Data Insights on Postpartum Depression Among Women in Niger State, Nigeria
Description
This dataset provides comprehensive insights into the influence of information sources and social factors on postpartum depression (PPD) awareness and help-seeking behavior among women in Northern Nigeria. Collected through a cross-sectional survey of 384 women aged 18-49 in Niger State, the data captures demographic profiles, preferred communication channels (interpersonal, electronic, and print media), and their impact on PPD knowledge and health-seeking actions. Key features of the dataset include: Demographic details: Age, education, employment, marital status, religion, and parity, highlighting disparities in access to information and support. Measurement models: Confirmatory Factor Analysis (CFA) and Structural Equation Modelling (PLS-SEM) results validate the reliability and validity of constructs like PPD knowledge and help-seeking behaviour. Path analysis: Reveals significant relationships, such as interpersonal and electronic media's strong influence on PPD awareness (β = 0.525–0.551), compared to print media's marginal impact (β = 0.200–0.225). Methodological rigor: Multistage stratified sampling ensured urban-rural representation, while SmartPLS-SEM and bootstrapping tested theoretical frameworks like the Theory of Planned Behaviour. The dataset, hosted on the Mendeley Data repository, supports the development of customised maternal mental health interventions, policy design, and cross-cultural research in low-resource settings. Its granularity enables replication, secondary analysis, and comparative studies on health communication efficacy. Keywords: Postpartum depression, help-seeking behaviour, information sources, SmartPLS-SEM, Northern Nigeria.
Files
Steps to reproduce
Steps to Reproduce the Results To replicate the findings of this study, follow these structured steps based on the methodology outlined in the document: 1. Data Access & Preparation: Download the raw dataset from Mendeley Data: a. Clean and preprocess the data: b. Verify completeness (N = 384 responses). c. Check for missing values and outliers (none reported in the study). d. Code categorical variables (e.g., education, employment status) as per Table 1. 2. Software Setup a. Use SmartPLS-SEM (v4.x) for structural equation modelling (PLS-SEM). Alternative: R (with `plspm` or `semPLS` packages) or Python (PyPLS). 3. Measurement Model Validation a. Confirmatory Factor Analysis (CFA): b. Load observed variables (e.g., "Healthcare Officials," "Radio") onto latent constructs (Table 2). c. Ensure factor loadings >0.7 (Table 2) and AVE > 0.5 (Table 3). d. Assess composite reliability (CR > 0.8) and Cronbach’s alpha (>0.7) (Table 3). e. Discriminant validity: Confirm square root of AVE > inter-construct correlations (Table 4). 4. Structural Model Analysis a. Test hypothesized paths (Table 6) using PLS-SEM: Key relationships: Interpersonal sources → PPD knowledge (β = 0.551, p < 0.001). Electronic media → Help-seeking behaviour (β = 0.394, p = 0.007). Non-significant paths: Print media effects (p > 0.05). Model fit: Report R² values (e.g., 0.276 for IIS → HSB; Table 6). b. Calculate effect size (f²) and predictive relevance (Q²) (Table 5). 5. Bootstrapping for Significance a. Run 5,000 bootstrap samples (as implied by the methodology). b. Extract t-statistics and p-values for path coefficients (Table 6). 6. Reporting a. Tabulate results to mirror Tables 2–6. b. Visualize the PLS-SEM model (Figure 1) using SmartPLS or equivalent tools. 7. Sensitivity Checks (Optional) a. Stratify analysis by demographics (e.g., urban vs. rural) if subgroup comparisons are needed. Ensure ethical compliance (no respondent identifiers in raw data) and cite the original dataset DOI. For exact variable labels, refer to the questionnaire items in the repository.
Institutions
- Federal University of Technology Minna
- Covenant University