 1939;Tepper, 2000), no effort has yet been made to adapt Bhavsar et al.'s (2019)

A multidimensional perspective on supervisors' behaviors
In their original validation study conducted in the sport area, Bhavsar et al. (2019) found support for a three-factor structure of athletes' ratings of their coaches' need-related behaviors based on the type of behavior (supportive, thwarting, and indifferent) but not further broken down by needs (autonomy, competence, and relatedness).In other words, Bhavsar et al. (2019) found support for a three-factor model, which is interesting as it also matches Lewin et al.'s (1939) tripartite view of leadership behaviors.However, to clearly support the superiority of this solution, alternative representations, dominated by the type of need, the type of behavior, or both, have to be considered.
For instance, it could be argued that supervisors' interpersonal behaviors can represent an underlying continuum ranging from need supportive to need thwarting behaviors, with need indifferent behaviors falling in between these two extremes (i.e., a one-factor structure).Alternatively, it is possible for qualitatively distinct types of behaviors that are not mutually exclusive to co-occur.For instance, a supervisor can be understanding with their subordinates (autonomy support) while also inducing guilt to make them act a certain way (autonomy thwarting) and failing to provide them with a clear rationale for task engagement (autonomy indifference).Yet, both options are not mutually exclusive, and this possibility can be tested by examining psychometric multidimensionality.

Psychometric multidimensionality
In psychometric measurement, multidimensionality occurs when specific items tap into more than one construct (Morin, Arens, & Marsh, 2016;Morin, Boudrias, et al., 2016).Two types of construct-relevant multidimensionality are common in multidimensional instruments such as the TMIB-S.The first implies the evaluation of coexisting global and specific facets of a construct.For instance, recent research has explored the dimensionality of need-related behaviors, revealing that ratings on the TMIB-C (Bhavsar et al., 2019) and other measures of interpersonal behaviors (Interpersonal Behaviours Questionnaire; T oth-Kir aly et al., 2021) can be disaggregated into two independent components (global and specific).The first (global) reflects individuals' overarching perceptions of their supervisors' need-related behaviors across all dimensions.This global component thus captures the commonalities shared by the distinct need-related behaviors, irrespective of their positive or negative nature (Bhavsar et al., 2019 1 ;  T oth-Kir aly et al., 2021).This global factor reflects workers' general impression of their supervisor as a "rather good or bad leader."The second (specific) component reflects subscale-specific levels of need indifferent, thwarting, and supportive behaviors and/or of behaviors related to the needs for autonomy, competence, and relatedness, left unaccounted for by this global perception.This specific component reflects what is unique to each type of behavior and reflects deviations from employees' global perception of their supervisors' need-related behaviors. 2In other words, a supervisor might be perceived as generally nurturing and yet also engage in behaviors that thwart employees' need for autonomy or be indifferent to their need for competence.
A second form of multidimensionality involves cross-loadings, which depict reliable associations between items and more than one facet of a multidimensional construct.Indeed, workers' perceptions of need indifferent behaviors may also affect their responses to items created to measure need thwarting behaviors.Such cross-loadings emerge because ratings are inherently imperfect and also because supportive, indifferent, and thwarting behaviors are conceptually intertwined (Bhavsar et al., 2019).Previous results (Bhavsar et al., 2019;T oth-Kir aly et al., 2021) have demonstrated the value of incorporating cross-loadings to achieve an accurate representation of need-related behaviors.
These two types of multidimensionality are neglected in typical confirmatory factor analyses (CFA) where items are assumed to reflect a single factor (Morin et al., 2013).This limitation can be overcome with a combination of bifactor models, which disaggregate S-factors from the global component (G-factor; Morin et al., 2016;Myers et al., 2014), and exploratory structural equation models (ESEM), which estimate cross-loadings between items and conceptually related constructs based on a confirmatory specification of the main indicators of each factor (Morin et al., 2020).This combination, bifactor-ESEM, allows for the joint consideration of both forms of multidimensionality.
Ignoring either form of multidimensionality has important practical implications.On the one hand, when neglecting the global/specific nature of employees' ratings of their supervisors' behaviors, one is likely to erroneously conclude that each type of behavior plays a similar role in prediction, which would in fact mainly reflect the role played by the unmodeled global component (Morin, Arens, & Marsh, 2016).It would, therefore, be impossible to assess the unique effect of each behavior beyond the contribution of the global component (T oth-Kir aly et al., 2021).On the other hand, evidence has shown that neglecting cross-loadings may yield an erroneous assessment of the relations between a construct's dimensions (Asparouhov et al., 2015) and also of this construct's associations with other variables (Mai et al., 2018).This could explain why research based on the full-range model of leadership, which has typically ignored these two forms of multidimensionality, has shown that the behaviors assessed by the MLQ display similar levels of predictive validity across several criteria and tend to be highly correlated (Judge & Piccolo, 2004).In sum, overlooking construct-related multidimensionality may result in an inaccurate assessment of the psychometric properties of the measure and of the reality under study.As such, the second goal of this research was to investigate the multidimensionality underlying employees' ratings of their supervisor's need-related behaviors.

Supervisors' supportive, thwarting, and indifferent behaviors and employees' functioning
Leadership studies have typically focused on the role played by isolated leadership behaviors (Piccolo et al., 2012).In doing so, these studies have failed to consider how much of employees' functioning can be explained by other behaviors or how each specific type of behavior provides incremental predictive value, once the role of other types of behaviors is considered.In a metaanalysis of the MLQ, Judge and Piccolo (2004) noted that controlling for other leadership behaviors tended to undermine the incremental value of each specific type of behavior in a study.This is due, in part, to the high correlations observed among various leadership behaviors, which make it harder to detect unique effects.The adoption of a bifactor approach, allowing for the explicit disaggregation of employees' ratings into independent global and specific components, makes it possible to test this incremental contribution in a way that is not contaminated by inflated factor correlations.
The ability to jointly consider the relative contribution of these global and specific components has important theoretical and practical implications, especially when it comes to distinguishing between less desirable types of leadership behaviors.For instance, it is unclear whether authoritarian or passive types of leadership (i.e., need thwarting and indifferent behaviors) have clearly differentiated consequences.One could theoretically argue (Bhavsar et al., 2019) that need indifferent behaviors may not be as adversely experienced as need thwarting behaviors.Some evidence even indicates that the consequences of these two types of behaviors may differ in nature (Cheon et al., 2019), with need indifferent behaviors being more likely to predict outcomes reflecting a disinterest in one's work (e.g., job boredom) and need thwarting behaviors being more likely to associate with more adverse outcomes reflecting resource depletion (e.g., emotional exhaustion).Yet, these proposed differentiated effects remain to be clarified, which is the third and final objective of this research.More precisely, we sought to consider the unique effect of each form of supervisory behavior proposed by SDT (e.g., indifferent behaviors) while controlling for the specific and shared effects of the other two forms (e.g., thwarting and supportive behaviors).We also sought to control for the effects of alternative leadership behaviors proposed by other theoretical frameworks (abusive supervision, passive leadership, transformational leadership, and LMX), in order to examine the discriminant validity of the TMIB-S.
In doing so, this research also addresses a gap in the leadership literature, which has widely documented the consequences of supervisors' behaviors in terms of employee performance, but has generally overlooked the relationship between supervisors' behaviors and employees' welland ill-being (Inceoglu et al., 2018).In the few studies in which subordinates' psychological health was considered, scholars mostly tested it as a mediator to help explain the leadershipperformance relationship (e.g., Montano et al., 2017).In this research, we consider employees' psychological health as an end in and of itself, as opposed to a means to performance.Additionally, when considering employees' psychological health, leadership researchers have mostly looked at a very narrow set of outcomes (e.g., job satisfaction; see Inceoglu et al., 2018).Consequently, research has yet to document how supervisors' behaviors relate to a more diversified set of variables related to employees' well-and ill-being.This research addresses this gap by extending the nomological network associated with supervisory behaviors via the consideration of a wide array of outcomes.

OVERVIEW OF STUDIES 1 TO 3
This research examines (1) whether a recent SDT-based approach focused on need-related behaviors, developed in the sport context (Bhavsar et al., 2019), could provide an alternative theoretical perspective to guide our understanding of supervisors' behaviors and (2) the validity of a questionnaire anchored in this theoretical approach in the work context.In doing so, we examine (3) the multidimensionality of subordinates' ratings of their supervisors' need supportive, indifferent, and thwarting behaviors in order to better disentangle the global and specific components of these perceptions (Morin, Arens, & Marsh, 2016).Finally, we also seek to fill a gap in leadership research (see Inceoglu et al., 2018) by ( 4) investigating the motivational processes and well-being consequences associated with these different components of supervisors' behaviors.
The present research addresses these objectives through three studies.Study 1 seeks to provide validity evidence for the TMIB-S in a sample of English-speaking workers.Study 2 tests this measure's validity in a sample of French-speaking workers and offers a first test of the criterion-related (need satisfaction and frustration, job boredom, work engagement, and emotional exhaustion) and discriminant (abusive supervision, passive leadership, LMX, and transformational leadership) validity of the TMIB-S.Finally, Study 3 replicates and extends the investigation of criterion-related (need satisfaction and frustration, job satisfaction, job boredom, and work-related rumination) and discriminant (abusive supervision, passive leadership, LMX, and transformational leadership) validity of the TMIB-S in a new independent sample of English-speaking employees.

