Graduating students are often advised to research more careers, attend more networking events and keep every possible option open. Yet for someone who has spent months revisiting the same doubts, more exploration can become part of the problem. A randomized trial involving 486 final-year university students suggests that a different intervention may help: writing in concrete detail about the person they hope to become at work.
Published on 7 October 2026 in Frontiers in Psychology, the study found that four weekly writing sessions, each lasting approximately 20 minutes, reduced a particular kind of repetitive career thinking called ruminative career exploration. The benefit was detectable against both a group writing about unrelated future plans and a group waiting to receive the intervention.
Compared with the waitlist, the standardized effect on career rumination was d = 0.41 (95% confidence interval 0.22 to 0.61). The intervention also reduced difficulties related to feeling unprepared to make a career decision, with an effect of d = 0.43 (95% confidence interval 0.25 to 0.62). These were modest to moderate psychological changes, not proof that students secured better jobs. In fact, none of the measured employment outcomes remained statistically significant after correction for multiple tests.
The distinction matters for universities, careers advisers and students. The trial points towards a possible way of interrupting unproductive thought, but it does not establish a shortcut to employment or a substitute for practical career information. In a different university context, research on students’ perceptions of learning in Sweden likewise illustrates why measured outcomes must be distinguished from subjective experience.
Why more career exploration is not always better
Career exploration sounds unambiguously useful. Students can compare occupations, investigate qualifications, speak to employers and evaluate their own interests. Such activities may make an uncertain decision more informed. But the researchers, Shang Sun and Di Yang, argued that the quantity of exploration is not the same as its quality.
They distinguished reflective exploration, which involves gathering and evaluating useful information, from ruminative exploration, which involves repeatedly questioning possible futures without reaching a workable conclusion. A student might productively compare the requirements of two jobs, for example, or might spend hours cycling through the fear of choosing the wrong one without learning anything new.
The second pattern can look like diligence from the outside. It involves sustained attention to careers, yet the thinking may be repetitive rather than progressive. The study therefore tested whether making a future working identity clearer would reduce unproductive rumination, rather than simply increase the amount of career exploration.
The underlying concept is a future work self: a vivid and accessible mental picture of the person someone hopes to become in their professional life. It differs from an isolated goal such as earning a particular salary. It concerns what everyday work would look like, what capabilities a person would use and how present actions connect with that future identity.
This research adds to a wider discussion about how people form vocational identities. A separate Research Today report on 529 police cadets examined associations between emotional intelligence and a sense of career calling. The present study asks a different question: whether a brief, randomized writing exercise can change how students think about an approaching career decision.
How the researchers tested the intervention
The team recruited final-year undergraduate and postgraduate students from four universities in eastern China, including two nationally prominent universities and two regional institutions. A total of 612 students were screened; 486 met the requirements and were randomly allocated in equal groups of 162.
The trial used three conditions. In the first, students completed four weekly writing exercises about their future professional identity. The opening session asked them to imagine their working life several years ahead and describe the future self concretely. Subsequent sessions covered a typical working day, the capabilities and relationships associated with that identity, and one practical step connecting their present and future selves.
The second group was an active control. Its members also wrote for approximately 20 minutes each week, but about general future plans unrelated to work, including a journey, a living arrangement, a learning project and a community activity. This comparison was important because it helped distinguish the content of the career exercise from the effects of setting aside time to write and reflect.
The third group was placed on a waitlist and completed the assessments without taking part in the writing programme during the study. Its members were offered the career-writing programme after the final follow-up.
Participants were assessed at four points: before the intervention, immediately after the four-week programme, one month later and three months later. The researchers registered the trial in advance under ChiCTR2500074449. The allocation sequence was generated independently, and analysts were initially blinded to the group labels, although students could not be unaware of the writing tasks they were completing.
The average participant age was approximately 22 years, and roughly six in ten participants were women. The groups were similar at baseline on the measured characteristics. The psychological outcomes were assessed using questionnaires about future work identity, self-exploration, exploration of the occupational environment, ruminative career exploration and three forms of career decision-making difficulty.
Those three difficulty measures were deliberately kept separate. Lack of readiness included low motivation, general indecisiveness and unhelpful beliefs about decision-making. Lack of information captured gaps in knowledge about oneself, occupations and the decision process. Inconsistent information covered conflicting or unreliable information and competing pressures.
For the main analysis, the researchers used longitudinal growth models to compare how these outcomes changed across the four assessments. The models included all 486 randomized participants under an intention-to-treat approach, using a missing-data method rather than restricting the analysis to those who completed every survey.
