Helping Trajectories During Role Transitions: How They Vary and Why It Matters

Sensemaking and the Baseline Curve of Helping During Role Transition

The TTOCB proposes sensemaking and identity-crafting as key frameworks for understanding how the helping behavior of those transitioning shifts over the course of their onboarding into a new role. Role transitions, recognized as significant events triggering sensemaking and identity work (Ibarra & Barbulescu, 2010), follow three stages as individuals evolve from “outsiders” to “insiders.” Methot et al. (2017) suggest that as they advance through these stages, individuals exhibit a curvilinear pattern of helping. Upon role entry, employees are often highly motivated to help others as a means of asserting benevolent values and identity, and managing others’ impressions of them as a good citizen (Bolino, 1999). However, in this initial transition stage, aside from prioritizing their own mastery of task-related skills and knowledge, they often lack the requisite skills and knowledge to effectively assist others, and may not even grasp how to discern the types of assistance needed (Methot et al., 2017). Additionally, given nascent social networks (Park & Nawakitphaitoon, 2018), opportunities to engage in helping may be limited, thus constraining their engagement in helping at entry. However, as those in transition familiarize themselves with the rules, norms, and networks of their workplace, helping rises. Motivated to project their best selves, as they master job-related tasks, integrate into their roles, and interact with a broader range of colleagues, transitioning employees gradually intensify the level of assistance provided to others (Methot et al., 2017). However, as they become more acquainted with their new role, they also become more attuned to the potential unintended, negative consequences of their actions and more cognizant of cues and signals deviating from their initial expectations (Boswell, Boudreau, & Tichy, 2005). Additionally, they may sense that their assistance is being taken for granted, resulting in an expansion of the job itself and in-role demands (Bolino & Grant, 2016). Hence, they may avoid any further intensification of their helping, resulting in its stabilization. Moreover, Methot et al. (2017: 15) suggest that in response to certain cues (e.g., the intensification of task responsibilities accompanying the end of onboarding), one option may even be to “reevaluate” or even “discontinue” helping. Accordingly, we propose:

How Trajectories May Differ for Newcomers vs. Transitioning Incumbents

Just as the baseline curvilinear helping trajectory should not be taken to mean that those transitioning necessarily return to the level of helping with which they entered, the TTOCB also suggests that not all those transitioning into a new role will exhibit the same trajectory. Rather it suggests that the degree of identity discrepancy (how discrepant one’s experience is with one’s identity) and the significance of discrete and continuous experiences may predict varying helping trajectories over time. Given likely differences in identity discrepancies and the cues experienced by those transitioning from within the same organization (i.e., transitioning incumbents) versus those entering the organization for the first time (i.e., newcomers), we propose that the former manifest a trajectory markedly different than that of the latter, namely one that is higher but flatter.

In contrast to newcomers, transitioning incumbents tend to have some understanding of the cultural norms surrounding how help is offered and provided within the organization (Feldman & Brett, 1983). Moreover, because they are often recruited into their new role precisely because of their firm-specific human and/or social capital (Zhu, Wanberg, Harrison, & Diehn, 2016), relative to newcomers, they are likely to be afforded a substantially greater number of helping opportunities. That is, with a deeper understanding of the organization’s relational norms, greater experience in applying their human capital to firm-specific problems, and familiarity with securing firm-specific resources to solve such problems, transitioning incumbents are likely to be better positioned at entry than newcomers to identify how their competencies can assist others, match these competencies to others’ needs, and frame (and time) their offers of assistance to maximize impact (Ashforth, 2001). Additionally, given their familiarity with the organization and higher levels of firm-specific human and social capital, relative to newcomers, allocating time and effort to helping others is likely to pose less risk to their mastering of new tasks (Ashforth, 2001). To the extent that the perceived costs of helping are lower when entering their new role, relative to newcomers, transitioning incumbents likely experience a smaller discrepancy in their helping identity and are thus likely to engage in a higher level of helping upon entering their new role. Accordingly, we posit:

Beyond this difference in starting levels of helping upon entry, for three reasons the overall helping trajectories of these two groups are likely to differ significantly, with the trajectory of transitioning incumbents tending to be flatter than that of newcomers over the course of their first year on the job. First, although both groups of transitioning employees are likely to be motivated to close any helping-based identity discrepancy before the end of their period of intensive on-boarding (in most organizations, between 3 to 6 months following job entry; Ellis, Bauer, & Erdogan, 2014), because their initial level of helping is lower than that of transitioning incumbents, newcomers are likely to have to engage in a steeper upward slope of helping to approach what they perceive to be normative or expected levels of helping in a timely manner. Second, given their superior understanding of helping related needs, norms, and expectations in the organization, transitioning incumbents are likely to be better able to calibrate their helping initiatives to those opportunities allowing for optimal impact (Feldman & Brett, 1983). Accordingly, while all those transitioning attempt to adjust their helping to meet concurrent role expectations and thus reduce helping-based identity discrepancies, relative to newcomers transitioning incumbents are likely to do so in a more strategic, regulated manner, manifesting in a relatively slower increase in helping over the initial transition period. In contrast, newcomers are likely to be less adept at calibrating and targeting their helping (Methot et al., 2017), leading them to “fire in all directions,” seeking out and exploiting as many opportunities as possible to assist others so as to more rapidly meet normative helping expectations and align harmoniously with their identity (Martin, 2016; Rubenstein, 2014). Hence, while Methot et al. (2017) propose that this initial upward slope in helping among new employees peaks and stabilizes at the point at which environmental signals suggest they have gone too far in addressing identity discrepancies, and/or when limited resources preclude any further increase in helping, this inflection point may occur at a higher level for newcomers due to their disadvantage in reading environmental signals and accurately calibrating their helping to environmental needs.

Finally, Methot et al. (2017) propose that helping trajectories are influenced by the kinds of cues individuals experience along the way. The TTOCB differentiates between event-based cues (disjointed experiences, serving as markers that “have the potential to radically affect the trajectory of one’s development and propensity to identify with a role” [Ashforth, 2001: 174]) and process cues, which they define (p. 18) as “arcs of continuous experience.” It also (p. 20) distinguishes between cues by level of significance, with more major cues being those “of critical importance” and having the potential to “elicit intense reactions.”

After engaging in increasingly high levels of helping during the initial on-boarding period, both newcomers and transitioning incumbents may experience a process cue in the form of “citizenship fatigue,” which can be defined as a state in which employees experience exhaustion, weariness, or heightened tension as a result of engaging in OCB (Bolino, Hsiung, Harvey, & LePine, 2015), resulting in a gradual diminishing of engagement in helping to conserve resources and/or facilitate the allocation of these resources to other work or non-work priorities (Methot et al., 2017). However, beyond any such process cue, newcomers may also experience a highly significant event cue at the point in the on-boarding process in which the intensity of socialization declines (i.e., at 6 months according to Ellis et al., 2014), and/or when they are no longer formally or informally considered “on probation,” and are recognized as “regular” employees. To the extent that this event cue often serves as an indicator of acceptance, a marker of identity fulfillment, and the removal of a significant source of uncertainty for those transitioning into a new role (i.e., precariousness of employment), it likely diminishes some of the pressure on newcomers to try to accede to every help request received, thus potentially resulting in a sharper inflection point, characterized by a more immediate and sharper stabilization in helping relative to that which would be expected in the case of a process cue such as citizenship fatigue. Moreover, to the extent that probationary terms are less severe for transitioning incumbents than for newcomers (i.e., they are more likely to be reassigned rather than fired if the probationary terms are not met), completing one’s probation is likely to be a more salient milestone or “major” event cue for newcomers, with the potential to elicit behavioral shifts that are sharper and of greater magnitude.

Taken in combination, these factors suggest that although both newcomers and transitioning incumbents manifest a curvilinear helping trajectory, this trajectory will be flatter for transitioning incumbents than newcomers. Hence:

Supportive Leadership and Peer Descriptive Helping Norms

As noted, a foundational proposition of the TTOCB is that over time OCB varies as a function of sensemaking cues. Extending this idea, we propose that key transition stakeholders, namely the leaders and peers that individuals encounter when moving into a new work role, serve as key cue senders, guiding the sensemaking of those in transition by engaging in sense-giving (Maitlis & Lawrence, 2007: 57). Consistent with Attraction-Selection-Attrition (ASA) theory (Schneider, Smith, & Goldstein, 2000), leader and peer sense-giving during the hiring process can influence candidate self-selection. For example, for transitioning incumbents and newcomers self-identifying as more helping-oriented, opportunities to join work units led by supportive leaders and/or characterized by higher peer helping norms may be deemed preferential. To the extent that this occurs, it would be expected to reduce the magnitude of identity discrepancy experienced by those transitioning into a new role and, as a result, manifest in a higher initial level of helping manifested upon role entry. Furthermore, as leaders and peers engage in helping-related sense-giving during the transition, the signals they transmit may serve as process cues, altering the overall arc of the helping trajectory.