STUDY 1
This study sought to provide a preliminary examination of the factor structure of the TMIB-S in a sample of English-speaking workers.More precisely, this study contrasts different representations of TMIB-S ratings to achieve a more accurate representation of this measure's multidimensionality.

Procedure and participants
This research (Studies 1 to 3) was exempt from ethical review, according to local regulations.The Prolific Academic crowdsourcing service was used to recruit participants to Study 1.This online platform has been shown to provide quality data for researchers (Palan & Schitter, 2018;Peer et al., 2017).Participants were recruited based on several prescreening criteria: (1) being presently employed; (2) working part-time or full-time, (3) not working without pay nor being self-employed; (4) living and working in the USA (United States of America), Canada, or the United Kingdom (UK); and (5) speaking English as a first language.The general goal of the study was disclosed, and participants were assured of the anonymity of their responses.They were compensated £0.34 for completing a 3-minute questionnaire and provided written consent to participate before completing the survey.
Because participation was motivated by an external reward, we could not be confident that participants would respond to the questionnaire in a fully autonomously driven and attentive manner (Ryan & Deci, 2017).Therefore, in line with prior organizational research conducted using Prolific and involving monetary rewards (e.g., Lagios et al., 2021), the trustworthiness of participants' responses was ensured through an attention check included midpoint through the survey (i.e., "It is important that you pay attention to our survey, please tick strongly agree").Given the brevity of this survey (i.e., only 22 items, see below), only one attention check was required.Moreover, in line with prior Prolific-based research, a final control question asked participants whether they were presently employed by an organization (although Prolific allows to recruit participants based on specific criteria, sometimes their Prolific profile is not up to date, hence the necessity to ensure this criterion was met).They were guaranteed that this question only served scientific purposes and that their response would not change their compensation.Two participants reported not being currently employed and 14 failed the first attention check.These 16 participants were excluded.
A total of 350 participants (M age = 38.66;SD = 11.61;54.9% women) completed the survey.Participants lived and worked in the UK (66.9%),USA (26.9%), or Canada (6.3%); most of them had a permanent position (92%) and worked full-time (77.4%), for an average of 35.99 hours per week (SD = 9.56).Participants' average job tenure was 5.53 years (SD = 5.29), and, in average, their supervisors had been supervising them for 3.57 years (SD = 3.51).Roughly a third (38%) of the participants held supervisory positions.Participants mainly worked in the private sector (66.9%).

Measures
Supervisor interpersonal behaviors were measured with the 22-item TMIB-S.In order to make the scale validated by Bhavsar et al. (2019) suitable for the work domain, we adjusted the items by changing the word "activities" to "tasks" or "assignments" and by replacing the stem "My coach …" to "My supervisor …" (see Appendix A at the beginning of the supporting information file).Workers were requested to think of their interactions with their ongoing supervisor in order to report how much they agreed with each statement (1-strongly disagree; 7-strongly agree).Eight items measured need supportive behaviors (α = .95;e.g., autonomy support: " shows that he/she understands my perspective"; competence support: "recognizes my efforts and accomplishments"; and relatedness support: "shows care and concern"), eight items assessed need thwarting behaviors (α = .94;e.g., autonomy thwarting: "tries to control everything I do"; competence thwarting: "belittles my abilities"; and relatedness thwarting: "deliberately ignores me"), and six items measured need indifferent behaviors (α = .83;e.g., autonomy indifference: "is unresponsive to my opinions"; competence indifference: "can be disorganized"; and relatedness indifference: "is indifferent to how I feel").

Analyses
Mplus 8 (Muthén & Muthén, 2018) and the maximum likelihood robust (MLR) estimator were used in all analyses.The fit indices, parameter estimates, and standard errors, obtained via this estimator, are robust nonnormality.No missing responses were allowed in the online questionnaire.A sequence of a priori CFA and ESEM were conducted (see Table 1), in line with prior studies (e.g., Bhavsar et al., 2019;T oth-Kir aly et al., 2021).These models and the rationale underlying each of them are detailed in the Study 1 section of the supporting information file.Model fit was examined using the Tucker-Lewis index (TLI), the comparative fit index (CFI), and the root mean square error of approximation (RMSEA) (Hu & Bentler, 1999).Adequate model fit is indicated by RMSEA values below .08 and TLI/CFI value above .90,whereas excellent model fit is indicated by RMSEA values below .06 and TLI/CFI values above .95.Although we also report the chi-square test of exact fit (χ 2 ), this last indicator is not interpreted given its high sensitivity to minor misspecifications and sample-size dependency (e.g., Marsh et al., 2005).Nonetheless, model fit is not a sufficient criterion to select the optimal measurement model in the context of comparisons between CFA, ESEM, and bifactor alternatives (Morin et al., 2020), which need to also consider parameter estimates.First, CFA and ESEM solutions are compared, and the ESEM solution should be retained when it results in welldefined factors and reduced factor correlations when compared with CFA (Asparouhov et al., 2015;Morin et al., 2020).Second, the retained alternative (CFA or ESEM) is then compared with its bifactor equivalent, which should be retained when it results in the following: (1) higher model fit, (2) a G-factor that is well-defined (i.e., moderate to strong loadings), and (3) a subset of well-defined S-factors.The observation that multiple cross-loadings >.10 or .20 in the ESEM solution are smaller in the bifactor-ESEM solution also supports the bifactor-ESEM solution (Morin et al., 2020).
T A B L E 1 Goodness-of-fit statistics for the estimated measurement models (Study 1)

Results
When comparing CFA and ESEM solutions, model fit (reported in Table 1) and parameter estimates (reported in Table S1 of the supporting information file) supported the superiority of the three-factor ESEM solution (M2).Moving to the bifactor solutions, model fit (reported in Table 1) and parameter estimates (reported in Table S2 of the supporting information file) supported the superiority of this model's bifactor counterpart (i.e., bifactor-ESEM model with one G-factor and three S factors, M10).We provide a more extensive discussion of the superiority of M2 and M10, when compared with other solutions, in the Study 1 section of the supporting information file.More specifically, the three-factor ESEM solution (M2) resulted in well-defined factors (λ = .362to .998,M λ = .717for supportive behaviors; λ = .562to .969,M λ = .762for thwarting behaviors and λ = .396to .779,M λ = .602for indifferent behaviors).Similarly, the bifactor-ESEM solution with one G-factor and three S-factors (M10) included a G-factor that was welldefined and reflected employees' global perceptions of their supervisor's need-related interpersonal behaviors, defined by negative loadings from the supportive behaviors items (λ = À.774 to À.633, M λ = À.708), positive loadings from the thwarting behaviors (λ = .651to .873,M λ = .796),and slightly smaller positive loadings from the indifferent behaviors (λ = .344to .859,M λ = .566).Because of this specific pattern of loadings (positive for need thwarting and indifferent behaviors, and negative for need supportive behaviors), we hereafter refer to this Gfactor as reflecting supervisors' global levels of need hampering behaviors.Moreover, although a few items reflected the G-factor more strongly than their S-factor, all S-factors maintained a meaningful amount of specificity (supportive: λ = .187to .577,M λ = .417;thwarting: λ = .055to .441,M λ = .264,and indifferent: jλj = .191to .591,M jλj = .381).Taken together, these results supported the bifactor-ESEM model with one G-factor and three S-factors (M10).Additional information on this solution is provided in Study 1 section of the supporting information file.

