2026-09-26 Weekly Good Reads: The Hidden Promotion Tax of Remote Work

Why I picked it

Over the past week, WeChat forwards kept circling back to a few interlocking questions: whether management is actually creating value or just building fragile power structures, how ordinary people can carve out a future for themselves, what catches you when opportunity appears, and whether job opportunities themselves will become scarcer as technology shifts. Together they point to one real problem: organizations don't identify people by output alone—they also use visibility, presence, and existing impressions to judge who looks more like "someone who's truly committed to their work."

An open-access paper published in September 2026 by Hellen Gießen and René Schmoll, Think Work – Think Office?, lands this question squarely on remote work and promotion judgments. The two authors don't simply compare actual promotion rates between remote and office employees. Instead, they first measure whether managers automatically associate "office" with work and "home" with leisure, then randomly vary how many days a week a hypothetical employee works from home, and observe whether managers' promotion recommendations shift accordingly.

I picked it not because it proves "remote work always costs you," but because it pulls apart three judgments that often get lumped together: fully remote and hybrid work are not the same thing; the stereotypes managers openly acknowledge and their automatic associations are not the same thing; and promotion recommendations in an experiment and promotion decisions in a real organization are not the same thing. For anyone who needs to build competence and chase opportunities inside an organization, this kind of disaggregation is more useful than a one-liner about "showing your face more."

The paper's core argument

The study recruited U.S. managers through Prolific in May 2025. Participants had to reside in the U.S., hold U.S. citizenship, be fluent in English, not be students, hold a management position, and have a prior study approval rate on the platform of at least 90%. The study received 423 complete questionnaires. After removing samples who were unemployed, failed attention checks or scenario identification checks, or had implausible work hours or age information, 306 managers remained. The sample averaged 41.51 years old with 17.88 years of work experience; they worked an average of 38.43 hours per week, about 11.93 of which were done from home.

The first step was measuring the association "think work, think office." The authors first developed a dedicated implicit association test through two pilot studies: participants had to quickly categorize "people who work from home" and "people who work in an office" alongside "work" and "leisure." If pairing "office" and "work" on the same key was noticeably faster, that association was more fluent. The main sample's average implicit association score was 0.59, which counts as moderate strength by the paper's effect-size standard; the average explicit questionnaire score was 5.71, also above the scale's neutral point of 4. In other words, among these managers, "office feels more like work, home feels more like leisure" showed up both in direct answers and in reaction speed.

But the explicit and implicit measures were not significantly correlated with each other. This matters. A person may be unwilling, unable, or unaware of holding a certain view on a questionnaire, yet the automatic association can still exist; conversely, directly expressed attitudes don't necessarily match rapid categorization responses. The paper didn't merge the two into a single "bias index" but analyzed them separately.

The second step was a randomized scenario experiment. Managers saw materials about a hypothetical employee. The study randomly assigned that employee to work from home 0, 2, or 5 days per week, and also randomly varied the employee's gender, then asked managers to give a promotion recommendation. The average recommendation scores for the three groups were approximately 7.91, 7.65, and 6.62. Fully remote employees scored significantly lower than both fully in-office and two-days-at-home employees; the difference between hybrid and fully in-office was not significant. The overall effect size for promotion differences explained by the three arrangements was about 0.10.

So the paper's safest conclusion isn't "you can't get promoted unless you sit in an office," but rather: in this controlled scenario, the penalty concentrated in the five-days-at-home condition, and hybrid arrangements did not replicate the same difference. The so-called flexibility stigma isn't a straight line uniformly increasing from zero to five days—it looks more like crossing some cognitive threshold once an employee is completely detached from the office.

The study also found that managers who themselves worked from home more had weaker explicit stereotypes; but this relationship didn't appear for implicit associations. Familiarity with remote work may make people more willing to correct the "home equals leisure" narrative in explicit judgments, but that doesn't mean the automatic association has disappeared. Since managers' own work arrangements weren't randomly assigned, this part can only be treated as correlation: people who already disagreed with remote-work stigma might also be more likely to choose to work from home.

The authors further tested whether stereotypes amplify the promotion disadvantage of fully remote employees. Neither the implicit nor explicit measure produced a significant overall interaction; only the specific comparison of "fully remote vs. fully in-office" showed significant moderation. The paper accordingly calls this partial support, rather than claiming all stereotypes reliably translate into promotion discrimination at all remote-work intensities. This conservative framing is worth preserving.

Points worth questioning

First, the experiment measured recommendations for hypothetical employees, not real promotions. Randomly assigning office days can support the causal claim that "a fully remote label in a scenario shifted average recommendations," but it can't directly extrapolate to the same magnitude of pay, promotion, or layoff consequences in real companies. Real managers also see long-term performance, job nature, team collaboration, and business results—information that could either weaken or amplify the labeling effect.

Second, the sample came from U.S. managers on Prolific who met screening criteria—not the U.S. manager population as a whole, and even less representative of Chinese companies, public institutions, or highly distributed software teams. The U.S. has its own context for remote-work prevalence, labor institutions, and management culture. The paper describes a judgment mechanism that can be triggered experimentally, not a cross-nationally universal penalty magnitude.