The role of supportive leadership: The role of leaders as sense-givers begins already during the selection process as, consistent with ASA theory, candidates are drawn to workplaces whose leaders offer cues that their helping-related values and expectancies align with those of the candidate (Cristofaro, 2022). Moreover, the signaling leaders offer regarding the expected utility of helping (Saks & Ashforth, 1997) likely serve as process cues, shifting the general arc of the curvilinear helping trajectory.

Supportive leadership is defined as “behavior directed toward the satisfaction of subordinates’ needs and preferences, such as displaying concern for subordinates’ welfare and creating a friendly and psychologically supportive work environment” (House, 1996: 327). It places a premium on encouraging “other-oriented help” (Rubenstein, Kammeyer-Mueller, & Thundiyil, 2020: 1480), taking account of followers’ individual needs and preferences when making decisions (Rafferty & Griffin, 2006), the modeling of respect and concern for both individual well-being and development, and the development of supportive relationships among unit members as a basis for enhancing the collective good (Kim, Atwater, Jolly, Ugwuanyi, Baik, & Yu, 2021). A key relational element of various leadership theories (e.g., the individualized consideration dimension of Transformational Leadership; Rafferty & Griffin, 2006), supportive leaders explicitly highlight the value of helping as aligned with organizational goals and implicitly reinforce this through their own prosocial modeling (Grant, 2012; Kim et al., 2021). Research indicates that by modeling such behavior themselves, supportive leaders create a supportive climate among those they supervise, manifesting in terms of higher levels of team collaboration and a collective mindset oriented towards assisting one another as a basis for collective progress and long-term success (Kim et al., 2021).

From an ASA perspective, those self-selecting to work under more supportive leaders are likely to self-identify with the leader’s prosocial expectations and consequently exhibit a higher level of helping already upon entering the new role than those opting to serve under less supportive leaders (Lorinkova & Perry, 2019). Moreover, supportive leadership is, over time, likely to stabilize this higher level of helping, resulting in a “flatter” or more linear trajectory, for two reasons. First, such leaders are likely to continuously signal—in part, through their own modeling of such behavior—the value they place on prosocial behavior, thus reducing the concerns of those in transition that there may be a personal penalty to pay by allocated time and effort to assisting others rather than focusing on their own assigned tasks (Effelsberg, Solga, & Gurt, 2014). Second, given their focus on supporting their subordinates, such leaders are likely to facilitate helping by providing new entrants with greater helping opportunities (Zhu & Akhtar, 2014). With those transitioning into units with more supportive leaders tending to manifest higher levels of helping already upon entry, and experiencing leadership consistently signaling the value of helping, and providing the kind of individualized support allowing them to maintain steady engagement in helping even in the face of specific challenges, we propose:

The role of peer descriptive helping norms: Because leaders’ signaling can often be broad and generic, those in transition may face considerable uncertainty when attempting to translate and apply these signals to their specific situation (Ashforth, Sluss, & Harrison, 2007; Wang, Kammeyer-Mueller, Liu, & Li, 2015). Hence, as noted by Liu, Wang, Bamberger, Shi, and Bacharach (2015: 335), those transitioning into a new role, “often have little choice but to glean insights into how to achieve these imperatives and objectives from informants around them,” especially veteran peers. It is through social learning (Bandura, 1977)—by observing the behavior of veteran peers—that those in transition glean an understanding of others’ expectations and insight into the boundaries of normative (i.e., socially acceptable) behaviors and responses (Ashford & Black, 1996). Consequently, in situations where the most suitable course of action is unclear, descriptive norms as an indicator of collective action (Cialdini & Trost, 1998) reduce uncertainty (Schwandt, 2005), and provide a means through which those in transition glean insight into the most effective and justifiable responses to particular situations (Bamberger & Biron, 2007).

Like leaders, veteran peers may, through their helping norms, influence the initial level of helping that those in transition engage in upon entering their new position (Hurst et al., 2012). First, veteran peer descriptive helping norms (herein referred to as helping norms) may influence the self-selection of candidates for an open position by signaling normative helping expectations and giving an impression of the unit “helping culture” (Amabile, Fisher, & Pillemer, 2014). Candidates inferring that their helping-related values are incongruent with those of the veterans with whom they would be working may thus self-select out of consideration (Schneider, 1987). Second, upon entering a new position, consistent with sensemaking theory, those in transition are likely to glean understandings and draw meaning based on the easily observable helping behaviors of their veteran peers (i.e., helping norms; Liu et al., 2015; Louis, 1980). In units in which helping norms are higher, those transitioning are likely to have to aim higher to reduce any identity discrepancy than they would were they to enter a unit with lower helping norms. Similarly, in such units, veteran helping norms are likely to serve as a continuing flow of process cues that, over time, stabilize these higher levels of helping among those transitioning into new work roles. For instance, helping norms may cue the stabilization of higher levels of helping by signaling that even when facing increased task demands, there are benefits to retaining high levels of helping, as well as costs to reducing it (Gonzalez-Mulé, DeGeest, McCormick, Seong, & Brown, 2014). Based on this logic, we posit:

The Impact of Helping Trajectories on Transition Outcomes

Research on transition-related adjustment has emphasized three outcomes, namely task performance, social integration, and retention (Bauer & Erdogan, 2012). Given that socialization is a process involving learning and assimilation, task performance and social integration emerge as pivotal outcomes for those transitioning (Bauer & Erdogan, 2012). Furthermore, with up to a third of employees quitting their job within six months (Maurer, 2015), and that failure in building relationships with others accounts for such early turnover (Zhou, Park, Kammeyer-Mueller, Shah, & Campbell, 2022), the turnover intentions of those transitioning serve as an additional criterion for transition adjustment. Thus, we next theorize how variations in helping trajectories are connected to these outcomes.

Level of Helping Trajectory and Socialization Outcomes: Individuals displaying more elevated levels of OCB and helping tend to receive higher ratings of task performance from their leaders (e.g., Podsakoff, MacKenzie, Paine, & Bachrach, 2000). One explanation for this may be that by more immediately providing higher levels of helping, those in transition stand out as reliable and trustworthy partners, thus motivating others to reciprocate with greater and higher quality task-related feedback, and an enhanced level of sharing of information and other task-related resources, enabling those in transition to master their new role and display heightened levels of task performance (Jia et al., 2021; Zhou et al., 2022) more quickly. This suggests that a high level of helping is positively associated with task performance.

Moreover, to the extent that a heightened level of helping may endow those in a new role with greater social capital resources (facilitating access to information and support [Ashford & Black, 1996]), such individuals may be better positioned to jump-start the kind of reciprocity-based social exchange central to establishing and developing interpersonal relationships (Grant, 2013; Jia et al., 2021). Studies indicate that cycles of reciprocated helping not only elicit and strengthen instrumental relationships (Bolino et al., 2002), but that over time these instrumental relationships morph into more intimate relationships (Colbert, Bono, & Purvanova, 2016), thus cementing social integration—the degree to which individuals experience a sense of social acceptance and belonging. By providing higher levels of assistance to others, transitioners may quicken the pace of trust- and relationship-building with insiders (De Jong, Van der Vegt, & Molleman, 2007), thus enhancing their social integration at work.

With the tendency of others to reciprocate prosocial acts by those transitioning, research suggests that helping by those transitioning into a new role is instrumental in the transitioners’ ability to develop strong and positive interpersonal relationships in the new work unit (Grant, 2013; Jia et al., 2021). Studies indicate that the presence of such relationships are among the key determinants of employee retention (Dutton & Ragins, 2017). Indeed, embeddedness theory (Mitchell, Holtom, Lee, Sablynski, & Erez, 2001) suggests that intimate relationships with other employees (i.e., “links”) combine with other attachment factors to elicit a sense among employees that they are “stuck” in a web of attachments making it difficult to imagine leaving (Mitchell et al., 2001: 1104). Accordingly, a heightened level of helping among new entrants is also likely to be inversely associated with their intention to quit.

Taken together, the logic above suggests that:

Curvature of Helping Trajectory and Socialization Outcomes: In contrast to the beneficial consequences associated with a heightened level of helping noted above, helping trajectories characterized by a heightened degree of curvature are likely to indicate instability and inconsistency in helping. Moreover, from a prospect theory perspective (Kahneman & Tversky, 1979), to the degree that the high point of the trajectory serves as reference point for any subsequent assessment of the change in new entrant’s helping, the downward trend is likely to have a robust influence on any attributions drawn (Alessandri, Cortina, Sheng, & Borgogni, 2021; Chen, Ployhart, Thomas, Anderson, & Bliese, 2011), thus potentially eliciting more adverse outcomes.