Discussion
This study provided preliminary support for the ability of SDT (Ryan & Deci, 2017) to offer an alternative integrative understanding of leadership behaviors (Anderson & Sun, 2017).Specifically, we provided preliminary evidence supporting the factor validity of a measure of supervisors' need supportive, thwarting, and indifferent behaviors (the TMIB-S), adapted from the sport context (TMIB-C; Bhavsar et al., 2019), in a sample of English-speaking workers.In doing so, we expand upon prior instruments in the work setting (e.g., Baard et al., 2004;Moreau & Mageau, 2012), by offering a multidimensional measure encompassing not only autonomysupportive and autonomy thwarting (controlling) behaviors from supervisors but also their supportive and thwarting behaviors toward the other psychological needs of their employees (relatedness and competence), together with their indifferent behaviors toward employees' psychological needs.
We investigated the multidimensionality of employees' ratings of their supervisors' behaviors and found support for a bifactor-ESEM representation.Although prior research has represented interpersonal behaviors as multidimensional (e.g., Rocchi, Pelletier, Cheung, et al., 2017;Rocchi, Pelletier, & Desmarais, 2017;T oth-Kir aly et al., 2021) or as global (e.g., Bartholomew, Ntoumanis, Ryan, Bosch, & Thøgersen-Ntoumani, 2011;Gillet, Fouquereau, Forest, et al., 2012) constructs, our study supports a third representation.This third option bridges the gap between the two previous ones: Interpersonal behaviors may be represented through a global entity (global levels of need hampering behaviors; G-factor) coexisting with three specific facets (need supportive, indifferent, and thwarting behaviors; S-factors).It should be emphasized that, though we labeled the G-factor as reflecting global levels of "need hampering behaviors" based on the valence of the factor loadings (negative for need supportive behaviors and positive for need indifferent and need thwarting behaviors), this factor still reflects the whole range of supervisors' positive and negative need-related interpersonal behaviors.As such, although higher scores on this G-factor reflect exposure to need hampering supervisors, lower scores reflect exposure to need nurturing supervisors.Moreover, it is important to keep in mind that the S-factors cannot be interpreted as one would interpret a first-order factor.Rather, one should keep in mind that these S-factors reflect the extent to which subordinates' perceptions of each interpersonal style (supportive, thwarting, and indifferent) deviate from their global perception of their supervisor's interpersonal behaviors.In other words, S-factors reflect what is unique to each interpersonal style, over and above employees' global perception of their supervisor's interpersonal behaviors.
Despite the superiority of this bifactor-ESEM solution, the three-factor ESEM solution was also, though to a lesser extent, satisfactory.This alternative model matched Bhavsar et al.'s (2019) results, suggesting that researchers and practitioners who require a more traditional representation of each type of behavior could rely on this alternative representation (M2).

STUDY 2
Although Study 1 provided validity evidence for the TMIB-S, this evidence was limited to an initial demonstration of factor validity in an English-Speaking sample.Yet, leadership is a global issue, requiring measures to be adaptable across linguistic populations (e.g., Millsap, 2011).Moreover, the universality of need-related behaviors across contexts (e.g., sports and work) and cultures is a key tenet of SDT (Vansteenkiste et al., 2020).As such, the ability to establish the validity of another linguistic version of the TMIB-S is critical to support cross-cultural comparisons based on this instrument.Study 2 examines the validity of a French version of the TMIB-S by verifying whether the bifactor-ESEM solution retained in Study 1 would be replicated in Study 2 and by documenting the invariance of this solution across samples of English-and French-speaking employees.
The ability to test for other forms of validity (i.e., discriminant and criterion-related) of employees' responses to the TMIB-S also constitutes a crucial step toward documenting the potential utility of this measure of supervisory behaviors.Study 2 addresses these issues.First, the capacity to demonstrate the added-value of a measure beyond that of already established ones represents a critical step toward the establishment of the measure's discriminant validity (Judge & Piccolo, 2004).Therefore, we examined the discriminant validity of the TMIB-S by controlling for four other well-established leadership behaviors: (1) abusive supervision (i.e., "sustained display of hostile verbal and nonverbal behaviors, excluding physical contact"; Tepper, 2000, p. 178), (2) passive leadership (i.e., "a passive mode of reaction or the lack of response from the leader in the face of a variety of situations"; Chênevert et al., 2013, p. 278), (3) LMX (high quality supervisor-subordinate relationships, implying perceptions of support, guidance, and appreciation from one's supervisor; Graen & Uhl-Bien, 1995), and (4) transformational leadership (i.e., "the extent to which a leader is visionary, innovative, supportive, participative and worthy of respect"; Carless et al., 2000, p. 401).Showing that these established and widely studied leadership styles do not contribute beyond what can be explained by the supportive, thwarting, and indifferent styles measured by the TMIB-S would support the TMIB-S as a useful alternative to existing leadership measures.Moreover, showing that scores on the TMIB-S are equivalent, or even more effective, in prediction than other types of leadership behaviors would provide an opportunity to overcome the theoretical, methodological, and practical limitations associated with the tendency to measure positive, negative, and passive leadership with instruments developed from distinct theoretical frameworks (Klasmeier et al., 2021;Xu et al., 2015).Indeed, such instruments rely on different types of items, instructions, and response scales and were created through distinct types of procedures.By using scales developed so differently, one risks measuring constructs that are hardly comparable or overlapping, ending up with a poor reflection of the reality under study.
In addition, Study 2 investigates the criterion-related validity of the TMIB-S using several indicators of employees' well-(work engagement) and ill-being (emotional exhaustion and job boredom).In doing so, we aim to examine the unique implications of need indifferent, supportive, and thwarting behaviors in terms of employee functioning.Importantly, some have previously argued that passive leadership styles (e.g., need indifferent behaviors) could be as destructive as more actively negative types (e.g., need thwarting behaviors) of leadership (Skogstad et al., 2007).Others have shown that passive leadership behaviors were as important as other behaviors in predicting various outcome variables and called for more research on these passive forms of supervisory behaviors (Judge & Piccolo, 2004).Yet, empirical studies documenting the psychological health consequences of passive forms of leadership behaviors remain scarce.As such, organizations, supervisors, and researchers may remain unaware of the potentially devastating effects of such behaviors.
We chose to measure work engagement, emotional exhaustion, and job boredom in this study, as they have been shown to share associations with supervisory behaviors (Breevaart et al., 2016;Krasniqi et al., 2019;Whitman et al., 2014) and to have important consequences for organizations, including sickness absence, turnover, counterproductive work behaviors, job performance and service quality (Schaufeli & Salanova, 2014).Based on prior research, we expected that work engagement, a work-related indicator of well-being that is characterized by vigor, dedication, and absorption (Schaufeli et al., 2019), would result more importantly from need supportive interpersonal behaviors (Bhavsar et al., 2019;Slemp et al., 2018).Emotional exhaustion being an intense state of fatigue and energy depletion, resulting from chronic exposure to work stressors such as work overload, emotional demands, and interpersonal conflict (Schaufeli & Salanova, 2014), we expected it to be best predicted by thwarting behaviors as they imply pressure, criticism, and rejection.Finally, job boredom also reflects an undesirable outcome, yet not one that is as actively adverse as emotional exhaustion.Because job boredom is defined as "an unpleasant state of relatively low arousal and dissatisfaction" (Mikulas & Vodanovich, 1993, p. 3) resulting from a work environment that is not sufficiently stimulating and challenging (Schaufeli & Salanova, 2014), we expected it to be best predicted by need indifferent behaviors, which reflect an understimulating supervisory style.
Finally, we considered the psychological mechanisms that may explain these associations.On the basis of SDT (Bartholomew, Ntoumanis, Ryan, Bosch, & Thøgersen-Ntoumani, 2011), we proposed that the relations between supervisory behaviors (supportive, thwarting, and indifferent) and employees' work engagement, job boredom, and emotional exhaustion would be mediated by employees' need satisfaction and frustration.Based on prior research (Bhavsar et al., 2019;Gillet, Fouquereau, Forest, et al., 2012), we anticipated that need supportive behaviors would best predict need satisfaction, which would itself associate more importantly with work engagement.Contrastingly, we hypothesized that need frustration would result more importantly from need thwarting behaviors and to a lesser extent from need indifferent behaviors (Bhavsar et al., 2019) and in turn associate more strongly with emotional exhaustion and, to a lesser extent, job boredom.

Procedure and participants
Participants were recruited by trained research assistants across a variety of organizations located in France.Participants had to be employed, in France, and to have a supervisor.Eligible participants received an email clarifying the general goal of the research and offering them to take part in an online survey, while explaining that participation was voluntary.They were also guaranteed that their responses would remain anonymous and were asked to provide active written consent.Given that each of these participants was personally and carefully recruited by trained research assistants and given that participation was most likely driven by autonomous motives, we were confident that participants would complete the questionnaire with attention (Ryan & Deci, 2017) and thus did not include any attention checks, in line with prior research on similar constructs (e.g., Bhavsar et al., 2019).
A total of 512 French participants (M age = 31.12;SD = 11.74;59.8% women) completed the survey.A majority of the participants had permanent positions (68.6%) and worked full-time (82.2%) for an average of 37.28 hours a week (SD = 10.19).Participants' average job tenure was 5.67 years (SD = 7.40), and they had been working under the management of their current supervisor for an average of 3.56 years (SD = 4.96).Roughly a fourth (22.1%) of the participants held supervisory positions.Participants mainly worked in the private sector (71.3%).