What changed after four weeks of writing?
The clearest result concerned ruminative career exploration. Relative to the waitlist, the future-work-self group showed a model-estimated difference of 0.34 points lower on a five-point rumination scale at the final assessment, equivalent to d = 0.41 (95% CI 0.22 to 0.61; p < 0.001). Compared with the active writing control, the corresponding difference was 0.29 points, or d = 0.35 (95% CI 0.16 to 0.55; p < 0.001).
The active control makes the finding more informative than a simple comparison with students who did nothing. Both writing groups devoted similar time to structured reflection. Only one was prompted to construct a detailed picture of a future working identity. The difference between those groups suggests that the occupational focus of the exercise mattered.
The study did not find convincing increases in the two more constructive forms of career exploration. Compared with the waitlist, the effect on self-exploration was d = 0.10 (95% CI -0.10 to 0.30; p = 0.31), and the effect on exploring the occupational environment was d = 0.13 (95% CI -0.05 to 0.32; p = 0.14).
Additional equivalence tests supported the narrower conclusion that any changes in these reflective exploration measures were smaller than the researchers’ prespecified practical threshold of d = 0.30. This does not mean the exercise stopped students from seeking information. It means that the trial detected a change in repetitive doubts without detecting a comparably meaningful increase in exploration activity as measured by those scales.
One implication is that a clearer career identity may work more like a filter than an accelerator. It may help a student decide which questions are worth pursuing and which doubts no longer require another round of abstract reconsideration. That interpretation fits the observed pattern, but the trial did not directly observe students’ thought processes outside the questionnaires.
Decision readiness improved, but information gaps remained
The three decision-difficulty measures did not move in the same way. Against the waitlist, the future-work-self exercise reduced lack of readiness by d = 0.43 (95% CI 0.25 to 0.62; p < 0.001). Against the active control, the effect was d = 0.38 (95% CI 0.19 to 0.57; p < 0.001).
It also reduced difficulties related to inconsistent information. The standardized effect was d = 0.35 (95% CI 0.16 to 0.54; p = 0.001) relative to the waitlist, with a smaller but still detectable contrast against the active control.
By contrast, the effect on lack of information was small and statistically inconclusive: d = 0.11 (95% CI -0.08 to 0.30; p = 0.24) against the waitlist, and d = 0.09 against the active control. A writing exercise cannot be assumed to supply missing knowledge about employers, qualifications, vacancies or labour-market conditions.
That difference is practically important. Someone who already knows the main options but feels trapped in a cycle of indecision may face a different problem from someone who genuinely lacks information about how to enter a profession. The first person might benefit from a structured identity exercise; the second may still need interviews, occupational research, internships or expert guidance.
The authors caution that the effects across these three dimensions were estimated separately. They did not formally test whether one dimension improved statistically more than another. The reported pattern therefore supports a targeted interpretation, not a definitive ranking of which type of difficulty responds best.
Did less rumination explain the improvement?
The researchers went beyond asking whether the intervention worked. They tested a statistical pathway linking the writing exercise to changes in career readiness through three kinds of exploration: rumination, self-exploration and environmental exploration.
The model estimated an indirect effect of -0.18 through reduced ruminative exploration, with a 95% bootstrap confidence interval from -0.29 to -0.09. The comparable indirect effects through self-exploration (-0.02) and environmental exploration (-0.03) were small, and their confidence intervals included zero.
Within this fitted model, the ruminative pathway represented approximately 35% of the total estimated effect on lack of readiness. The remaining direct effect was -0.28, meaning much of the observed difference was not accounted for by the three measured exploration pathways.
There is an essential distinction between these results and a demonstrated mechanism. Students were randomly assigned to writing conditions, so the comparison of those conditions can support causal claims about the intervention as a whole. The degree of rumination was measured, not randomly manipulated. Unmeasured factors could therefore influence both changes in rumination and subsequent readiness. The mediation analysis is consistent with the researchers’ explanation, but it cannot prove that reducing rumination caused the improvement.
Did students actually get better jobs?
The trial also measured four exploratory employment indicators three months after the intervention: whether students had signed an employment agreement, the number of offers received, satisfaction with employment quality and the perceived match between work and field of study.
Some descriptive differences favoured the future-work-self group. For example, 54.3% reported having signed an employment agreement, compared with 47.1% in the active control and 45.8% in the waitlist group. However, the overall test for this indicator was not statistically significant (p = 0.20). The average number of offers was 1.74, 1.58 and 1.51 respectively, again without a statistically significant group difference (p = 0.16).