Studies indicate that leaders’ attributions of subordinates helping explain a substantial portion of the variance in leader-assessed performance (Podsakoff et al., 2000). However, in the face of subordinate behavioral inconsistency, leaders may be less able to generate such attributions (Crittenden, 1983), assuming any high level to be temporary and/or basing their impression more on the downward trend than on the level of helping at any particular point in time (particularly if performance is assessed after one year on the job, with new entrant helping peaking at six months) (Alessandri et al., 2021; Chen et al., 2011). In contrast, as a sustained level of helping by an individual transitioning into a new role is rather distinctive behavior, it is likely to be more easily recognized and recalled by leaders, on the basis of attribute substitution, generalizing from it to their overall ratings of subordinate task performance (Mackenzie, Podsakoff, & Fetter, 1991).

Furthermore while, as noted above, higher levels of helping may grease the wheels of reciprocation and facilitate the development of more intimate interpersonal relationships at work and hence the social integration of the newcomer, research suggests that such relationships may suffer if insiders deem those in transition to be inconsistent or unreliable sources of assistance (sometimes delivering, sometimes not) (Halbesleben & Wheeler, 2015). Moreover, to the extent that peers may over-weigh the downward slope in helping relative to the actual level of helping at any point in time (Alessandri et al., 2021; Chen et al., 2011), their relationship with the new entrant is likely to suffer, thus putting the new entrant’s social integration at risk.

In the same way, a heightened degree of curvature in the helping trajectory may pose a risk to the new entrant’s job embeddedness and thus heighten their intention to quit. While “links” are logically likely to strengthen as insiders become reliant on the assistance transitioners provide, to the extent that others question the reliability of a new entrant to provide assistance when needed, social links between new entrants and their peers are likely to erode over time. That is, the strength of these relationships may decline over time if those in transition fail to deliver or maintain the level of assistance others have learned to expect, or are otherwise deemed to be unreliable sources of assistance. Although Mitchell et al. (2001) theorize that such a decline in embeddedness likely only serves to increase the risks that some other event may drive turnover, Jiang, Liu, McKay, Lee, and Mitchell (2012) conclude from their meta-analytic results that lower job-based embeddedness is directly associated with increased turnover intentions.

Taken together, the logic above suggests that:

Participants and Procedures

Data were collected from new and veteran employees in a large Chinese IT company (total workforce of approximately 20,000). Those employed in this company (largely engineers) worked interdependently in the context of project-focused teams. Initial data collection occurred as newcomers entered their new roles in the fall of 2017 in the context of the company’s annual recruitment program. All internal transitions occurred at the end of the fiscal year (i.e., December), allowing us to similarly sample transitioning incumbents at the end of the fiscal year in 2017 and starting in their new roles by February 2018. As a reference group, we also randomly selected organizational veterans (having at least one year of job tenure as of February 2018) from the same team as those to which the newcomers and/or transitioning incumbents entered.

Data were collected in the same manner across all three groups. At T0 (newcomers or transitioning incumbent’s started working within a week), we sent invitations and an initial survey focusing on demographics. New entrants and veterans also reported on their unit leaders’ supportive leadership at T0. Eight hundred and forty three invitations were sent to newcomers, with 685 responding. Three hundred transitioning incumbents were randomly invited, with 253 responding. Of the 840 randomly sampled veterans invited to participate, 716 responded to the initial survey. Two hundred and twenty one invitations were also sent to team leaders at T0, with 166 responding. Within a week of completing the T0 instrument, participants across the three groups received a second T1 survey containing items regarding their helping. Subsequent questionnaires containing these same helping items were sent to participants every three months over the next year (i.e., T2-T5) with links to each quarterly survey sent to participants through their personal email. At T5, participants were requested to assess their social integration and express their intention to quit. Simultaneously, team leaders were asked to evaluate their subordinates’ task performance. Participants received a small gift each time they completed a survey. Wave-specific response rates ranged from 61% to 85% depending upon the participant group (details available from the authors). Per convention for the assessment of time-based trajectories (Bollen & Curran, 2006), we retained in our final sample all individuals with at least four data points. Accordingly, the final sample included 356 newcomers, 135 transitioning incumbents, 407 veterans, and 102 leaders drawn from 102 teams, with the final response rates for newcomers, transitioning incumbents, and veterans being 52.0%, 53.4%, and 56.8%, respectively. Using t-tests to compare participants completing at least four surveys with those who missed two or more follow-up measurements subsequent to T0 data, attrition analyses indicated no significant differences in participants’ demographic characteristics. In the final analyzed sample, 58.7%, 60.7%, 58.0%, and 84.3% of newcomer, transitioning incumbents, veterans, and team leaders (respectively) were men, with an average age of 23.9 (SD = 2.21), 34.1 (SD = 6.47), 36.6 (SD = 4.17) and 38.2 (SD = 5.78) years, respectively, and with 64.6%, 43.0%, 60.7% and 43.1% of them having a college degree.

Measures

Responses to all instruments were assessed on the basis of a 7-point Likert-like scale with anchors ranging from 1 = strongly disagree to 7 = strongly agree. Given that our study was conducted in China, we followed Brislin’s (1970) recommendations by employing translation and back-translation procedures from English to Chinese to ensure the accuracy and cultural appropriateness of all translated measures ¹

Analytical Strategy

Given the multilevel and longitudinal structure of our data—with repeated measures nested within individuals who are nested within teams—we employed Bayesian random coefficient modeling in Mplus 8.4 to test our hypotheses. To address the curvilinear pattern of helping behavior during role transitions (H1) and differences between transitioning incumbents and newcomers (H2a-b), we specified a three-level Bayesian growth model (Model 1). At the within-person level (Level 1), helping behavior was modeled as a quadratic function of time, with random linear (β_1ij) and quadratic (β_2ij) slopes capturing individual-specific trajectories. At the between-person level (Level 2), we incorporated control variables (gender, age, education) to predict the helping intercept and random slopes. Here, transition status variables (dummies for newcomers and transitioning incumbents) were key predictors, allowing us to quantify baseline differences in entry helping levels (H2a) and trajectory shapes (H2b). Finally, at the between-team level (Level 3), random intercepts for the helping trajectory parameters and transition effects were estimated, providing a team-level lens on helping behavior. To facilitate interpretation, time was centered at the study midpoint (time at T3 = 0), while age and education were group-mean centered within teams.

Building on this foundation, we tested Hypotheses 3a-b (supportive leadership as a moderator) and 4a-b (peer helping norms as a moderator) in two distinct three-level models. For Hypothesis 3, we extended Model 1 by adding team-level supportive leadership (grand-mean centered) at Level 3 (Model 2a), specifying paths where supportive leadership directly influenced the helping intercept and moderated the linear (β_1ij) and quadratic (β_2ij) slopes. Similarly, for Hypothesis 4, we created a parallel model (Model 2b) with peer helping norms (grand-mean centered) at Level 3, examining its effects on the intercept and slope parameters. Importantly, in both models, we incorporated the team-level moderators to alter the magnitude of transition status effects on the helping intercept and slopes. This allowed us to assess the degree to which supportive leadership or peer norms influence newcomers’ and transitioning incumbents’ initial helping levels and their trajectory dynamics. Conditional estimates at ±1 SD of moderators were derived for hypothesis testing (e.g., helping level differences at entry under high/low moderators), with significance determined by 95% credible intervals. All other specifications—including control variables, centering approaches, and Bayesian estimation procedures—remained consistent with Model 1.

To examine the consequences of helping trajectories (H5a-b), we fitted a separate, two-level Bayesian model (Model 3). At Level 1 (within-person), helping behavior was again modeled as a quadratic trajectory with random linear and quadratic slopes. At Level 2 (between-person), the random intercept (β_0ij), linear slope (β_1ij), and quadratic slope (β_2ij) from the helping trajectory served as predictors of three distal outcomes: task performance, social integration, and turnover intentions. Control variables (gender, age, education, transition status) were included as predictors of both the helping parameters and the distal outcomes. For consistency, age and education were grand-mean centered.

Across all models, we employed Bayesian estimation with non-informative priors, enhancing stability in the context of complex variance structures. We verified convergence rigorously using the potential scale reduction (PSR) statistic (<1.01), autocorrelation plots and trace plots. Missing data were addressed through Bayesian full-information estimation under missing-at-random assumptions. Since Bayesian estimation in Mplus naturally accommodates outliers via non-normal, heavy-tailed distributions (Kruschke, 2013, 2015; Muthén & Muthén, 1998-2017), we did not screen for or remove outliers to avoid selective bias and artificially reduced variance in the data from alteration. Finally, effect estimates are reported as posterior medians with 95% credible intervals (CIs), with significance inferred when CIs excluded zero. For hypotheses 3-5, we estimated a pseudo R² based on the proportion reduction in residual variance (Singer, 1998). ²

Descriptive Statistics and Preliminary Tests

Means, standard deviations, and correlations among the study variables are presented in Table 1 (with correlation tables for each role transition group reported in Supplemental Appendix 1). As shown, positive correlations were consistently observed at the team level over time between both supportive leadership and helping (.584 < r < .670 over 5 time points; all at p < 0.01), and between helping norms and helping (.720 < r < .831 over 5 time points; all at p < 0.01). Furthermore, at the individual level, in all five waves helping positively correlated with task performance (.209 < r < .404) and social integration (.287 < r < .411), and was negatively correlated with the intention to quit (−.325 < r < −.171) (p < 0.01 for all outcomes at all time points).