Measures
Through a translation back-translation procedure (Beaton et al., 2000), measures not available in French (need-related behaviors, abusive supervision, job boredom) were adapted from English.
Supervisor interpersonal behaviors were measured with the French adaptation (see Appendix B at the beginning of the supporting information file) of the questionnaire described in Study 1 (need indifferent behaviors: α = .86;need thwarting behaviors: α = .92;and need supportive behaviors: α = .93).
Abusive supervision was measured with three items (α = .74;e.g., "Tells me my thoughts or feelings are stupid") adapted from Tepper (2000) by Detert et al. (2007).Participants indicated how frequently their current supervisor engaged in a series of behaviors (1-never to 5-very often).
Passive leadership was assessed with five items (α = .87)adapted in French (Chênevert et al., 2013).Participants were asked to think of their supervisor to rate their degree of agreement (e.g., "My direct supervisor avoids making decisions"; 1-strongly disagree to 7-strongly agree).
LMX was measured with seven items (α = .91;Graen & Uhl-Bien, 1995) adapted to French by Montani et al. (2017).Respondents were asked to think about their experience with their supervisor to reply using a 5-point response scale, for which the anchors differed across items (e.g., "How well does your supervisor understand your job problems and needs?"; 1-not a bit to 5-a great deal).
Need satisfaction was measured with a nine-item scale created in French by Gillet et al. (2008).Items were contextualized with the stem "At work…" Three three-item subscales were used to measure the satisfaction of the need for autonomy (α = .76;e.g., "I have the opportunity to make decisions about the tasks that I have to perform"), competence (α = .72;e.g., "I feel like I am able to meet the demands of the tasks that I have to perform"), and relatedness (α = .83;e.g., "I get along well with the people whom I interact with").Items were rated from 1 (strongly disagree) to 7 (strongly agree).
Need frustration was evaluated with the original French nine-item scale created by Gillet, Fouquereau, Lequeurre, et al. (2012).Three three-item subscales were used to measure the frustration of the need for autonomy (α = .82;e.g., "I feel forced to behave in a certain way"), competence (α = .78;e.g., "It happens that I hear things that make me feel incompetent"), and relatedness (α = .78;e.g., "I think other people hate me").Participants rated their agreement with each statement (1-strongly disagree to 7-strongly agree).
Work engagement was measured via the three-item version of the Utrecht Work Engagement Scale (UWES-3: Schaufeli et al., 2019; α = .86;e.g., "At my work, I feel bursting with energy").Participants indicated their response on a frequency scale (1-never to 7-always).
Emotional exhaustion was evaluated with a three-item subscale (α = .74;e.g., "I feel I am unable to be sensitive to the needs of coworkers and customers") from the French version of the Shirom-Melamed Burnout Measure (Sassi & Neveu, 2010), rated on a frequency scale (1-never to 7-always).

Analyses
Analyses were conducted using the same procedures as in Study 1.The same set of measurement models already described in Study 1 was first used to identify the optimal measurement structure for the TMIB-S.We then proceeded to test the invariance of this measurement model (Millsap, 2011) across samples from Studies 1 and 2 (more details can be found in the Study 2 section of the supporting information file).Before moving to predictive models, we also contrasted a priori measurement models to identify the optimal structure for the measures of need fulfillment (need satisfaction and frustration), following a sequence of models described in T oth-Kir aly et al. (2018).These additional analyses appear in Study 2 section of the supporting information file and replicated T oth-Kir aly et al.'s (2018) results, leading us to retain a bifactor-ESEM model encompassing one global need fulfillment G-factor and six S-factors (the frustration and satisfaction of the needs for relatedness, competence, and autonomy).Predictive analyses were then conducted, based on the model described in Figure 1: We contrasted a total mediation solution with a partial mediation alternative.We analyzed mediation by estimating the indirect effects of supervisors' behaviors on job boredom, work engagement, and emotional exhaustion, as mediated by employees' need fulfillment (Morin et al., 2013).
As a final step, the discriminant validity of the TMIB-S was examined relative to measures of passive leadership, abusive supervision, LMX, and transformational leadership.This final step was conducted by incorporating these variables to the final retained mediation model as additional predictors, and by contrasting four alternative models: (1) null: The effects of these additional variables on the mediators and outcomes were constrained to be exactly 0; (2) partial mediation: The effects of these additional variables on the mediators and outcomes were freely estimated; (3) total mediation: The effects of these additional variables on the mediators, but not the outcomes, were freely estimated; and (4) direct: The effects of these additional variables on the outcomes, but not the mediators, were freely estimated.We conducted these analyses independently 3 for the four variables (passive leadership, abusive supervision, LMX, and transformational leadership) to limit possible multicollinearity due to their high correlations (r = .54to .88).Latent correlations among all variables included in this study are presented in Table S4 of Study 2 section of the supporting information file.

Results
The fit of the preliminary measurement models appears in the upper section of Table S3 in the supporting information file.Additional details regarding these models are disclosed in the Study 2 section of the supporting information file.These results reinforced Study 1's conclusions in supporting the superiority of the bifactor-ESEM solution with one G-factor and three S-factors (M10).This solution (M10) was therefore retained and tested for measurement invariance.Results from these analyses provided support for the invariance (configural, weak, strong, partial strict, variance-covariance, and means) of this solution across the English (Study 1) and French (Study 2) respondents (see Table S3 of the supporting information file, middle section).In the model of partial strict invariance, the uniquenesses of two items had to be allowed to be freely estimated across samples, indicating the presence of slightly higher measurement errors associated with these two items in the French questionnaire (i.e., Study 2).
Model fit from the different predictive models was satisfactory and similar for the partial and total mediation models, albeit slightly better for the former relative to the latter (see Table S3 of the supporting information file, middle section).Statistically significant direct links between the predictors (supervisors' interpersonal behaviors) and the outcomes were also evidenced in the partial mediation model, which was thus retained for interpretation.Results from this model are detailed in Table 2 and in the Study 2 section of the supporting information file (see Figure S1 for an illustration) and thoroughly examined in the following discussion section.Taken together, results suggested the presence of 18 potential indirect effects, which were all supported through tests of statistical significance, via bias-corrected bootstrap confidence intervals (see Table S8 of Study 2 section of the supporting information file).
Results from the tests of discriminant validity are reported in the Study 2 section of the supporting information file (also see Table S3, bottom section) and examined in detail in the following discussion..078

Note:
This table is divided into three horizontal parts: The upper part displays associations between predictors (on the top) and the mediators and outcomes (on the left); the middle and lower parts display associations between mediators (on the top) and the outcomes (on the left).The statistical significance of the differences in the size of the path coefficients for variables predicted in a statistically significant manner by more than one other variable from the same set (predictors or mediators) were tested using the Mplus Model Constraint Function.For any single line, coefficients with an identical superscript letter do not differ from one another in a statistically significant manner (p

