Audience Heterogeneity in Responses to Brief Depression Psychoeducational Messages: Self-Rated Understanding and Fatalistic Beliefs Among Chinese Young Adults

Description

This dataset examines whether the relationship between self-rated understanding of depression and depression-related beliefs differed between questionnaires with and without a brief passage, and whether Fog and Shackles framing produced frame-congruent responses. The final analyses were not preregistered. A formative pretest was conducted with 74 adults recruited by convenience sampling. After Fog and Shackles had been introduced as depictions of depression, respondents rated eight coping actions and four candidate titles from 1 (more Shackles-like) to 7 (more Fog-like), with 4 indicating no clear leaning. The ratings informed material selection and represent prompted semantic mapping, not spontaneous metaphor associations. The main study included 868 Chinese adults aged 18–30 recruited through Credamo, a Chinese online respondent platform. Participants completed separately released No Message (n = 238), Literal (n = 209), Fog (n = 223), or Shackles (n = 198) questionnaires. The three Message versions contained length-matched passages about depression; No Message omitted the passage. Respondents were not randomly assigned. Self-rated understanding was recorded before any passage. Later measures covered fatalism, perceived treatability, title choice, eight coping actions, and post-message checks. The Message × understanding interaction for fatalism was negative (b = −0.189, 95% CI [−0.328, −0.051], Holm-adjusted p = .015). Understanding was positively associated with fatalism in No Message but essentially unrelated to fatalism across the pooled Message versions. Item analyses located the pattern mainly in pessimism about becoming completely free of depression. Perceived treatability showed a positive interaction (b = 0.173, Holm-adjusted p = .019). Among Fog and Shackles recipients choosing between the corresponding titles, Fog recipients had higher odds of choosing the Fog-aligned title (OR = 1.78, Holm-adjusted p = .050). Coping balance did not differ. The data can reproduce the reported analyses and support further work on audience heterogeneity, depression-related beliefs, title choice, and coping evaluations. Version comparisons are conditional associations, not causal effects: questionnaires were released mostly at different times, beliefs were measured only after any passage, and No Message also differed in reading activity and attention. Self-rated understanding is subjective, not an objective knowledge test. No diagnosis, symptom inventory, or treatment history was collected, so the data do not characterise a clinical sample.

Steps to reproduce

All questionnaires were administered in Chinese. The formative pretest was conducted with 74 consenting convenience respondents recruited through the researchers' networks. Respondents were introduced to Fog and Shackles as two depictions of depression and rated eight coping actions and four candidate titles from 1 (more Shackles-like) to 7 (more Fog-like), with 4 indicating no clear leaning. Means were compared with 4 using two-sided one-sample t tests. Holm adjustment was applied separately to the actions and titles. The pretest served as formative semantic mapping for stimulus selection. The main study was conducted through Credamo, a Chinese online survey and respondent-recruitment platform. Adults aged 18–30 completed one of four separately released questionnaires: No Message, Literal, Fog, or Shackles. The three Message versions contained length-matched descriptive passages about depression; No Message omitted the passage. Respondents were not individually randomised because the questionnaires were released in separate batches. Documented eligibility and duration criteria were applied to the exported records. When a platform account appeared in more than one version, the earliest eligible response was retained. This removed 27 later responses from people who completed both Fog and Shackles. One age-ineligible response was also excluded. The final sample comprised 868 respondents: No Message, n = 238; Literal, n = 209; Fog, n = 223; and Shackles, n = 198. The retained first responses from the 27 cross-version respondents are marked by n_conditions_completed = 2, enabling the remove-both sensitivity analysis to be reproduced from the public file. Direct identifiers and exact response times were removed; start time is represented in ten-minute bins. The public datasets are main_study_analysis_n868.csv and formative_pretest_analysis_n74.csv. Download and extract Depression_Message_Public_Reproducibility.zip, open Depression_Message_Reanalysis.Rproj in RStudio, and run: source("R/install_packages.R") source("run_public_analysis.R") The first command installs any missing required packages. The second validates the datasets and runs all measurement, pretest, primary, supporting, exploratory, frame-congruence, sensitivity, table, and figure scripts in sequence. Numerical outputs are written to results/tables/, and the main tables and figure to results/publication/. The final script verifies the locked sample sizes and principal estimates and stops if they are not reproduced. A successful run ends with “Public analysis completed successfully.” The reported analyses used R 4.6.0.

Institutions

• Jilin Agricultural University

  Jilin, Changchun

Categories

Licence