OPEN IN VIEWER

Table 1

Mean, Standard Deviation, and Correlation Coefficient of Each Variable

	Meanperson (SD)	Meanteam (SD)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
Individual
1. Gender	.413 (.493)	.427 (.188)	—	−.074	.011	.108	.136	.129	.181	.113	.150	.005	−.067	−.201*	.258*	.005	.079	.180
2. Age	31.184 (7.164)	31.387 (1.645)	−.006	—	−.130	−.080	−.096	−.195*	−.179	−.235*	−.169	−.109	.141	−.450**	.315**	.276**	−.047	−.132
3. Education	3.159 (.876)	3.141 (.304)	.028	−.205**	—	−.031	−.009	.044	−.091	−.055	−.014	.072	.028	.144	−.103	−.089	.009	−.039
4. Helping Wave1	4.604 (1.176)	4.645 (.739)	.036	.384**	−.126**	—	.909**	.884**	.872**	.857**	.676**	.624**	−.595**	−.188	−.039	.282**	.643**	.831**
5. Helping Wave2	4.912 (.947)	4.949 (.662)	.043	.183**	−.056	.711**	—	.923**	.885**	.884**	.666**	.666**	−.601**	−.116	−.033	.184	.670**	.735**
6. Helping Wave3	5.317 (.811)	5.336 (.497)	.039	−.139**	.025	.399**	.673**	—	.900**	.870**	.646**	.638**	−.618**	.029	−.103	.064	.584**	.720**
7. Helping Wave4	5.232 (.861)	5.244 (.503)	.002	−.035	−.030	.401**	.602**	.739**	—	.881**	.647**	.633**	−.571**	−.043	−.028	.082	.637**	.723**
8. Helping Wave5	4.822 (.936)	4.840 (.583)	−.005	.100**	−.068*	.527**	.685**	.661**	.697**	—	.648**	.691**	−.635**	−.037	−.047	.090	.628**	.769**
9. Task Performance	5.610 (1.087)	5.621 (.496)	.018	.125**	−.040	.310**	.317**	.209**	.219**	.404**	—	.585**	−.604**	−.012	−.034	.045	.537**	.578**
10. Social Integration	5.919 (.930)	5.933 (.447)	.022	−.004	−.003	.287**	.334**	.301**	.306**	.411**	.454**	—	−.604**	.117	−.131	−.021	.524**	.494**
11. Intention to Quit	3.243 (1.115)	3.257 (.455)	−.038	−.098**	.064	−.251**	−.248**	−.178**	−.171**	−.325**	−.487**	−.459**	—	−.126	.170	.007	−.501**	−.511**
12. Newcomer dummy	.396 (.489)	.384 (.063)	.000	−.821**	.191**	−.464**	−.219**	.134**	−.026	−.161**	−.171**	.003	.135**	—	−.650**	−.669**	−.089	−.165
13. Incumbent dummy	.150 (.358)	.159 (.046)	−.018	.169**	−.030	−.175**	−.089**	.005	.199**	.078*	.013	−.038	.003	−.341**	—	−.119	−.030	−.003
14. Veteran dummy	.453 (.498)	.458 (.049)	.013	.685**	−.166**	.581**	.279**	−.135**	−.117**	.103**	.159**	.025	−.135**	−.738**	−.383**	—	.165	.207*
Team
15. Supportive Leadership		4.325 (1.038)	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	.438**
16. Helping Norms		5.364 (0.636)	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—

Note: Correlations below the diagonal represent individual-level correlations (N = 898). Correlations above the diagonal represent team level correlations (N = 102). Mean_person (SD) and Mean_team (SD) are means and standard deviations computed at individual level and team level, respectively. Newcomer/Incumbent/Veteran: Binary variables (1 = group member, 0 = otherwise); * represents p < 0.05; ** represents p < 0.01.

Before testing our hypotheses, we established the fundamental comparability of our measures across time and employee categories by assessing measurement invariance. The results confirmed sufficient measurement equivalence (detailed findings are provided in Supplemental Appendix 2). To rigorously evaluate the construct validity of our measures within the multilevel structure of the data, we conducted multilevel confirmatory factor analysis (multilevel CFA). This model specified distinct latent factors at each level: a single factor for helping behaviors at the within-person level (Level 1), four factors (helping behaviors, task performance, intention to quit, and social integration) at the between-person level (Level 2), and three factors (helping behaviors, supportive leadership, and peer helping norms) at the between-team level (Level 3). The results indicated a good fit to the data (χ² = 439.443, DF = 389, CFI = .997, TLI = .996, RMSEA = .005, SRMR for within Level 1 = .010, SRMR for between Level 2 = .044, SRMR for between Level 3 = .041). To further examine the discriminant validity of our constructs, we tested three alternative models, with the findings supporting the discriminant validity of the measures (see Supplemental Appendix 3). Finally, a variance partitioning analysis for helping behaviors revealed significant variance components distributed across all three levels: a large portion resided at the within-person level (variance = .382, accounting for approximately 44.2% of the total variance), followed by the between-person level (variance = .205, approximately 23.7%), and the between-team level (variance = .278, approximately 32.1%). This substantial and statistically significant variance at each hierarchical level provides strong empirical justification for employing a three-level analytical approach in subsequent hypothesis testing.

Hypothesis Testing

All models demonstrated robust convergence, as indicated by a PSR factor close to 1 for each parameter, small Bayesian autocorrelation, and trace plots from the MCMC simulation showing good mixing (Jebb & Woo, 2015; Kaplan, 2014; Zyphur & Oswald, 2015).

Baseline Curvilinear Pattern and Group Differences (Hypotheses 1–2b)

Model 1 (Table 2) examined the baseline helping trajectories and differences between transitioning incumbents and newcomers (Hypotheses 1, 2a, and 2b). Hypothesis 1 predicted that helping behavior during role transitions would follow a curvilinear trajectory, initially increasing before stabilizing or declining. Supporting this, results revealed significant negative quadratic slopes for both newcomers (β_2ij = −.258, 95% CI [−.285, −.231]; Table 3) and transitioning incumbents (β_2ij) = −.194, 95% CI [−.221, −.166]; Table 3), confirming an inverted U-shaped pattern. The trajectory peaked at Time 3 for newcomers (M = 5.345, 95% CI [5.175, 5.512]; Table 4) and Time 4 for transitioning incumbents (M = 5.439, 95% CI [5.270, 5.612]; Table 4), followed by declines. Veterans (non-transitioning members) showed no significant curvature (β_2ij = −.005, 95% CI [−.026, .016]; Table 3), highlighting the uniqueness of transition dynamics. Figure 1 displays the plotted helping trajectories for newcomers, transitioning incumbents and veterans, demonstrating a pattern consistent with the curvilinear trajectory proposed in Hypothesis 1. These findings provide robust support for Hypothesis 1.

OPEN IN VIEWER

Table 2

Multilevel Analysis of Helping Behavior Trajectories

Level & component	Parameter	Estimate (SD)	95% Credibility Intervals (CIs)
Within-person (Time)
Helping behavior	Intercept (β0ij)
	Time (β1ij)
	Time² (β2ij)
	Residual variance	.213* (.006)	[.201, .226]
Between-person
β0ij	Newcomer dummy (γ01j)
	Incumbent dummy (γ02j)
	Gender	.014 (.041)	[−.066, .094]
	Age	.002 (.005)	[−.008, .012]
	Education	−.019 (.024)	[−.067, .029]
	Residual variance	.191* (.015)	[.164, .221]
β1ij	Newcomer dummy (γ11j)
	Incumbent dummy (γ12j)
	Gender	−.024 (.013)	[−.050, .002]
	Age	.001 (.002)	[−.002, .005]
	Education	−.006 (.008)	[−.022, .010]
	Residual variance	.010* (.002)	[.007, .014]
β2ij	Newcomer dummy (γ11j)
	Incumbent dummy (γ22j)
	Gender	−.010 (.009)	[−.028, .008]
	Age	.000 (.001)	[−.003, .002]
	Education	−.004 (.006)	[−.014, .007]
	Residual variance	.001* (.000)	[.001, .002]
Between-team
β0j	Intercept	5.166* (.070)	[5.027, 5.302]
	Residual variance	.296* (.051)	[.214, .415]
β1j	Intercept	−.093* (.015)	[−.122, −.064]
	Residual variance	.001* (.000)	[.001, .002]
β2j	Intercept	−.005 (.011)	[−.026, .016]
	Residual variance	.001* (.000)	[.001, .002]
Newcomer effects
γ01j	Intercept	.180* (.089)	[.005, .351]
	Residual variance	.177* (.045)	[.105, .278]
γ11j	Intercept	.306* (.028)	[.250, .361]
	Residual variance	.012* (.004)	[.007, .021]
γ21j	Intercept	−.253* (.019)	[−.290, −.216]
	Residual variance	.003* (.001)	[.001, .007]
Incumbent effects
γ02j	Intercept	.204* (.071)	[.064, .340]
	Residual variance	.166* (.062)	[.064, .310]
γ12j	Intercept	.357* (.025)	[.307, .407]
	Residual variance	.026* (.008)	[.014, .044]
γ22j	Intercept	−.189* (.015)	[−.217, −.160]
	Residual variance	.003* (.002)	[.001, .007]

Note: SD = posterior standard deviation. Italicized parameters were modeled as random effects. Estimates marked with * have 95% Credibility Intervals (CIs) excluding zero. Gender: Male (0) vs. Female (1). Newcomer/Incumbent: Binary variables (1 = group member, 0 = otherwise).