Discussion
Study 2 provided replication evidence for the bifactor-ESEM factor structure of the TMIB-S identified in Study 1, showing that a similar structure appeared to provide an equally adequate representation of the underlying structure of the French version of this instrument.In doing so, Study 2 further supported the ability of SDT to be a valuable theoretical framework to understand leadership behaviors across countries and cultures (Vansteenkiste et al., 2020).Furthermore, and also supporting results from Study 1, the alternative three-factor ESEM structure (M2) also seemed to represent an acceptable alternative for researchers seeking to achieve a more traditional type of measurement.Study 2 also provided preliminary support for the criterion-related validity of the TMIB-S.In this regard, our results showed that global levels of need hampering behaviors directly and indirectly (through global need fulfillment and specific autonomy satisfaction) predicted more emotional exhaustion and boredom and less work engagement.Our results also revealed that, once the variance explained by global need hampering behaviors was considered, specific need supportive, thwarting, and indifferent behaviors also predicted various indicators of psychological functioning.More precisely, specific levels of imbalance in each of the supportive, thwarting, and indifferent behaviors, relative to all other types of behaviors, also explained unique variability in outcomes' levels, over and above that already explained by global level of need hampering behaviors.This has important implications for leadership researchers.Indeed, modeling the global/specific nature of subordinates' perceptions of their supervisors' need-related interpersonal behaviors makes it possible to capture the fact that the specificities inherent to each interpersonal style have distinct consequences.Interestingly, we observed that, whereas the effects of the specific levels of need supportive and thwarting behaviors on psychological functioning were indirect (mediated via global and specific levels of need fulfillment), those of need indifferent behaviors were direct.
On the one hand, as expected, need supportive behaviors predicted higher global levels of need fulfillment and specific relatedness satisfaction, which in turn both predicted more work engagement and less emotional exhaustion and boredom.Likewise, need thwarting behaviors predicted lower global need fulfillment, which in turn led to less work engagement and more emotional exhaustion and boredom.On the other hand, need thwarting behaviors also unexpectedly predicted higher specific relatedness satisfaction, which was in turn related to more work engagement and less emotional exhaustion and boredom.This result could suggest that exposure to supervisors that are perceived as displaying a high level of need thwarting behaviors may encourage employees to seek fulfillment of their need for relatedness from other work-related sources, which may in turn protect and promote their psychological functioning.In addition, although need thwarting behaviors, as expected, were also related to higher levels of specific competence frustration, these levels were, in turn, surprisingly related to higher work engagement.This suggests that subordinates who feel that their need for competence is frustrated at work as a result of their supervisor's need thwarting behaviors might become more engaged at work.A possible explanation for these unexpected results could be that some subordinates may engage in effective coping strategies, such as directly confronting their supervisor about their thwarting behaviors (e.g., Frieder et al., 2015) or using ingratiation when interacting with their supervisor (Harvey et al., 2007), as tactics to gain an increased sense of control over their work environment when facing need thwarting supervisors.Yet, such coping strategies require resource-mobilization (Frieder et al., 2015), which is why subordinates might still feel energized at work (i.e., work engagement).However, we did not consider these factors in the present study, and this hypothesis would need to be verified in future research.An alternative explanation for this unexpected result could be that employees who perceive their need for competence to be frustrated could engage more into their work as a way to compensate and increase their mastery.For instance, Sheldon and Gunz (2009) showed that the more individuals perceive their needs for competence and relatedness to be threatened, the more they desire to experience the satisfaction of these needs and orient their behaviors toward that goal.Other studies have provided support for this "need restoration" process (e.g., Radel et al., 2013), showing that the deprivation of specific needs does have a motivational force yielding behaviors aiming at restoring the deprived needs.Yet, because this need restoration process occurs over time (Radel et al., 2013) and because our study is cross-sectional, this explanation remains speculative.These possibilities thus require verification in future longitudinal studies.
Contrasting with these observations, need indifferent behaviors only had direct associations with the indicators of psychological functioning.Specifically, need indifferent behaviors were positively associated with employees' emotional exhaustion and job boredom and negatively associated with their work engagement.These direct associations emphasize the value of considering this third type of supervisory behaviors when seeking to understand the drivers of employees' psychological functioning and reinforce the idea that a lack of leadership might be as harmful as the presence of inadequate leadership behaviors (Judge & Piccolo, 2004).Moreover, as expected, need indifferent behaviors were strongly associated with job boredom, thus supporting the idea that understimulating supervisory behaviors can predict "an unpleasant state of relatively low arousal and dissatisfaction" (Mikulas & Vodanovich, 1993, p. 3).Finally, results also supported the distinctive psychological implications of exposure to need thwarting versus indifferent behaviors in showing that, whereas the former had implications for employees' levels of psychological need fulfillment, the latter did not significantly predict this psychological mediator.This result thus suggests that need indifferent behaviors, rather than contribute to employees' need frustration or satisfaction, might rather be associated with a distinct need state.Indeed, recent research suggested that need unfulfillment (i.e., a negative psychological experience involving feelings of disconnection, dullness, and uncertainty) could be represented as a third and distinctive need state when examined together with employees' need frustration and satisfaction (Cheon et al., 2019;Huyghebaert-Zouaghi et al., 2021).
Finally, Study 2 provided preliminary support for the discriminant validity of the TMIB-S relative to ratings of abusive supervision, passive leadership, LMX, and transformational leadership, which did not share any statistically significant association with the outcomes once the effects of supervisors' interpersonal behaviors were taken into account.Passive leadership did share one statistically significant association with a mediator (i.e., autonomy satisfaction) which is not surprising because passive supervisors offer no structure, thus allowing employees to make decisions (too) autonomously.Yet, out of a total of 40 specified links between the four alternative leadership measures and the 10 mediator/outcome factors included in our study, only this one association turned out to be significant, whereas many significant associations were found between the behaviors measured by the TMIB-S and the mediators and outcomes.These results thus support the added-value of the TMIB-S, when compared with wellestablished measures of leadership, in predicting various outcomes of interest. 4,5 This meticulous consideration of the differentiated effects of several forms of interpersonal behaviors, while controlling for other leadership behaviors (Piccolo et al., 2012), strongly supports the discriminant and criterion-related validity of the TMIB-S.This study also extends our understanding of the consequences of passive types of leadership (Judge & Piccolo, 2004;Skogstad et al., 2007).

STUDY 3
Study 3 builds on Study 2, seeking to accomplish the following objectives: (1) provide replication evidence for the bifactor-ESEM factor structure of the TMIB-S among a new sample of English-speaking workers, (2) extend the nomological network of the TMIB-S via further analyses of its criterion-related validity, and (3) provide further evidence for the discriminant validity of the TMIB-S.
In terms of criterion-related validity, we attempted to replicate findings from Study 2 by measuring need satisfaction and frustration using different measures (i.e., validated in a different culture and language) than those used in Study 2.Moreover, we retained job boredom given its less actively adverse nature on the ill-being continuum and its unique association with need indifferent behaviors identified in Study 2. In addition, to expand upon results from Study 2, we also considered employees' ratings of job satisfaction (as an alternative manifestation of wellbeing) and work-related rumination (as an alternative manifestation of ill-being).This decision is underpinned by the documented associations between various forms of supervisory behaviors and these outcomes (e.g., Gillet, Fouquereau, Forest, et al., 2012;Pereira et al., 2015) and by the impact of these outcomes in terms of individual (e.g., problem-solving, impaired concentration) and organizational (e.g., job performance, organizational commitment) functioning (Lu et al., 2019;Lyubomirsky & Tkach, 2003).Based on prior research, we expected job satisfaction to be best predicted by need supportive behaviors through the mediation of need satisfaction (Gillet, Fouquereau, Forest, et al., 2012).We also expected rumination (i.e., the experience of recurring and invasive thoughts about work-related matters in the absence of job demands necessitating these thoughts; Martin & Tesser, 1996), to be best predicted by need thwarting behaviors through the mediation of need frustration.Indeed, when employees face a supervisor who rejects, criticizes, and pressures them, they may experience feelings of exclusion, uselessness, and coercion, which could in turn spill over into their off-job hours in the form of ubiquitous thoughts.Finally, as in Study 2, we expected job boredom to share stronger and more direct associations with indifferent behaviors.
As in Study 2, discriminant validity was assessed in relation to measures of actively adverse (abusive supervision) and passive/neglecting (passive leadership) forms of leadership, as well as to more desirable forms of leadership captured by LMX and transformational leadership.

Procedure and participants
The Prolific Academic online platform was used to recruit participants (compensated £1.00 for completing a 10-minute questionnaire), relying on inclusion criteria and on a procedure identical to Study 1.Because participation was, as in Study 1, based on external rewards, the same attention check and control question used in Study 1 were included in Study 3. Given that this survey was longer than the one used in Study 1, an additional attention check was also included (i.e., "It is important that you pay attention to our survey, please tick strongly disagree"), allowing us to divide the survey into three sections of roughly the same length, each with their own attention check.This made it possible to control for participants' possible drop in attention in the last sections of the survey.Thirty participants failed the first check, 21 failed the second, and two reported not being currently employed.In total, 53 errors were made by 43 individuals who were excluded from the analyses.
The final sample included 449 participants (M age = 38.91;SD = 10.95;54.9% women).Participants lived and worked either in the UK (70.6%),USA (26.5%), or Canada (2.9%), and most of them had permanent positions (92.9%) and worked full-time (100%) for an average of 36.04 hours per week (SD = 9.17).Respondents' job tenure was of 6.22 years in average (SD = 5.01), and they had been working for their supervisor for 4.18 years in average (SD = 3.85).Roughly half of the participants held supervisory positions (45.9%).Participants mainly worked in the private sector (62.4%).
Need satisfaction.The 18-item scale validated by Van den Broeck et al. (2010) was used.Three six-item subscales were used to measure the satisfaction of the need for autonomy (α = .81;e.g., "I feel free to do my job the way I think it could best be done"), competence (α = .90;e.g., "I feel competent at my job"), and relatedness (α = .90;e.g., "At work, I feel part of a group"), on a Likert scale (1-strongly disagree; 7-strongly agree).
Need frustration.The 12-item scale developed by Bartholomew, Ntoumanis, Ryan, and Thøgersen-Ntoumani (2011) was here used and contextualized with the stem "At work …" Three four-item subscales, respectively, assessed the frustration of the needs for competence (α = .89;e.g., "There are situations where I am made to feel inadequate"), autonomy (α = .89;e.g., "I feel prevented from making choices with regards to the way I work"), and relatedness (α = .84;e.g., "I feel I am rejected by those around me"), on a Likert scale (1-strongly disagree; 7-strongly agree).
Job satisfaction.Participants were asked to rate their degree of satisfaction with their present job (1-dissatisfied to 4-satisfied) on a single item measure (Huyghebaert-Zouaghi et al., 2021).
Work-related rumination was measured via two items (α = .83;e.g., "I worry about things that have to be done at work"; de Bloom et al., 2014) asking employees to indicate how much they agreed with both statements (1-strongly disagree; 5-strongly agree).

Analyses
Analyses were conducted using the same procedure as in Studies 1 and 2 (measurement models, invariance, predictive analyses, and discriminant validity).The details of these analyses are disclosed in the supporting information file (Study 3 section), while the model tested in the predictive analyses is also illustrated in Figure 1 (correlations are reported in Table S10 of the supporting information file).