Third, the study retained 306 of 423 complete questionnaires—about a quarter were removed, including 67 who failed the scenario identification check. Strict cleaning can improve comprehension of the experimental materials, but such substantial sample attrition may also leave a final sample that's better at answering questionnaires and more careful at identifying scenarios. Readers shouldn't treat the 306 as a naturally occurring management population with no filtering.

Fourth, the implicit association test can only show associations in categorization speed—it doesn't directly equal real discriminatory behavior. The paper developed a new test specific to remote work and reported good internal consistency, but any new measure needs more samples and scenario replications. Especially when implicit and explicit measures are uncorrelated and the overall moderation term isn't significant, you can't interpret a moderate-strength average association as "managers' true inner thoughts have been read out."

Fifth, the experiment only offered three levels: 0, 2, and 5 days per week. It can't tell us what 1, 3, or 4 days would look like, nor can it distinguish fixed hybrid, on-demand office attendance, cross-city distributed teams, and temporary caregiving arrangements. What's actually at work might not be the location of home but visibility, synchronous collaboration, signals of compliance with organizational norms, or managers' expectations about communication costs.

Sixth, the study focused on recommenders' stereotypical associations but didn't directly test bias-reduction measures. Managers working remotely more themselves correlated with weaker explicit stereotypes, but that's not yet a randomized intervention. Whether standardized promotion evidence, clear role outputs, documented key collaborations, and calibrated evaluation scales across teams are more effective than "go to the office more" still requires new research.

Takeaways tied to recent interests

This paper adds an uncomfortable but necessary variable to the question of "how do people without connections carve out a future for themselves": competence isn't automatically seen by organizations. Organizational judgment depends on both results and a set of low-cost identification cues. Physical presence in the office is the cheapest, most familiar cue; it isn't necessarily accurate, but it can substitute for judging real contributions when evidence is thin.

For individuals, the conclusion isn't to mechanically increase desk time, but to avoid letting your role's value depend on your manager guessing. One possible mechanism behind the fully remote penalty is precisely that daily work processes, ad-hoc collaboration, and commitment are less visible. A transferable response is to turn contributions into verifiable evidence: what facts revealed the key problem, what proposals you made, what rework you reduced, which decision you pushed forward, who adopted the result. Showing your face only provides a "person is present" signal; evidence is what shows "things changed because of you."

For managers, this study's reminder isn't to pull all remote employees back to the office either. Hybrid and fully in-office showed no significant difference in the experiment—which itself argues against the crude narrative that "remote workers are all slacking." What really needs checking is whether promotion criteria have quietly swapped visibility for performance: whether who you bump into in the hallway more often, who replies to instant messages faster, who adapts better to your working style—these get treated as conscientiousness and leadership.

This also connects to recent forwards about "where power comes from." Managers hold not just task-assignment power but also the power to interpret ambiguous signals. When performance standards are unclear, who looks more like the ideal employee influences who gets opportunities; and what the "ideal employee" looks like is often shaped by managers' own experiences, team conventions, and visibility preferences. Reducing the arbitrariness of this power doesn't require managers to have no intuition—it requires important decisions to return to consistent, auditable evidence.

For hobbyists maintaining personal projects that need fragmented work, the paper offers another reminder: don't copy big-company presence competition. Personal projects have no promotion committee; the scarcest resources are continuous attention and real delivery. At that point, adding synchronous time just to look busy actually hurts results. But once you need to collaborate with external partners, clients, or organizations, proactively design visibility: short, clear progress records, decision rationale, reproducible results, and next-step ownership are more reliable than being online all day.

How to read it

Start with the abstract and results section, grabbing just three sets of numbers: 306 managers; average implicit association of 0.59; promotion recommendations of approximately 7.91, 7.65, and 6.62. Then immediately write the three office conditions next to those numbers, to avoid misreading "experimental penalty for fully remote" as "any work-from-home gets penalized."

Second, look at the study design and distinguish which variables were randomly assigned and which were merely measured. Employees working 0, 2, or 5 days from home is a randomized scenario, so you can discuss its experimental effect on recommendations; how much managers themselves work from home isn't random, so you can only say it correlates with explicit stereotypes.

Third, look at the full interaction tests. The paper's title makes you expect a tidy mechanism, but the overall "stereotype × remote intensity" interaction isn't significant—only the specific comparison of fully remote vs. fully in-office shows moderation. Reading this clearly is what prevents a mechanism story from overshadowing the statistical results.

Finally, run a small experiment on a performance, promotion, or collaboration judgment you're familiar with: strip out names, office-presence impressions, and response speed, keeping only goals, constraints, actions, results, and retrospectives—then ask again who you'd recommend. If the ranking changes, it means the original judgment was mixed with cues that aren't equivalent to output. Managers can use this to revise evaluation forms; individuals can use it to fill in their own evidence chain.

Open-access original: Gießen & Schmoll: Think Work – Think Office?

Paper PDF: Journal of Applied Social Psychology open-access PDF