OPEN IN VIEWER

Table 3.

Conditional Trajectory Parameters for Role Transition Groups

Group	Intercept (SD) [95% CI]	Linear Slope (SD) [95% CI]	Quadratic Slope (SD) [95% CI]
Newcomer	5.345* (.086)[5.175, 5.512]	.213* (.021)[.172, .253]	−.258* (.014)[−.285, −.231]
Incumbent	5.368* (.087)[5.200, 5.537]	.265* (.024)[.217, .312]	−.194* (.014)[−.221, −.166]
Veteran	5.166* (.070)[5.027, 5.302]	−.093* (.015)[−.122, −.064]	−.005 (.011)[−.026, .016]
Difference (Incumbent − Newcomer)	.022 (.097)[−.167, .215]	.052 (.032)[−.011, .116]	.064* (.020)[.026, .103]

Note: SD = posterior standard deviation. Estimates marked with * have 95% Credibility Intervals (CIs) excluding zero.

OPEN IN VIEWER

Table 4

Predicted Helping Behavior Over Time by Role Transition Group

Group	Time 1(Entry)	Time 2	Time 3(Midpoint)	Time 4	Time 5
Newcomer	3.889 (.102)	4.875 (.087)	5.345 (.086)	5.300 (.087)	4.738 (.101)
	[3.690, 4.088]	[4.701, 5.045]	[5.175, 5.512]	[5.130, 5.470]	[4.538, 4.939]
Incumbent	4.064 (.104)	4.910 (.088)	5.368 (.087)	5.439 (.088)	5.121 (.103)
	[3.859, 4.268]	[4.740, 5.082]	[5.200, 5.537]	[5.270, 5.612]	[4.923, 5.323]
Veteran	5.332 (.078)	5.253 (.070)	5.166 (.070)	5.067 (.069)	4.959 (.077)
	[5.177, 5.486]	[5.117, 5.387]	[5.027, 5.302]	[4.933, 5.204]	[4.807, 5.112]

Note: Values are posterior means (SD) and [95% Credibility Intervals]. All estimates are significant (95% CIs exclude zero).

OPEN IN VIEWER

Figure 1

Helping Behavior Trajectories by Role Transition Group

Hypothesis 2a posited that transitioning incumbents would exhibit higher initial helping levels at entry (Time 1) than newcomers. To test this, we compared predicted helping levels at Time 1 between groups. As Table 4 shows, transitioning incumbents exhibited higher helping levels at entry (M = 4.064, 95% CI [3.859, 4.268]) compared to newcomers (M = 3.889, 95% CI [3.690, 4.088]). However, the model-constrained difference between groups was non-significant (difference = .174, 95% CI [−.067, .421]), and thus Hypothesis 2a was not supported.

Hypothesis 2b proposed that transitioning incumbents’ trajectories would be flatter than newcomers’. We tested this through differences in quadratic slopes. The significant quadratic difference (difference = .064, 95% CI [.026, .103]; Table 3) confirmed transitioning incumbents’ trajectories were flatter (less steep decline), supporting Hypothesis 2b. This pattern is further evident in Table 4: newcomers declined sharply from Time 3 to Time 5 (Δ = −.607), while incumbents declined more gradually (Δ = −.247).

Moderating Effects (Hypotheses 3a-b and 4a-b)

Next, Model 2a and 2b (Table 5) tested how team contexts—supportive leadership (Hypotheses 3a-b) and peer helping norms (Hypotheses 4a-b)—shaped helping trajectories. Hypothesis 3a posited that supportive leadership would boost initial helping at role entry (Time 1) for both newcomers and transitioning incumbents. Results strongly supported this: under high supportive leadership (1 SD above mean), newcomers exhibited significantly higher helping at entry (M_high = 4.549, 95% CI [4.318, 4.783]) than under low supportive leadership (M_low = 3.296, 95% CI [3.073, 3.526]; Table 6), with a difference of 1.252 (95% CI [.965, 1.538]). Similarly, transitioning incumbents showed higher entry helping under high supportive leadership (M_high = 4.721, 95% CI [4.463, 4.974]) versus low (M_low = 3.375, 95% CI [3.125, 3.636]; difference = 1.344, 95% CI [.985, 1.699]).

OPEN IN VIEWER

Table 5

Moderating Effects on Helping Behavior Trajectories

		Supportive leadershipas moderator		Peer descriptive helping normsas moderator
Level & component	Parameter	Estimate (SD)	95% Credibility Intervals (CIs)	Estimate (SD)	95% Credibility Intervals (CIs)
Within-person (Time)
Helping behavior	Intercept (β0ij)
	Time (β1ij)
	Time² (β2ij)
	Residual variance	.210* (.006)	[.198, .223]	.211* (.006)	[.199, .224]
Between-person
β0ij	Newcomer dummy (γ01j)
	Incumbent dummy (γ02j)
	Gender	.016 (.041)	[−.062, .097]	.028 (.039)	[−.047, .105]
	Age	.000 (.005)	[−.010, .010]	.000 (.005)	[−.010, .010]
	Education	−.022 (.024)	[−.070, .027]	−.025 (.024)	[−.072, .022]
	Residual variance	.196* (.015)	[.169, .227]	.177* (.013)	[.154, .204]
β1ij	Newcomer dummy (γ11j)
	Incumbent dummy (γ12j)
	Gender	−.023 (.013)	[−.049, .003]	−.019 (.013)	[−.045, .007]
	Age	.000 (.002)	[−.003, .004]	.001 (.002)	[−.003, .004]
	Education	−.006 (.008)	[−.021, .010]	−.007 (.008)	[−.022, .009]
	Residual variance	.010* (.002)	[.007, .015]	.011* (.002)	[.007, .015]
β2ij	Newcomer dummy (γ21j)
	Incumbent dummy (γ22j)
	Gender	−.010 (.009)	[−.028, .008]	−.014 (.009)	[−.032, .004]
	Age	.001 (.001)	[−.002, .003]	.000 (.001)	[−.002, .002]
	Education	−.003 (.006)	[−.014, .008]	−.002 (.006)	[−.014, .008]
	Residual variance	.001* (.000)	[.001, .002]	.001* (.000)	[.001, .002]
Between-team
β0j	Intercept	5.176* (.059)	[5.061, 5.290]	5.173* (.046)	[5.080, 5.264]
	Moderator	.399* (.047)	[.308, .493]	.985* (.056)	[.874, 1.093]
	Residual variance	.161* (.032)	[.110, .235]	.042* (.013)	[.021, .074]
β1j	Intercept	−.088* (.015)	[−.118, −.059]	−.092* (.015)	[−.122, −.063]
	Moderator	.018* (.009)	[.000, .036]	.012 (.016)	[−.020, .042]
	Residual variance	.001* (.000)	[.001, .002]	.001* (.000)	[.001, .002]
β2j	Intercept	−.011 (.010)	[−.031, .010]	−.006 (.010)	[−.026, .014]
	Moderator	−.019* (.007)	[−.032, −.005]	.006 (.012)	[−.018, .029]
	Residual variance	.001* (.000)	[.001, .002]	.001* (.000)	[.001, .001]
Newcomer effects
γ01j	Intercept	.148 (.088)	[−.023, .322]	.142 (.087)	[−.029, .315]
	Moderator	−.124* (.055)	[−.233, −.017]	−.656* (.093)	[−.837, −.474]
	Residual variance	.156* (.041)	[.088, .252]	.164* (.040)	[.099, .257]
γ11j	Intercept	.292* (.028)	[.237, .346]	.297* (.027)	[.243, .351]
	Moderator	−.073* (.016)	[−.105, −.042]	−.125* (.026)	[−.177, −.073]
	Residual variance	.009* (.003)	[.004, .016]	.007* (.003)	[.003, .014]
γ21j	Intercept	−.237* (.018)	[−.273, −.202]	−.244* (.018)	[−.281, −.208]
	Moderator	.074* (.010)	[.054, .093]	.101* (.017)	[.068, .136]
	Residual variance	.001* (.001)	[.001, .003]	.002* (.001)	[.001, .004]
Incumbent effects
γ02j	Intercept	.196* (.069)	[.062, .333]	.189* (.069)	[.056, .328]
	Moderator	−.190* (.067)	[−.322, −.060]	−.719* (.111)	[−.941, −.506]
	Residual variance	.147* (.060)	[.051, .284]	.161* (.056)	[.072, .290]
γ12j	Intercept	.356* (.024)	[.310, .403]	.355* (.024)	[.309, .401]
	Moderator	−.099* (.023)	[−.143, −.053]	−.144* (.038)	[−.217, −.071]
	Residual variance	.019* (.007)	[.008, .035]	.019* (.007)	[.008, .035]
γ22j	Intercept	−.186* (.014)	[−.214, −.158]	−.186* (.015)	[−.215, −.157]
	Moderator	.089* (.013)	[.063, .115]	.086* (.023)	[.040, .131]
	Residual variance	.002* (.001)	[.001, .004]	.003* (.002)	[.001, .007]

