TL;DR
This study validates the Bergen Social Media Addiction Scale (BSMAS) as a consistent and reliable tool for comparing social media addiction across genders. By confirming "measurement invariance" in a sample of 1,120 university students, the researchers demonstrate that any observed differences in addiction scores between men and women reflect genuine behavioral trends rather than flaws in the test itself.
What Problem Does This Paper Address?
In psychological research, comparing two groups (like men and women) isn't as simple as comparing their average test scores. You first have to prove that the test "means" the same thing to both groups. This is a concept called measurement invariance.
If a scale lacks measurement invariance, a higher score in one group might be due to "measurement bias"—perhaps the wording of a question resonates differently with women than with men. While the Bergen Social Media Addiction Scale (BSMAS) is one of the most widely used tools in the field, previous research on gender differences in social media addiction has been contradictory. Some studies suggested women are more prone to addiction, others pointed to men, and some found no difference at all. The authors of this paper set out to determine if these discrepancies were real or simply artifacts of a biased measurement tool.
Methods
The researchers employed Multi-Group Confirmatory Factor Analysis (CFA) using the R programming language (specifically the lavaan and semTools packages). This is a rigorous statistical procedure that tests for invariance across four increasingly strict levels:
- Configural Invariance: Does the basic structure (one single factor of "addiction") hold for both groups?
- Metric Invariance: Do the individual items on the scale contribute to the overall score with the same "weight" or strength for both men and women?
- Scalar Invariance: Do men and women with the same level of actual addiction give the same answers to the questions (i.e., are the "starting points" or intercepts equal)?
- Strict Invariance: Is the amount of "noise" or measurement error consistent across both groups?
The authors used the Weighted Least Squares with Mean and Variance Adjusted (WLSMV) estimator, which is specifically designed for ordinal data like the 5-point Likert scales used in the BSMAS.
Datasets
The study utilized a primary dataset collected from 1,120 participants recruited from five universities in Inner Mongolia, China.
- Composition: 513 men and 607 women.
- Age Range: 18 to 26 years (Mean = 20.89).
- Criteria: Participants had to be at least 18 years old and possess at least one social media account (e.g., WeChat, QQ, Weibo).
Key Insights and Ideas
- The "Apples to Apples" Confirmation: The most significant insight is that the BSMAS passed all four levels of invariance. This means researchers can finally be confident that gender comparisons using this scale are psychometrically sound.
- Universal Structure: The study confirms that the "six core components" of addiction (salience, mood modification, tolerance, withdrawal, conflict, and relapse) manifest identically in both male and female university students in mainland China.
- Robustness of the BSMAS: Despite being a short, 6-item scale, the BSMAS showed high internal consistency (Cronbach’s alpha = 0.822), proving it is a "psychometrically robust" instrument even in different cultural contexts.
Key Results
The statistical "fit" of the model was excellent across the board, exceeding standard benchmarks for psychological research:
- Model Fit: The Comparative Fit Index (CFI) was 0.991 and the Root Mean Square Error of Approximation (RMSEA) was 0.068 for the baseline model.
- Invariance Testing: When moving from one level of invariance to the next, the change in CFI ($\Delta$CFI) was consistently $\le$ 0.002, well below the recommended cutoff of 0.01. This indicates that adding constraints (forcing the model to be equal across genders) did not significantly degrade the model's accuracy.
- Factor Loadings: Standardized factor loadings for all items ranged from 0.629 to 0.735, indicating that every item on the scale is a strong indicator of the underlying addiction construct.
Limitations and Future Work
While the study provides a strong foundation for gender-based research, the authors acknowledge several limitations:
- Sample Specificity: The data comes exclusively from university students in one region of China. The results may not generalize to other age groups (like the elderly or young children) or different cultural backgrounds.
- Self-Reporting Bias: Like all studies relying on questionnaires, the data is subject to "social desirability bias," where participants might underreport their social media use.
- Cross-Sectional Design: The study captures a snapshot in time. The authors suggest that future research should look at longitudinal invariance—checking if the scale remains consistent for the same individuals as they age or as social media technology evolves.
Summary updated