Results
Table S9 of the supporting information file (upper section) details the model fit from the measurement models used to investigate the optimal structure for the TMIB-S and its invariance across samples.Additional details regarding these models are disclosed in the Study 3 section of the supporting information file.These results corroborated the conclusions from Studies 1 and 2, leading us to retain M10 (i.e., the bifactor-ESEM solution encompassing one G-factor and three S-factors).Measurement invariance was thus examined based on this solution (M10).This solution demonstrated complete invariance across the samples of English-respondents used in Studies 1 and 3.
The fit of the predictive models is reported in Table S9 of the supporting information file (middle section).These results revealed that the total mediation model was unable to achieve acceptable levels of fit based on the TLI value, whereas the partial mediation model was able to do so.Examination of the parameter estimates from these models revealed statistically significant direct effects of the predictors (supervisors' interpersonal behaviors) on the outcomes.The partial mediation model was therefore selected for interpretation.Results from this model are detailed in Table 3 (for a more extensive presentation, see Study 2 section of the supporting information file and Figure S2 for an illustration) and thoroughly examined in the following discussion section.These results suggested the presence of 19 potential indirect effects, which were all supported through tests of statistical significance, via bias-corrected bootstrap confidence intervals (see Table S14 of Study 3 section of the supporting information file).
Results from the tests of discriminant validity are reported in the Study 3 section of the supporting information file (also see Table S9, bottom section) and examined in detail in the following discussion.

Discussion
Study 3 replicated the bifactor-ESEM structure of the TMIB-S demonstrated in Studies 1 and 2 and further established the adequacy of the English version of this instrument.In addition, as in Studies 1 and 2, Study 3 showed the adequacy of the alternative three-factor ESEM structure (M2) retained by Bhavsar et al. (2019).Study 3 also extended the nomological network of the TMIB-S, revealing (as in Study 2), that all four factors shared well-differentiated relations with the covariates.These results supported the idea that need hampering behaviors exert their effects on the least desirable outcome (i.e., work-related rumination) in part through "the darker side" of psychological needs (i.e., need frustration; Bartholomew, Ntoumanis, Ryan, Bosch, & Thøgersen-Ntoumani, 2011;Bartholomew, Ntoumanis, Ryan, & Thøgersen-Ntoumani, 2011).Importantly, results also showed that, once the contribution of global need hampering behaviors was considered, specific need supportive, indifferent, and thwarting behaviors also meaningfully predicted various indicators of psychological functioning.
As expected, need supportive behaviors indirectly predicted lower levels of rumination and job boredom and higher levels of job satisfaction through global need fulfillment.The effects of need supportive behaviors on job satisfaction and boredom were also explained by employees' specific autonomy satisfaction, thus highlighting the unique role played by the need for autonomy in explaining the beneficial effect of positive contextual factors (supportive behaviors).Specific need supportive behaviors also directly predicted lower levels of rumination and, unexpectedly, shared direct positive associations with higher levels of job boredom.This surprising result thus suggests that too much support might directly predict boredom.Keeping in mind the bifactor nature of the measurement model underpinning TMIB-S ratings, this result may suggest that supportive behaviors alone do not suffice to prevent maladaptive functioning.Rather, they may need to be complemented with low levels of indifferent and thwarting behaviors, as reflected by low global levels of need hampering behaviors.In other words, this result could indicate that supervisors who wish to prevent job boredom among their subordinates should not only try to enact more need supportive behaviors but should also simultaneously make an effort to reduce their need thwarting and indifferent behaviors, to help subordinates to perceive them as globally displaying lower levels of need hampering behaviors.It also implies that researchers who wish to get a comprehensive understanding of the effects of supervisors' need-related behaviors should simultaneously measure all three types of behaviors.
Specific levels of need thwarting behaviors only indirectly related to more job satisfaction and less job boredom, via the mediating role of employees' specific relatedness satisfaction.This unexpected result replicates a similar association already identified in Study 2, suggesting that exposure to need thwarting supervisors may encourage employees to seek fulfillment of their need for relatedness from other work-related sources.Moreover, specific levels of thwarting behaviors also shared no statistically significant association with the most deleterious outcome considered in this study (work-related rumination).It may be that need thwarting behaviors need to combine with other facets of interpersonal behaviors (e.g., high levels of global need hampering and specific indifferent behaviors, and low levels of specific supportive behaviors) in order to create the most adverse consequences, a hypothesis which we further address in the General Discussion.
Specific need indifferent behaviors were indirectly linked to more job boredom and rumination and less job satisfaction through the mediating role of need fulfillment.Although no evidence for such an indirect effect was found in Study 2, Study 3 suggests that, at least in some circumstances, the effect of need indifferent behaviors on employees' psychological functioning may be explained by employees' psychological need frustration and satisfaction.Yet, and supporting results from Study 2, specific need indifferent behaviors were also directly associated with higher levels of job boredom.These results highlight how predictive this passive leadership style can be (Skogstad et al., 2007).
Finally, adding support for the TMIB-S 0 discriminant validity, LMX, transformational leadership, abusive supervision, and passive leadership did not relate to any of the outcome variables once the effects of supervisors' interpersonal behaviors were taken into account.However, LMX did share two associations with the mediators, namely, higher global need fulfillment and specific competence satisfaction.Similarly, transformational leadership was associated with higher specific levels of competence satisfaction and lower specific levels of competence frustration.These results are neither surprising, nor concerning, as (1) they are consistent with prior research showing LMX and transformational leadership to positively associate with need satisfaction (Kovjanic et al., 2013;Lian et al., 2012); (2) LMX and transformational leadership did not significantly predict any of the outcomes included in our study (job satisfaction, job boredom, and work-related rumination); and (3) out of a total of 40 specified links between the four alternative leadership measures and the 10 mediator/outcome factors included in our study, only four associations turned out to be significant, while numerous significant associations were found between the supervisory behaviors measured by the TMIB-S and those same mediator/ outcome factors.Taken together, these results thus show that the TMIB-S competes well with other measures of leadership in predicting various outcomes..040

Note:
This table is divided into three horizontal parts: The upper part displays associations between predictors (on the top) and the mediators and outcomes (on the left); the middle and lower parts display associations between mediators (on the top) and the outcomes (on the left).The statistical significance of the differences in the size of the path coefficients for variables predicted in a statistically significant manner by more than one other variable from the same set (predictors or mediators) were tested using the Mplus Model Constraint Function.For any single line, coefficients with an identical superscript letter do not differ from one another in a statistically significant manner (p

GENERAL DISCUSSION
The present work aimed to replicate recent conceptual and methodological developments from the sport context (Bhavsar et al., 2019) to offer a first-in-the-literature work-specific instrument, anchored in SDT (Ryan & Deci, 2017), allowing to simultaneously measure perceived supervisors' supportive, indifferent, and thwarting behaviors toward their employees' psychological needs.
A tripartite approach of supervisory behaviors Through three studies, this research demonstrated SDT (Ryan & Deci, 2017) to be a valuable framework to provide an alternative integrative perspective on leadership behaviors (Anderson & Sun, 2017).Indeed, we provided the first demonstration, in the work context that supervisors' indifferent behaviors toward their subordinates' basic psychological needs can be modeled as a distinctive category of supervisory behaviors when tested alongside supervisors' need supportive and thwarting behaviors.In doing so, our research extends work-related SDT research by showing that supervisors' behaviors are not dichotomous or extreme in nature (i.e., adverse thwarting behaviors and beneficial supportive behaviors; e.g., Gillet, Fouquereau, Forest, et al., 2012).Rather, workers may also experience more passive and nuanced negative behaviors from their supervisors in the form of need indifferent behaviors.Moreover, we showed that the TMIB-S constitutes a valuable tool for researchers and practitioners from English-and French-speaking countries to jointly assess those three forms of supervisory behaviors (see Appendices A and B, in the supporting information file).