Note: SD = posterior standard deviation. Italicized parameters were modeled as random effects. Estimates marked with * have 95% Credibility Intervals (CIs) excluding zero. Gender: Male (0) vs. Female (1). Newcomer/Incumbent: Binary variables (1 = group member, 0 = otherwise).

OPEN IN VIEWER

Table 6

Conditional Trajectory Parameters under Team Context Moderators

Moderator	Group	Level	Intercept (SD) [95% CI]	Linear Slope (SD) [95% CI]	Quadratic Slope (SD) [95% CI]
Supportiveleadership	Newcomer	Low	5.040* (.100) [4.843, 5.236]	.261* (.024) [.214, .309]	−.306* (.015) [−.335, −.275]
		High	5.607* (.101) [5.409, 5.809]	.147* (.025) [.098, .196]	−.191* (.016) [−.223, −.160]
	Incumbent	Low	5.155* (.107) [4.948, 5.367]	.351* (.032) [.286, .414]	−.269* (.019) [−.305, −.232]
		High	5.589* (.107) [5.379, 5.800]	.185* (.031) [.125, .241]	−.124* (.018) [−.160, −.088]
	Veteran	Low	4.765* (.075) [4.616, 4.911]	−.107* (.017) [−.141, −.073]	.009 (.013) [−.016, .034]
		High	5.589* (.077) [5.437, 5.741]	−.069* (.018) [−.105, −.034]	−.030* (.013) [−.055, −.004]
Peerhelping norms	Newcomer	Low	5.107* (.088) [4.932, 5.282]	.277* (.024) [.231, .324]	−.318* (.015) [−.349, −.288]
		High	5.524* (.089) [5.350, 5.701]	.133* (.025) [.086, .182]	−.182* (.016) [−.215, −.150]
	Incumbent	Low	5.194* (.095) [5.008, 5.383]	.347* (.032) [.285, .409]	−.250* (.019) [−.287, −.213]
		High	5.533* (.096) [5.344, 5.723]	.179* (.031) [.118, .241]	−.134* (.018) [−.170, −.097]
	Veteran	Low	4.549* (.058) [4.438, 4.663]	−.099* (.018) [−.134, −.065]	−.010 (.013) [−.034, .015]
		High	5.797* (.059) [5.681, 5.912]	−.084* (.018) [−.120, −.049]	−.002 (.013) [−.027, .023]

Note: SD = posterior standard deviation. Estimates marked with * have 95% Credibility Intervals (CIs) excluding zero.

Hypothesis 3b predicted that supportive leadership would flatten trajectories (i.e., weaken curvilinearity) for both groups. This was tested through differences in quadratic slopes between high/low leadership conditions. For newcomers, the quadratic slope was significantly less negative under high supportive leadership (β_2ij = −.191, 95% CI [−.223, −.160]) than under low supportive leadership (β_2ij = −.306, 95% CI [−.335, −.275]; Table 6), with a difference of .114 (95% CI [.080, .149]). Transitioning incumbents mirrored this pattern (high: β_2ij = −.124, 95% CI [−.160, −.088]; low: β_2ij = −.269, 95% CI [−.305, −.232]; difference = .145, 95% CI [.097, .195]), confirming flatter trajectories under supportive leadership. Supportive leadership explained 11.9% (pseudo R² of 0.119) of between-team variance in newcomers’ intercept effects (γ_01j) and 11.4% for transitioning incumbents’ (γ_02j), indicating modest moderation of the overall level of helping. More substantially, it accounted for 66.7% of between-team variance in newcomers’ quadratic slope effects (γ_21j) and 33.3% for transitioning incumbents’ (γ_22j), underscoring its strong role in sustaining helping behavior. It also explained 25.0% (newcomers, γ_11j) and 26.9% (transitioning incumbents, γ_12j) of linear slope variance, highlighting broad influence on trajectory dynamics.

Hypotheses 4a-b replicated these tests for peer helping norms. Hypothesis 4a (peer norms → initial helping) was fully supported: newcomers (M_high = 4.527, 95% CI [4.317, 4.745]; M_low = 3.282, 95% CI [3.078, 3.492]; difference = 1.246, 95% CI [.985, 1.508]) and transitioning incumbents (M_high = 4.635, 95% CI [4.391, 4.878]; M_low = 3.500, 95% CI [3.262, 3.739]; difference = 1.135, 95% CI [.797, 1.468]) showed stronger entry helping under high norms (Table 6). Hypothesis 4b (peer norms → flatter trajectories) was also supported: quadratic slopes were less negative for newcomers (high: β_2ij = −.182, 95% CI [−.215, −.150]; low: β_2ij = −.318, 95% CI [−.349, −.288]; difference = .136, 95% CI [.100, .172]) and transitioning incumbents (high: β_2ij = −.134, 95% CI [−.172, −.097]; low: β_2ij = −.250, 95% CI [−.287, −.213]; difference = .115, 95% CI [.064, .168]) under high norms. Figure 2 illustrates the moderating effects of supportive leadership and peer helping norms on the plotted helping trajectories for newcomers, transitioning incumbents, and veterans, and Table 7 provides the corresponding predicted levels of helping behaviors over time. Peer helping norms explained less between-team variance in the overall level of helping than leadership (newcomers γ_01j: pseudo R² = 7.3%; transitioning incumbents γ_02j: pseudo R² = 3.0%). However, they accounted for 41.7% of newcomers’ linear slope variance (γ_11j) and 26.9% for transitioning incumbents’ (γ_12j), reflecting strong acceleration of early helping growth. Their association with quadratic slopes was more limited (newcomers γ_21j: pseudo R² = 33.3%; transitioning incumbents γ_22j: pseudo R² = 0%), suggesting weaker influence on long-term sustainability.

OPEN IN VIEWER

Figure 2

Moderating Effects on Helping Behavior Trajectories

OPEN IN VIEWER

Table 7.

Predicted Helping Behavior Over Time by Role Transition Group under Team Context Moderators

