📄 social_science

A longitudinal investigation of social norms and social networks in science

This longitudinal study of 93 Canadian academic trainees reveals that while social experiences and satisfaction reflect both enduring individual differences and responsiveness to changing conditions, academic support and the ability to discuss failure with supervisors are the strongest predictors of positive research outcomes, whereas network characteristics poorly predict adherence to Mertonian norms.

Original authors: Holly J. Kalyn Bogard, Justin Petluk, Calvin Lloyd, Louise Barrett

Published 2026-08-20

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Holly J. Kalyn Bogard, Justin Petluk, Calvin Lloyd, Louise Barrett

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Science is often imagined as a solitary pursuit, where a lone researcher in a quiet lab uncovers a hidden truth. In reality, scientific knowledge is built by communities. Researchers rely on one another for collaboration, for checking each other's work, and for the emotional resilience needed to navigate years of study. Within these communities, there are unwritten rules about how scientists should behave—ideals like sharing data openly, judging work based on merit rather than the author's fame, and prioritizing truth over personal gain. These are the social norms that hold the scientific enterprise together. But there is often a gap between how scientists say they should act and how they actually act in their daily work. Understanding this gap, and how it affects the people training to become scientists, is crucial for the future of research. If the environment feels hostile or unfair, talented individuals may leave, or the quality of the science itself may suffer.

To understand how these social dynamics play out over time, a team of researchers at the University of Lethbridge in Canada followed a group of 93 academic trainees—graduate students and postdoctoral fellows—over a period of eighteen months. Instead of taking a single snapshot of their experiences, the researchers checked in with them four times, at six-month intervals. This approach allowed them to see if the feelings and relationships they reported were just fleeting moods or stable parts of their lives. They asked the trainees about their satisfaction with their research, the support they received from their supervisors and peers, and their views on whether scientists are living up to the ideal standards of their profession. The goal was to determine what truly drives a trainee's happiness and success: is it the quality of their social connections, the stability of their support system, or something else entirely?

The study revealed that a trainee's experience is shaped by a mix of who they are as a person and the changing conditions around them. When the researchers looked at how trainees rated scientific norms, they found that these views were remarkably stable. A student who believed strongly in the importance of open sharing in the first survey tended to hold that same belief six months later. However, the researchers also detected small, meaningful shifts. These fluctuations suggest that while a person's core values remain steady, their daily experience of the scientific environment can change, and those changes matter. The most significant finding was that a trainee's satisfaction with their current experience was not primarily driven by the number of people in their network or the technical help they received. Instead, it hinged on the quality of their emotional relationships, particularly with their supervisor.

The single strongest predictor of whether a trainee felt satisfied with their academic life was their comfort level in discussing failure with their supervisor. Trainees who felt they could talk openly about their mistakes, setbacks, or struggles without fear of judgment reported much higher levels of satisfaction. This comfort was not just a temporary feeling; it was a stable trait that persisted over time. Yet, the study also showed that when a trainee's comfort level increased during a specific period, their satisfaction with their academic support and overall experience improved as well. This indicates that the relationship between a student and their supervisor is not fixed; it can evolve, and when it becomes more open and supportive, the trainee's entire academic experience improves.

While emotional support was the key to satisfaction, the study also highlighted the role of the broader social network. Having a few very close colleagues and fewer difficult relationships was linked to better feelings about academic support. Interestingly, the sheer number of people providing technical help was less important than the emotional closeness of the relationships. The researchers found that trainees who felt supported by a tight-knit, positive group were more likely to feel that their supervisors were helpful. However, the study also uncovered that certain groups of trainees faced distinct challenges. International students, women, and those who felt underrepresented in their field reported lower satisfaction with financial support and, in some cases, lower satisfaction with academic support. These groups also tended to perceive that scientists were less likely to follow the ideal norms of fairness and openness compared to their peers. This suggests that structural inequalities in the academic system are felt not just in material terms, like funding, but in the daily social experience of the trainee.

The researchers also looked at whether having a supportive network or being happy with one's training made a trainee more likely to believe that scientists were actually following the ideal norms of the profession. Surprisingly, the answer was no. A trainee's social environment and their personal satisfaction did not strongly predict how they viewed the behavior of scientists in general. This suggests that perceptions of how science works are shaped by larger, more systemic forces—such as funding models, departmental politics, or the culture of a specific discipline—rather than just the immediate relationships a student has with their peers and supervisor. The disconnect between a trainee's happy, supportive daily life and their view that the broader scientific community is failing to live up to its ideals points to a deeper issue. It implies that even when individual relationships are strong, the broader pressures of the academic system, such as intense competition for jobs and funding, may be eroding the shared values that hold the scientific community together.

Ultimately, this longitudinal investigation paints a picture of academic training as a journey where stability and change coexist. A trainee's satisfaction is deeply rooted in the emotional safety of their relationship with their supervisor and the quality of their close connections. While a student's core beliefs about science remain steady, their daily experience is responsive to the support they receive. The findings suggest that to improve the lives of future scientists, institutions and supervisors must focus on fostering environments where failure can be discussed openly and where the unique challenges faced by marginalized groups are acknowledged. By strengthening these emotional and relational foundations, the scientific community can better support the people who will carry its work forward, even as it grapples with the larger structural pressures that threaten its ideals.