Multidimensionality of supervisory behaviors
Our research supported both a bifactor-ESEM (one G-factor and three S-factors) and an ESEM (three factors) representation of ratings on the TMIB-S, thus highlighting the multidimensionality of ratings on this instrument.Both these solutions are consistent with the idea that employees' ratings of their supervisors' need-related interpersonal behaviors are dominated by the type of behavior considered (supportive, thwarting, and indifferent) rather than the type of need (competence, relatedness, and autonomy).In other words, our results emphasize the importance of supervisors' interpersonal styles toward subordinates' psychological needs as a whole.This conclusion aligns well with a premise of SDT emphasizing that subordinates' psychological needs for autonomy, competence, and relatedness represent equally important psychological nutriments for psychological functioning (Ryan & Deci, 2017).Just like plants need water, sunlight and soil to grow, SDT suggests that individuals require the satisfaction of all of their three needs for autonomy, competence, and relatedness to experience healthy development, integrity, and well-being (Deci & Ryan, 2000).This equal and additive importance of the three psychological needs has been demonstrated throughout decades of SDT research (e.g., Sheldon & Filak, 2008;Vansteenkiste et al., 2020).Hence, we advocate that it is the general need supportive, thwarting, or indifferent experience that matters most, rather than how each individual need is supported, thwarted, or neglected.This conclusion has important implications for supervisors and practitioners as it encourages them to consider all three needs as equally important when interacting with subordinates.Moreover, this tripartite structure is also in line with prior research conducted in the sport area (Bhavsar et al., 2019) and is consistent with Lewin et al.'s (1939) seminal proposition regarding the tripartite nature of leadership behaviors.
When considering the bifactor-ESEM solution, our results suggest the presence of a G-factor reflecting commonalities among all leadership behaviors but also indicate that the three behavioral dimensions retain some meaningful specificity over and above this G-factor.This implies that the G-factor is not sufficient to capture all information provided by employees' TMIB-S ratings (as indicated by the poor fit of the single factor model).This global/specific distinction is purely statistical rather than perceptual: Employees are not asked to report on their general impression of their supervisor's behaviors separately from their specific impressions of their supervisors' behaviors.They are simply asked to rate a series of items reflecting various behaviors.These ratings are then used to statistically distinguish between everything that is common to all of these ratings (G-factor) and what is unique to ratings of each specific type of behavior (S-factors) once the G-factor is accounted for.More precisely, when a supervisor's behavior is witnessed by their follower, this behavior contributes to both the follower's global impression of their supervisor as a "rather good or bad" leader and to their specific perceptions of their supervisor reliance on specific types of behavior in a way that deviates from their global impression.In a traditional first-order factor model, this global perception would be ignored, and employees' perceptions of need supportive, indifferent, or thwarting behaviors would overlap substantially because they would also reflect their unmodeled global perceptions, making it hard to clearly understand the unique role played by these different behaviors.In a bifactor model, rather than directly estimating employees' perceptions of the extent to which supervisors rely on each type of behavior, we statistically separate these perceptions from their global impression.
This bifactor representation has implications for leadership research.Indeed, by providing separate estimates of shared versus unique effects, this approach overcomes prior studies' difficulty to clarify which of the less desirable forms of leadership (e.g., transactional and laissezfaire) have the most detrimental effects (Judge & Piccolo, 2004;Skogstad et al., 2007) and sheds light on the distinct effects of supervisors' need thwarting and indifferent behaviors.Practitioners could consider this global/specific distinction by acknowledging that supervisors should first focus on their overall mode of functioning, as employees' global perceptions of this mode of functioning played the strongest role in prediction.Yet, this globally positive impression does not preclude their occasional reliance on need indifferent or thwarting behaviors to result in undesirable effects and should not be overlooked.
Still, because this novel approach remains underdocumented and statistically more complex, some researchers and practitioners may prefer to resort to the more traditional ESEM approach advocated by Bhavsar et al. (2019).Our research shows that both the B-ESEM (one Gfactor and three S-factors) and the ESEM (three factors, identical to those identified by Bhavsar et al., 2019) approaches are adequate and may be used by researchers and practitioners to represent ratings on the TMIB-S.

Consequences of supervisory behaviors
The TMIB-S: A useful alternative to existing approaches and measures Our conceptual perspective is anchored in SDT (Ryan & Deci, 2017), which depicts leadership as a way to promote subordinates' self-determined motivation and well-being.In contrast, most research has so far approached leadership as a way to influence subordinates' behaviors and performance (e.g., Avolio & Bass, 1991).It is true that some other conceptualizations also approach leadership behaviors through their possible link with subordinates' well-being, but these approaches typically do so only by examining isolated negative or positive leadership behaviors (e.g., engaging leadership; Schaufeli, 2015), rather than by considering the whole range of positive and negative interpersonal behaviors available to supervisors.Our results also supported the value of the TMIB-S as a solid alternative to existing leadership measures, given that transformational leadership, LMX, passive leadership, and abusive supervision did not share any association with a wide array of well-and ill-being variables (i.e., work engagement, job satisfaction, job boredom, work-related rumination, and exhaustion), once the effects of supervisors' need supportive, indifferent, and thwarting behaviors were accounted for.
These results position the TMIB-S as a viable alternative for researchers and practitioners who wish to explore the managerial antecedents of workers' psychological health.The TMIB-S not only accounts for a significant amount of variance in a wide array of well-and ill-being indicators but also has the advantage of simultaneously measuring the positive (supportive behaviors), negative (thwarting behaviors), and passive (indifferent behaviors) forms of supervisory behaviors, based on a single theoretical framework (SDT; Ryan & Deci, 2017) and measure.In contrast, by focusing on isolated leadership behaviors, existing instruments-such as those measuring LMX, passive leadership, and abusive supervision-lose some of their empirical and theoretical comparability (Avolio et al., 1999;Graen & Uhl-Bien, 1995;Tepper, 2000).As such, researchers and practitioners seeking to simultaneously capture positive, negative, and passive supervisory behaviors, when picking scales stemming from distinct conceptual frameworks, risk measuring overlapping or hardly comparable realities.The TMIB-S provides a solid alternative to this suboptimal situation by offering a valid integrative measure, based on a single, wellestablished, theoretical framework (Ryan & Deci, 2017).

Differentiated consequences of supervisory behaviors measured by the TMIB-S
By jointly considering the shared and specific effects of need supportive, thwarting, and indifferent behaviors, results from Studies 2 and 3 highlighted the well-differentiated consequences of these three interpersonal styles for subordinates.Importantly, we showed that actively negative and passive neglectful forms of leadership had clearly distinct consequences for employees' functioning, which remained unclear in the leadership literature (Judge & Piccolo, 2004).Results even showed that specific need indifferent behaviors may be more problematic than specific need thwarting behaviors.Indeed, specific need indifferent behaviors were directly and positively associated with the most problematic indicators of ill-being (i.e., emotional exhaustion and work-related rumination), whereas specific need thwarting behaviors had no direct associations with these outcomes.Specific need thwarting behaviors were even associated with some positive consequences (higher specific levels of relatedness satisfaction and of work engagement).These results reinforce the distinct implications of exposure to need indifferent versus thwarting behaviors from supervisors: Whereas the former results in the most detrimental consequences, the latter results in more varied psychological experiences.
Indeed, when facing a supervisor who threatens their psychological needs, subordinates could engage in need crafting (de Bloom et al., 2020) by seeking other sources of need support (e.g., colleagues; Moreau & Mageau, 2012) or could rely on other effective coping strategies (e.g., ingratiation; Harvey et al., 2007).This interpretation is consistent with the idea that individuals do not passively react to threatening social environments but can also proactively change the characteristics of their environment (Bakker & Demerouti, 2017;Lazarus & Folkman, 1984) in order to make their experiences more aligned with their needs and preferences.Research stemming from various theoretical frameworks has consensually indicated that such coping strategies can take the form of seeking social support from various sources (Bakker & Demerouti, 2017;de Bloom et al., 2020;Lazarus & Folkman, 1984).In sum, we suggest that the consequences of need thwarting behaviors may depend on whether subordinates yield to these actively adverse behaviors or decide to proactively attempt to overcome them.It should still be noted that such coping strategies may not be indefinitely protective as they require the investment of resources and may thus progressively take a toll on employees.More research into the temporal dynamics of employees' need-related experiences, need states, and need crafting behaviors is clearly needed to better understand these issues.
In contrast, need indifferent behaviors seemed to be more consistently harmful than need thwarting behaviors, which adds to the proposal that passive and neglectful forms of leadership may be more destructive than actively adverse ones (Skogstad et al., 2007).An explanation for this may be that by conveying uncertainty, chaos, and alienation, need indifferent supervisors create a climate of ambiguity, which is known to be particularly problematic for individuals' psychological functioning (see Skogstad et al., 2014).As a result of this general ambiguity and lack of clarity provided by their supervisor, subordinates may experience vulnerability (Lapidot et al., 2007), thus resulting in ill-being (Chênevert et al., 2013;Skogstad et al., 2014).Moreover, need indifferent behaviors may leave subordinates unable to categorize their supervisor's behaviors and create unpredictability in their social interactions.Indeed, the social identity theory of leadership suggests that leadership and its beneficial consequences derive from social identitybased perceptions of the leader as a group member (Steffens et al., 2021).Yet, need indifferent behaviors make it impossible for subordinates to categorize their leader either as an in-or outgroup member, which could explain their adverse consequences.
Indeed, uncertainty can create anxiety for some individuals who will then become motivated to attempt to reduce this psychological discomfort in order to return to a state of homeostasis (Hirsh et al., 2012).As such, when finding themselves in a state of uncertainty (lacking direct and explicit information) about their supervisor, subordinates will be motivated to reduce this experience to a manageable level and will thus activate a search for information to regain a sense of predictability in their relationship with their supervisor (Strom et al., 2014).Yet, these uncertainty reduction efforts are demanding in terms of resources and not always successful.As a result, they can place significant strain on employees, resulting in detrimental consequences.More generally, the adverse consequences of indifferent behaviors stress how important it is to take this neglectful form of supervisory behaviors into consideration (Chénard-Poirier et al., 2022).Although these neglectful behaviors may be trivialized in organizations, where they are often seen as less problematic than need thwarting behaviors, our results suggest that it might be critical for organizations to prevent them, as they have the power to produce even more detrimental consequences than need thwarting behaviors.
Unsurprisingly, our results showed that supervisors' supportive behaviors promote the most positive consequences for subordinates (e.g., higher need fulfillment and work engagement) and protect from the most detrimental ones (e.g., lower work-related rumination).These results are in line with prior research (e.g., Gillet, Fouquereau, Forest, et al., 2012) and more generally with SDT (Ryan & Deci, 2017) and encourage organizations to search for, nurture, and promote such supportive managerial behaviors.
More generally, the strong direct associations found between specific need indifferent behaviors and job boredom in our research suggest the existence of a passively deleterious pathway underpinning the effects of need indifferent behaviors.This health eroding process might complement the two processes established by the JD-R (job demands-resources; Bakker & Demerouti, 2017) model, with the health impairment process underpinning the effects of need thwarting behaviors and the motivational process underpinning those of the need supporting behaviors.