The future-work-self group reported somewhat higher satisfaction with employment quality, with an effect of d = 0.19 and p = 0.06. Perceived match with field of study initially showed a small nominal difference (d = 0.24; p = 0.03), but that result did not survive correction for testing four employment outcomes. After correction, the p values for employment agreement status, offer counts, satisfaction and field match were 0.80, 0.64, 0.24 and 0.12 respectively.
The responsible conclusion is therefore limited: the study found improvements in certain psychological measures, but no reliable evidence of improved employment outcomes during the three-month observation period. The exercise should not be marketed as a way to increase job offers or guarantee a better career match.
How reliable are the findings?
Several design choices strengthen the trial. Random allocation reduces the risk that pre-existing differences between students explain the outcomes. The active control tests whether career-specific writing offers anything beyond writing in general. Prospective registration makes the planned hypotheses more transparent, and four measurement occasions provide more information than a single before-and-after comparison.
Retention was also reasonably high. Of the 486 randomized participants, 457 (94.0%) provided post-intervention data, 436 (89.7%) were assessed at one month and 415 (85.4%) at three months. Attrition was not detectably different between the groups in the reported models. Nevertheless, missing responses could still introduce bias if students who dropped out differed in ways the analysis could not capture.
The main weaknesses concern measurement and generalizability. Every psychological outcome relied on self-report, and students knew the type of exercise they were completing. Their answers might partly reflect expectations or a desire to report progress. The measure of ruminative career exploration was adapted for this trial and has not yet been independently validated in a separate sample.
All four universities were in eastern China, and the country’s graduate recruitment calendar creates concentrated periods of pressure around job applications. Students in countries with different recruitment practices, different career-service systems or different labour markets may not respond in the same way. The results should not automatically be extended to working adults considering a mid-career change.
There are statistical qualifications too. The analysis did not explicitly model students nested within universities using a university-level random effect, which could make uncertainty estimates too narrow if students at the same institution were unusually similar. The main growth models assumed a linear change across time, although much of the observed improvement appeared soon after the writing programme and then levelled off. The researchers did not formally test whether the initial gains were maintained unchanged between the immediate and three-month assessments.
Finally, three months is a short period for judging lasting career decisions. A reduction in repetitive thoughts at graduation is valuable if replicated, but it does not establish that the effect endures for years or translates into improved wellbeing, earnings or job stability.
What students and universities can take from the research
The study suggests a potentially useful distinction for career support: students who need more information are not necessarily the same students who need help escaping repetitive doubt. Advisers may want to ask whether a student is still learning something new from their research, or simply revisiting the same feared outcomes.
The tested intervention was specific. It did not consist of vague positive affirmations. Students were asked to construct an occupational identity across four sessions: picture a future working self, describe a normal day, identify the capabilities and relationships that self would have, and identify a realistic next step from the present. The full programme took approximately 80 minutes spread over four weeks.
That format could be incorporated into university career-development courses or tested in digital guidance programmes. It should be treated as a candidate intervention rather than a proven universal service. Independent replication, external validation of the rumination scale, longer follow-up and comparisons across countries are needed before institutions can know whom it helps most and whether its benefits justify implementation costs.
For individual readers, the study offers a question rather than a prescription: when considering the future, is the next hour likely to produce a new fact, a clearer priority or a concrete action, or merely another round of the same uncertainty? The trial indicates that making a future working identity more concrete may reduce the latter. It does not remove the need for practical research, informed advice and realistic opportunities.
The broader lesson is that effective career preparation may depend not only on how much people think about work, but on what their thinking enables them to do next.
Source Information
Study: Sun, S., & Yang, D. (2026). “Effects of a future work self writing intervention on ruminative career exploration and career decision-making difficulties among final-year university students: a three-arm randomized controlled trial.”
Journal: Frontiers in Psychology, volume 17, article 1965900.
Published: 7 October 2026.
DOI: 10.3389/fpsyg.2026.1965900.
Study design: Prospectively registered, three-arm randomized controlled trial of 486 final-year undergraduate and postgraduate students at four universities in eastern China, comparing four weekly future-work-self writing sessions with general future-planning writing and a waitlist. Outcomes were assessed at baseline, after four weeks, and at one- and three-month follow-ups using longitudinal growth models and exploratory mediation analysis. Trial registration: ChiCTR2500074449.