Technical Summary: A Longitudinal Investigation of Social Norms and Social Networks in Science

Problem Statement
Empirical research regarding scientists' beliefs, norms, and social relationships has historically relied on cross-sectional survey designs. While effective for estimating the "current state" of a population, such designs cannot distinguish whether observed patterns represent stable, enduring features of academic culture or transient artifacts of a single measurement occasion. This limitation hinders the ability to evaluate developmental career-stage transitions or the dynamic flux of scientific social systems. Specifically, previous cross-sectional studies by the authors identified tensions between idealized Mertonian norms and enacted behaviors, as well as associations between trainee satisfaction and social resource exchange. However, the temporal stability of these findings and the causal ordering of these relationships remained untested. This study addresses that gap by treating response stability as an empirical question rather than an assumption.

Methodology
The study employed a longitudinal design utilizing four waves of survey data collected at six-month intervals between 2024 and 2026 from 93 Canadian academic trainees (graduate students and postdoctoral fellows). The sample included respondents from the Universities of Calgary, Lethbridge, and Alberta. To minimize anchoring effects, previous responses were cleared before each wave, though network composition (20 alters) was retained to reduce respondent fatigue, with options for modification.

The analysis utilized a Bayesian mixed-effects modeling framework (using the brms package in R) to handle small or unbalanced samples and explicitly capture uncertainty. The study applied three complementary analytical approaches:

Repeated-measures analyses: To assess the stability of associations over time and estimate between-person versus within-person variance.
Time-lagged analyses: To examine whether values at Wave $i$ predict outcomes at Wave $i+1$ , providing evidence for temporal ordering.
Variance decomposition: To distinguish stable individual differences from within-person changes using Variance Partition Coefficients (VPC).

Models were multilevel and multivariate, utilizing cumulative logit distributions for ordinal outcomes (e.g., norm ratings, satisfaction) and binomial or skew-normal distributions for network characteristics. Convergence was verified via standard Bayesian diagnostics (R-hat < 1.01, trace plots, posterior predictive checks). Sensitivity analyses were conducted to test for dropout bias, response anomalies, and prior sensitivity.

Key Contributions
This paper contributes to the literature by:

Providing the first longitudinal evaluation of the stability of social norm perceptions and network-satisfaction associations in academia.
Decomposing variance in social norm endorsements and satisfaction metrics into between-person (stable) and within-person (fluctuating) components.
Testing the robustness of previously identified cross-sectional findings regarding the tension between ideal and actual scientific norms and the predictors of trainee satisfaction.
Offering a nuanced view of how group differences (gender, international status, career stage) persist or shift over time in the context of academic training.

Results

Stability of Social Norms: Mean responses to 17 social norm items were robust across waves. Variance decomposition revealed that stable between-person differences accounted for 25% to 64% of total variance (VPC median range 0.25–0.64), indicating that while individual differences are substantial, there is also detectable within-person fluctuation. No consistent patterns of change emerged across the 53 repeated-measures analyses, suggesting that perceptions of norms are largely stable traits rather than transient states.
Predictors of Satisfaction: Academic support emerged as the central correlate of both current experience and research productivity. Financial support and research productivity contributed additional, smaller effects. Crucially, network characteristics were poor indicators of perceived adherence to Mertonian norms. However, network quality (specifically having more "very" close relationships and fewer difficult ones) predicted academic support more strongly than the volume of technical support provided.
Supervisor Emotional Support: The strongest single predictor of trainee satisfaction was the comfort level discussing academic failure with a supervisor. This relationship was graded (being "very" comfortable yielded significantly higher odds of satisfaction than "somewhat" comfortable) and showed strong autoregressive effects, indicating it is a stable feature of the supervisory relationship. However, within-person increases in comfort were also associated with improved satisfaction, suggesting the relationship is not entirely fixed.
Group Differences: Modest group differences were observed. International students, women, and those feeling underrepresented tended to report lower adherence to Mertonian norms in practice compared to ideals. Postdoctoral fellows reported lower satisfaction with financial support and fewer close network members, aligning with the precarity of their career stage.
Norm Perception vs. Social Environment: Contrary to the hypothesis that immediate social environments shape norm perceptions, network characteristics and satisfaction were poor indicators of perceived adherence to Mertonian norms. The authors suggest a mismatch between relational experience and norm perception, possibly because norm evaluations reflect broader structural or institutional features rather than micro-level network composition.

Significance and Claims
The authors claim that trainees' social experiences and satisfaction reflect a combination of enduring individual differences and responsiveness to changing conditions. The study demonstrates that while between-person differences are substantial, within-person fluctuations are detectable and meaningful, implying that trainees respond to changes in their material and social conditions.

The paper modestly concludes that these findings have implications for promoting positive research experiences:

Interventions should distinguish between stable individual differences (which require tailored approaches) and modifiable drivers of within-person change (which can be targeted by structural or relational improvements).
Emotional support, particularly the ability to discuss failure, is a central, stable, yet malleable dimension of effective supervision.
The observed gaps between ideal and actual norms likely reflect structural pressures (e.g., competition, funding models) that incentivize behaviors contrary to Mertonian ideals, rather than merely individual moral failings.
The lack of association between network characteristics and norm perception suggests that perceptions of scientific norms may be shaped more by institutional and disciplinary structures than by immediate interpersonal networks.

The authors emphasize that their results are exploratory and hypothesis-generating due to sample size constraints, particularly for subgroups like postdoctoral fellows, and caution against over-interpreting modest effect sizes.

Technical Summary: A Longitudinal Investigation of Social Norms and Social Networks in Science

More like this