Moderator	Group	Level	Time 1 (Entry)	Time 2	Time 3 (Midpoint)	Time 4	Time 5
Supportive Leadership	Newcomer	Low	3.296 (.115)	4.473 (.101)	5.040 (.100)	4.996 (.101)	4.340 (.115)
			[3.073, 3.526]	[4.276, 4.673]	[4.843, 5.236]	[4.796, 5.191]	[4.112, 4.565]
		High	4.549 (.118)	5.269 (.102)	5.607 (.101)	5.562 (.101)	5.135 (.116)
			[4.318, 4.783]	[5.070, 5.473]	[5.409, 5.809]	[5.365, 5.762]	[4.911, 5.367]
	Incumbent	Low	3.375 (.132)	4.534 (.109)	5.155 (.107)	5.238 (.108)	4.784 (.129)
			[3.125, 3.636]	[4.323, 4.750]	[4.948, 5.367]	[5.025, 5.450]	[4.524, 5.035]
		High	4.721 (.129)	5.279 (.108)	5.589 (.107)	5.649 (.109)	5.461 (.129)
			[4.463, 4.974]	[5.065, 5.494]	[5.379, 5.800]	[5.433, 5.859]	[5.203, 5.712]
	Veteran	Low	5.012 (.087)	4.880 (.076)	4.765 (.075)	4.666 (.075)	4.585 (.086)
			[4.841, 5.180]	[4.730, 5.026]	[4.616, 4.911]	[4.518, 4.815]	[4.418, 4.754]
		High	5.607 (.089)	5.628 (.077)	5.589 (.077)	5.489 (.077)	5.331 (.089)
			[5.434, 5.786]	[5.476, 5.780]	[5.437, 5.741]	[5.339, 5.642]	[5.157, 5.505]
Peer Helping Norms	Newcomer	Low	3.282 (.105)	4.512 (.089)	5.107 (.088)	5.065 (.088)	4.389 (.104)
			[3.078, 3.492]	[4.339, 4.688]	[4.932, 5.282]	[4.895, 5.241]	[4.185, 4.593]
		High	4.527 (.109)	5.208 (.090)	5.524 (.089)	5.475 (.089)	5.059 (.107)
			[4.317, 4.745]	[5.035, 5.387]	[5.350, 5.701]	[5.300, 5.652]	[4.852, 5.275]
	Incumbent	Low	3.500 (.121)	4.596 (.097)	5.194 (.095)	5.291 (.096)	4.890 (.120)
			[3.262, 3.739]	[4.404, 4.790]	[5.008, 5.383]	[5.103, 5.483]	[4.651, 5.128]
		High	4.635 (.123)	5.217 (.098)	5.533 (.096)	5.577 (.098)	5.352 (.122)
			[4.391, 4.878]	[5.026, 5.412]	[5.344, 5.723]	[5.383, 5.765]	[5.111, 5.592]
	Veteran	Low	4.711 (.072)	4.640 (.058)	4.549 (.058)	4.441 (.057)	4.313 (.071)
			[4.572, 4.854]	[4.527, 4.751]	[4.438, 4.663]	[4.329, 4.557]	[4.175, 4.456]
		High	5.956 (.073)	5.879 (.058)	5.797 (.059)	5.710 (.059)	5.620 (.073)
			[5.816, 6.102]	[5.764, 5.995]	[5.681, 5.912]	[5.596, 5.825]	[5.478, 5.763]

Note: SD = posterior standard deviation. All estimates are significant (95% CIs exclude zero).

Outcomes of Helping Trajectories

Model 3 (Table 8) evaluated how helping trajectories predicted three critical socialization outcomes: task performance, social integration, and intention to quit. Hypothesis 5a predicted that higher levels of helping are positively associated with higher task performance and social integration, and inversely associated with intention to quit. This hypothesis received significant support: Individuals who exhibited higher levels of helping received significantly higher task performance ratings from their team leader (intercept: β_0ij = .383, 95% CI [.281, .486]) and reported greater social integration (β_0ij = .454, 95% CI [.366, .543]). Additionally, higher helping levels were associated with lower intention to quit (β_0ij = −.317, 95% CI [−.426, −.207]). These findings suggest that displaying a high level of helping behavior during the transition period enhances both instrumental and relational outcomes, while also reducing withdrawal intentions.

OPEN IN VIEWER

Table 8

Effects of Helping Trajectories on Socialization Outcomes

Level & component	Parameter	Estimate (SD)	95% Credibility Intervals (CIs)
Within-person (Time)
Helping	Intercept (β0ij)
	Time (β1ij)
	Time² (β2ij)
	Residual variance	.214* (.007)	[.199, .229]
Between-person
β0ij	Intercept	5.203* (.058)	[5.090, 5.319]
	Newcomer dummy	.006 (.102)	[−.200, .203]
	Incumbent dummy	.150 (.081)	[−.010, .310]
	Gender	.060 (.054)	[−.043, .165]
	Age	−.010 (.007)	[−.023, .003]
	Education	−.020 (.031)	[−.082, .041]
	Residual variance	.519* (.028)	[.468, .577]
β1ij	Intercept	−.091* (.015)	[−.121, −.063]
	Newcomer dummy	.302* (.026)	[.252, .353]
	Incumbent dummy	.362* (.021)	[.322, .402]
	Gender	−.023 (.014)	[−.050, .004]
	Age	.001 (.002)	[−.003, .004]
	Education	−.003 (.008)	[−.018, .013]
	Residual variance	.019* (.002)	[.015, .024]
β2ij	Intercept	−.003 (.010)	[−.023, .018]
	Newcomer dummy	−.266* (.018)	[−.301, −.231]
	Incumbent dummy	−.191* (.015)	[−.220, −.163]
	Gender	−.006 (.010)	[−.025, .013]
	Age	−.001 (.001)	[−.004, .001]
	Education	−.006 (.005)	[−.017, .005]
	Residual variance	.004* (.001)	[.002, .006]
Outcome Models
Task performance	Intercept	3.979* (.296)	[3.406, 4.559]
	β0ij	.383* (.053)	[.281, .486]
	β1ij	1.354* (.390)	[.614, 2.135]
	β2ij	7.590* (2.665)	[4.636, 14.706]
	Newcomer dummy	1.063* (.718)	[.212, 2.940]
	Incumbent dummy	.742* (.528)	[.095, 2.099]
	Gender	.096 (.096)	[−.085, .293]
	Age	.002 (.012)	[−.019, .027]
	Education	.040 (.055)	[−.059, .160]
	Residual variance	.819* (.098)	[.584, .969]
Social Integration	Intercept	3.656* (.254)	[3.154, 4.152]
	β0ij	.454* (.045)	[.366, .543]
	β1ij	.953* (.325)	[.332, 1.616]
	β2ij	6.163* (2.093)	[3.741, 11.751]
	Newcomer dummy	1.306* (.568)	[.614, 2.827]
	Incumbent dummy	.657* (.419)	[.135, 1.746]
	Gender	.077 (.078)	[−.069, .232]
	Age	.011 (.010)	[−.007, .031]
	Education	.046 (.045)	[−.034, .143]
	Residual variance	.611* (.063)	[.463, .713]
Intention to Quit	Intercept	4.602* (.313)	[3.982, 5.210]
	β0ij	−.317* (.055)	[−.426, −.207]
	β1ij	−1.291* (.406)	[−2.120, −.533]
	β2ij	−6.899* (2.528)	[−13.770, −3.988]
	Newcomer dummy	−.978* (.687)	[−2.774, −.142]
	Incumbent dummy	−.619 (.508)	[−1.910, .030]
	Gender	−.147 (.095)	[−.346, .029]
	Age	−.001 (.012)	[−.026, .020]
	Education	.002 (.056)	[−.116, .102]
	Residual variance	.981* (.095)	[.756, 1.133]

Note: SD = posterior standard deviation. Italicized parameters were modeled as random effects. Estimates marked with * have 95% Credibility Intervals (CIs) excluding zero. Gender: Male (0) vs. Female (1). Newcomer/Incumbent: Binary variables (1 = group member, 0 = otherwise).

Hypothesis 5b posited that flatter helping trajectories—i.e., those characterized by lower curvature and thus greater behavioral consistency—would be linked to more positive socialization outcomes. Again, the results were consistent with this expectation. Individuals with a trajectory characterized by a flatter curvature demonstrated significantly better leader-rated task performance (quadratic slope: β_2ij = 7.590, 95% CI [4.636, 14.706]). A similar pattern emerged for social integration, which was greater when the trajectory was characterized by a flatter curvature (β_2ij = 6.163, 95% CI [3.741, 11.751]). Finally, intention to quit was lower as a function of diminished curvature (i.e., when helping was maintained more consistently over time; β_1ij = −1.291, 95% CI [−2.120, −.533]; β_2ij = −6.899, 95% CI [−13.770, −3.988]). These findings suggest that inconsistency in helping—especially when it declines sharply after an initial peak—may signal unreliability to others and thus undermine perceptions of competence, connectedness, and embeddedness.

Overall, the models explained 29.0% of the variance in task performance (pseudo R² = .290), 31.0% in social integration (pseudo R² = .310), and 21.6% in intention to quit (pseudo R² = .216). Among the trajectory components, quadratic slope (curvature) consistently showed the strongest effects across outcomes, underscoring the importance of helping stability in shaping how newcomers and transitioners are perceived and integrated.

Supplementary Analysis

We conducted several robustness tests to rule out alternative explanations. First, we reestimated the models testing Hypotheses 3a-b, and 4a-b taking into account (a) gender, age and education as alternative demography-based explanations, as well as (b) two personality factors, namely proactive personality (Parker, 1998) and relationship interdependent self-construal (Cross, Bacon, & Morris, 2000), and (c) team task interdependence (Bishop & Scott, 2000) as additional individual difference and contextual confounds. The inclusion of these variables in the newcomer and transitioning incumbent models did not result in any meaningful differences in the magnitude or significance of the estimates for either supportive leadership or helping norms (results available from the authors). Additionally, we ran an exploratory analysis comparing the effect sizes (variance explained) of supportive leadership and helping norms. Supportive leadership demonstrates a stronger moderating effect on the non-linear aspects (quadratic slopes) of helping behavior trajectories, particularly for newcomers, where it explains 66.7% of the variance ³ in curvature (γ_21j). In contrast, peer helping norms exert a stronger influence on the linear rates of change (linear slopes), especially for newcomers (41.7% variance explained in γ_11j), suggesting that norms more directly shape the consistent pace of behavioral adaptation over time. While both contextual factors significantly moderate key trajectory parameters (e.g., linear slopes for transitioning incumbents similarly affected by both), leadership uniquely buffers early volatility in newcomer behavior, whereas peer norms provide sustained guidance for ongoing behavioral change.

