Brief Depression Psychoeducation and Depression-Related Beliefs Among Chinese Young Adults: The Roles of Self-Perceived Knowledge and Message Framing

Description

Brief psychoeducational materials are widely used in public-facing mental health promotion, but their effects may depend on who receives them and how they are worded. This study examined whether brief depression psychoeducation was associated with two depression-related beliefs—fatalism and recovery expectancy—and whether these associations depended on self-perceived depression knowledge and message framing. Chinese young adults aged 18–30 years were recruited online and assigned to one of four conditions: No-text control, Neutral text, Fog-framed text, or Shackles-framed text (total N = 868). Participants completed an information-choice task, an 8-item coping-action endorsement measure, and belief measures. Primary analyses compared Any-text versus No-text using robust linear models with age, gender, and work status as covariates. Text-only analyses compared Neutral, Fog, and Shackles conditions. The association between Any-text exposure and both belief outcomes depended on self-perceived knowledge. Relative to No-text, Any-text was associated with lower fatalism and higher recovery expectancy primarily among participants with higher perceived knowledge, whereas corresponding differences at lower perceived knowledge were small and non-significant. Within the text conditions, metaphor did not provide a general advantage over literal wording. Instead, Fog framing selectively altered how perceived knowledge related to recovery expectancy relative to Neutral wording, whereas other framing contrasts were not consistently significant. The two more applied indicators were less sensitive in a single exposure: coping-action endorsement showed no significant overall text effect or text-by-knowledge interaction, and information choice showed only limited inferential evidence. Overall, the findings suggest that brief depression psychoeducation does not operate uniformly across recipients. Its association with key depression-related beliefs depends in part on self-perceived knowledge, and metaphorical framing should be treated as a selective wording feature rather than a universal enhancement.

Steps to reproduce

We conducted an online, between-participants message experiment with four conditions: a No-text control and three brief psychoeducational text versions (Neutral, Shackles, Fog). To ensure strict sample independence and prevent cross-participation, the four condition-specific questionnaires were fielded sequentially, which allowed us to apply the platform’s built-in exclusion filters to block repeat participation across conditions. We assessed cross-condition balance on observed covariates and found no substantial imbalances (Table S1). Participants reported demographics (gender, age, work status) and completed a single item assessing self-perceived depression knowledge and then completed one of four condition-specific survey versions corresponding to the experimental conditions. In the three text conditions, participants read a brief psychoeducational text about depression on a separate page. The No-text control condition received no text. Participants then completed outcomes in a fixed order intended to reduce immediate belief-question priming of the applied tasks18: (1) information choice, (2) coping-action endorsement, (3) beliefs about depression (fatalism and recovery expectancy), and (4) text-perception items (text conditions only). Analyses were conducted in R (v4.5.2). Models adjusted for age (years), gender (Female = 1, Male = 0; “Other/prefer not to say” coded missing for binary coding), and work status (Working = 1, Student = 0; “Other” coded missing). Primary models tested Any-text vs No-text associations and moderation by perceived knowledge (z). Ordinary least squares models included Any-text (coded 1 for Neutral/Fog/Shackles, 0 for No-text), perceived knowledge (z), their interaction, and covariates; inference used HC3 heteroskedasticity-consistent robust standard errors with 95% confidence intervals. Information choice was analysed using multinomial logistic regression (reference = biomedical), with condition indicators, perceived knowledge (z), and covariates; inference used sandwich robust standard errors with 95% confidence intervals. Secondary analyses restricted to the three text conditions compared Neutral/Fog/Shackles and their interactions with perceived knowledge using the same covariate set and robust SEs. Models used complete cases for binary-coded covariates, matching the Ns reported in the main regression tables.

Institutions

• Jilin Agricultural University

  Jilin, Changchun

Categories

Licence