Limitations and research perspectives
This research presents limitations worth considering in the interpretation of results.First, because our research included three samples of Western employees, it is unknown whether our results would be replicated in additional linguistic and cultural groups.This question is of particular relevance, for need supports' universality is one of SDT's fundamental tenets (Vansteenkiste et al., 2020).Second, because we used self-reported measures, social desirability and self-report biases could have influenced participants' responses.The use of more objective physiological indices of individual functioning (e.g., Bartholomew, Ntoumanis, Ryan, Bosch, & Thøgersen-Ntoumani, 2011) could allow to address this limitation.Third, we only considered the role of supervisors' need-related behaviors in relation to psychological needs and psychological health indicators.It would be interesting for future research to study the motivational (i.e., work motivation, Gagné et al., 2015) and behavioral (e.g., counterproductive behaviors; Detert et al., 2007) consequences of these behaviors, as well as their impact on the work-family interface (e.g., work-family conflict; Huyghebaert et al., 2018).This would allow for a broader understanding of the role played by these behaviors in relation to a wider array of consequences.Future research could also investigate how supervisors' need indifferent behaviors relate to subordinates' need unfulfillment, alongside their need frustration and satisfaction (Huyghebaert-Zouaghi et al., 2021).
Fourth, we did not examine how individual orientations could moderate the relations between supervisors' interpersonal behaviors and consequences for subordinates.For instance, individuals who are characterized by a very strong autonomy orientation (i.e., seeking self-initiation and activities that are interesting and challenging) may be less vulnerable to the adverse effects of need indifferent behaviors from their supervisor than workers with a more controlled orientation (i.e., individuals who prefer being controlled by rewards, deadlines, structures, and the directives of others; Ryan & Deci, 2017).Future research could further explore the fit between individual preferences and supervisors' interpersonal behaviors.Fifth, we solely considered outcomes of supervisors' need-related behaviors.It would be interesting to also consider their organizational antecedents: Scholars could examine how organizations create the conditions to develop optimal supervisory behaviors (need supportive) and prevent undesirable ones (need thwarting and indifferent).Based on the trickle-down effect (Eisenberger & Stinglhamber, 2011), perceived organizational support may convey a norm according to which every employee is expected to show support to others and spread in the form of need supportive behaviors.Conversely, organizational dehumanization (Lagios et al., 2021) may convey a norm of mistreatment tolerance and allow for need thwarting and indifferent behaviors to occur.
Finally, we relied solely on variable-centered analyses, which ignore the possible existence of qualitatively distinct subpopulations of employees exposed to specific configurations of supervisory behaviors.Future research would benefit from a person-centered approach (Morin et al., 2018) to better examine the combined effects of supervisors' need-related interpersonal behaviors on employees' functioning by simultaneously considering all types of supervisory behaviors and their possible co-existence (e.g., Leo et al., 2022).Pursuing this avenue would more generally address the relative dearth of person-centered studies in the leadership literature (e.g., Chénard Poirier et al., 2017).Moreover, longitudinal designs could be used to allow for a more thorough examination of how different combinations of supervisory behaviors evolve and relate to workers' functioning over time.Indeed, longitudinal designs conducted over shorter (e.g., daily; Breevaart et al., 2016) or longer (e.g., 4 months; Detert et al., 2007) time spans could allow for a better understanding of the temporal and dynamic nature of subordinates' perceptions of their supervisors' behaviors while making it possible to observe how distinct profiles of supervisors' need-related behaviors evolve and relate to subordinates' functioning over time.Indeed, over time, subordinates could see their supervisor as being supportive at times and thwarting or indifferent at other times, which could produce transitions in profile membership and result in health-related or motivational changes for subordinates.

CONCLUSION
This research adds to leadership research by supporting the validity, in the work context, of an integrative theoretical and operational framework to guide the understanding and measurement of supervisors' need-related interpersonal behaviors that may prove to be particularly useful for researchers interested in examining the psychological and motivational consequences of leadership.Indeed, rather than adding irrelevant details or an additional microscopic focus to an already overly saturated leadership field, we took a step back, and suggested a way to start anew, using a novel integrative perspective considering how leaders can act in relation to their subordinates' psychological needs.This perspective, anchored in the currently dominant theoretical framework on employee motivation and well-being (Ryan & Deci, 2017), can bring something new.Indeed, while research based on mainstream leadership theories has failed to document which of the less desirable forms of leadership are most deleterious for subordinates (Judge & Piccolo, 2004;Skogstad et al., 2007), our SDTbased approach provides strong evidence that need indifferent behaviors clearly have more detrimental effects on employees' psychological needs and well-being than need thwarting behaviors.In times that some refer to as a leadership talent crisis (Harter & Adkins, 2015), this perspective may help organizations to identify behaviors to eradicate in order to protect employee well-being.
items and one without such a global factor.Although both models resulted in an equivalent level of model fit and although their global factor was well-defined, they decided to reject the model including the global factor based on the observation that, in this model, the specific need thwarting factor did not retain any specificity of its own.However, this observation simply indicates that these items mainly serve to define the global factor and that once this global variance is considered, they do not retain any residual specificity (suggesting that ratings of need thwarting behaviors seldom deviate from global perceptions of interpersonal behaviors in the sample under investigation).Thus, observing a weak specific factor is not a valid reason to reject a model including a global factor, as long as some of the other specific factors are found to retain specificity (Morin et al., 2020). 2 It would be erroneous to see the specific (S-) factors as being in some small way unique factors sharing most of the shared variance among the indicators forming them.Rather, the S-factors are wholly made of what is uniquely left in these specific behaviors, beyond the shared variance captured by the global (G-) factor. 3Upon request from a reviewer, we conducted additional analyses where (1) all alternative measures (passive leadership, LMX, transformational leadership, and abusive supervision) were added to analyses already including the TMIB-S and (2) the TMIB-S was added to analyses already including all other measures.We contrasted the same four alternative models (i.e., null, partial mediation, total mediation, and direct).Results from these analyses are described in more details in the supporting information file and showed that adding the other leadership scales did not add to prediction once the TMIB-S was considered, whereas the TMIB-S explained well-and ill-being outcomes beyond these existing measures. 4Our final model including only the TMIB-S and the alternative models (also including the other measures) resulted in comparable R 2 (Table S16 of the supporting information file), showing that the other measures did not add to the prediction beyond what was explained by the TMIB-S.Similar results were found for Study 3. 5 We conducted additional analyses to test whether the TMIB-S factors contributed to prediction beyond the role played by the other leadership measures.Results from these additional analyses are reported in Table S15 and S17 of the supporting information file.These results showed that adding the TMIB-S resulted in a noteworthy increase in model fit, as well as in improvement in R 2 relative to a null model (in which the effects of the TMIB-S factors were set to be exactly zero while those of the other measures were freely estimated).These results indicate that the TMIB-S predicted well-and ill-being over existing measures.Similar results were found for Study 3.
tripartite conceptualization of supervisors' need supportive, thwarting, and indifferent behaviors to the work context.The present research was designed to directly address this limitation by proposing an adaptation of the Tripartite Measure of Interpersonal Behaviors-Coach (TMIB-C, Bhavsar et al., 2019) to supervisors' behaviors (TMIB-S) and to test the construct validity of this measure in two languages (English and French).

14640597, 0 ,
Downloaded from https://iaap-journals.onlinelibrary.wiley.com/doi/10.1111/apps.12423,Wiley Online Library on [28/01/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License F I G U R E 1 Predictive models tested in Study 2 and in Study 3

T
A B L E 2 Test of predictive validity: Results from the partial mediation model (Study 2)

T
A B L E 3 Test of predictive validity: Results from the partial mediation model (Study 3) .05), whereas those associated with a different *p < .05. **p < .01.