Theoretical Implications

Our study offers several important contributions to research and theory on helping and newcomer socialization. First, we demonstrate that already early in the transition process, those entering a new work role engage in helping behavior, with their level of engagement increasing over the course of their initial onboarding. From the perspective of organizational socialization, this is important in that it extends our understanding of the implications of the identity shifts often accompanying role transitions, suggesting that they have rather immediate implications for newcomer helping behavior (Ashforth, 2001). Moreover, it suggests the need to extend current conceptualizations of helping in socialization from a focus on newcomers as help-seekers and as the targets of helping to a more agentic conceptualization in which those in transition increasingly contribute from their own human and social capital to the units they join. From an OCB perspective, this important in that in contrast to much of the helping literature, which assumes helping to be relatively static over the longer-term (Grant, 2013), at least among those transitioning into a new work role, rather than just exhibiting small, short-term fluctuations, our findings indicate that helping is likely to manifest meaningful, longer-wave shifts. In that sense, our findings offer some of the first empirical support to the self-regulation model of OCB (Bolino, Harvey, & Bachrach, 2012) and TTOCB (Methot et al., 2017) which suggest that employee citizenship varies over time (e.g., minute by minute, over the course of a day, a year, or a career).

Second, our theorizing proposed (and findings confirmed) that although this shift among those transitioning generally mirrors the baseline, curvilinear trajectory proposed by the TTOCB, there are predictable variations in this baseline pattern, with these variations in trajectory having meaningful implications for three key adjustment-related outcomes. More specifically, higher and flatter helping trajectories are associated with heightened levels of task performance and social integration and lower intention to quit, explaining a substantial proportion of their variance. From a theoretical perspective, this is important in that it suggests that helping (and not just help-seeking) serves as an additional factor influencing the adjustment of those transitioning into a new role. Moreover it suggests the need to consider the overall trajectory of helping over the course of onboarding rather than just discrete or mean levels of helping at one point or another, with these patterns potentially offering important insights into how quickly those in transition socially integrate and develop the kind of supportive exchange relationships critical to task mastery and performance, not to mention the stability of these relationships.

Third, in the context of such long-wave effects, we proposed and demonstrated two key factors potentially influencing helping trajectories for those transitioning into a new work role. While these are certainly not the only work-based contextual factors influencing the helping trajectories of those in transition, the finding that both supportive leadership and helping norms impact not only starting levels of helping but also the stability of such behavior over time offers an important extension to our understanding of how insider (i.e., leader and peer) behavior, by shaping the helping trajectories of those in transition, can influence key socialization outcomes. Additionally, our post-hoc finding that peer norms influence helping trajectories differently than leadership style among newcomers also has important theoretical implications, suggesting that in understanding how helping emerges, changes over time, and ultimately shapes newcomer adjustment, we need to pay no less attention to the sense-giving role of veteran peer norms than to that of team leaders.

Finally, we offer an important contribution to both the TTOCB and research on work-related transitions in that we proposed and demonstrated that the helping trajectories of those entering a role from outside an organization (i.e., newcomers) differ from those of individuals transitioning from within (i.e., transitioning incumbents). Although researchers have speculated that the adjustment experiences of newcomers and transitioning incumbents differ, and that newcomers apply different socialization strategies than transitioning incumbents (Zhu et al., 2017), empirical research has to date largely focused on the former. Consequently, we know little about the experience of organizational incumbents transitioning into new roles and the role that collaborative behavior may play in affecting their transition success. Our findings confirm that while transitioning incumbents also manifest the baseline, inversed U-shaped helping trajectory predicted by TTOCB, this trajectory is flatter than that of newcomers, suggesting that collaborative behavior among these transitioning individuals tends to be more stable over time than among organizational newcomers. Moreover, our findings suggest that supportive leadership and peer norms are more influential in shaping these trajectories for newcomers than for transitioning incumbents.

Practical Implications

Our findings also have important practitioner implications. First, they suggest that the helping trajectories of those transitioning have meaningful implications not only on the task performance of the transitioning individual, but on their social integration and retention intentions as well. Accordingly, leaders should probably take a more proactive role in motivating positive helping patterns among new entrants. For example, during regular check-ins or performance reviews, leaders might discuss not only the help that transitioning employees receive but also the help they provide. By explicitly recognizing and valuing helping behavior, especially when sustained over time, leaders can signal its importance and embed it more firmly into team norms. Furthermore, leaders might consider developing more nuanced strategies by actively monitoring and identifying different helping trajectory patterns over time. Rather than applying a one-size-fits-all approach, managers could segment employees based on their helping patterns (e.g., consistently high, declining, or increasing) and tailor interventions accordingly. For instance, individuals showing early signs of declining helping might benefit from targeted developmental conversations, while those with rising helping patterns could be supported with opportunities for informal leadership or mentoring roles. By adopting a differentiated approach to motivating helping behaviors, organizations might more effectively foster and sustain high-functioning, supportive work environments during critical transition periods.

Second, beyond providing new entrants with the motivation to provide assistance, our findings suggest that higher and flatter helping trajectories may be elicited by providing new entrants with greater helping opportunities. To do this, onboarding programs might be redesigned to foster bi-directional helping from the start. For instance, leaders might strategically assign newcomers and transitioning incumbents to teams in which their experience and background may be of particular value. Highlighting this to both the new entrants and their teammates may make the latter more open to the assistance of the new entrants, thus encouraging them to not only seek help but provide it as well. Additionally, providing newcomers with early opportunities to share their prior knowledge, skills, or external perspectives can empower them as contributors, rather than passive recipients of guidance. In doing so, organizations may not only improve newcomer outcomes, but also cultivate stronger helping norms throughout the unit or team.

Indeed, organizational leaders may want to make a concerted effort to shift and strengthen peer helping norms so as to maximize the odds of eliciting higher and flatter helping trajectories. For example, leaders may want to be sure that organizational reward strategies and incentive structures are aligned to encourage prosocial behavior rather than discouraging it. Additionally, leaders may try to celebrate the successes of highly respected peers that consistently model high levels of helping, highlighting how their behavior not only enhanced their own individual outcomes, but furthered the interest of the team or work unit as a whole.

Finally, particularly for newcomers, leaders may consider taking steps to “flatten the curve” at the six-month inflection point such that peak levels of helping are stabilized rather than reduced. For instance, they might conduct a review of in-role expectations to ensure that those in transitioning have sufficient “slack” to maintain their high level of helping (Lin, Koopmann, & Wang, 2020).

Limitations and Future Research Directions

Despite these important contributions, several study limitations may limit the generalizability of our findings as well as present opportunities for future research. First, while newcomers and transitioning incumbents in Chinese firms typically enter their new positions as a cohort at set times of the year, and thus go through their on-boarding contemporaneously, this may be less typical for enterprises elsewhere. To the extent that the effects demonstrated may stem from such cohort-based on-boarding experiences, they may be less generalizable to organizations outside of China. Second, lacking data on changes in team leadership during the on-boarding period, we cannot rule out the possibility that team leadership may have changed midstream. While the impact of such changes should be examined in future research, to the extent that such changes occurred in the current sample they logically would have only increased the probability of Type II error, thus suggesting that, if anything, our results err on the conservative. Third, helping behavior is inherently a reciprocal process (Jia et al., 2021), yet this study focuses solely on the trajectories of helping given rather than helping seeking. While this does not undermine our conclusions, future research should consider how different forms of helping (e.g., proactive vs. reactive) influence socialization and explore the underlying mechanisms through which helping trajectories shape socialization outcomes. A more comprehensive approach to studying both helping given and helping sought would provide a deeper understanding of the reciprocal nature of helping behaviors.

Future research may also benefit by examining newcomer helping from the perspective of others. Although our data suggests a moderate correlation between self-assessments of helping and leader assessments of the help given by those transitioning, studies applying a more sociometric approach would be helpful. Additionally, studies are needed to examine other potential moderators of the helping trajectories of those in transition. For instance, future studies could investigate how pay structures and incentives (e.g., team-based pay; Bamberger & Levi, 2009) and the intensity of work pace might serve as additional cues moderating helping trajectories. Similarly, individual differences might also be investigated as moderators of helping trajectories. For example, individuals’ prosocial values (Grant & Rothbard, 2013) and helping identity (Farmer & Van Dyne, 2010) influence their ability to recognize and respond to others’ needs for assistance, thereby accounting for additional variance in helping trajectories.