Developing the E-MOI: A theory-driven multidimensional inventory for general and specific problematic technology use among adults

Technical Summary: Developing the E-MOI

Problem Statement
The rapid evolution of digital environments has outpaced the development of psychometric tools necessary to assess problematic technology use (PTU). Existing instruments suffer from several critical limitations: they often rely on outdated constructs (e.g., email usage) that lack ecological validity for contemporary platforms; they lack consensus on the operationalization of key constructs like "excessive use" versus "addiction"; and they predominantly utilize unidimensional or single-domain approaches (e.g., focusing solely on gaming or social media) rather than capturing the general severity and specific manifestations of PTU simultaneously. Furthermore, there is a scarcity of multidimensional instruments for adults that include multi-informant perspectives (observational reports) to mitigate self-report biases, and a general lack of standardized, population-based norms required for clinical interpretation of scores.

Methodology
The study aimed to develop and validate the Electronic Media Overusing Inventory (E-MOI) for adults, extending a previously validated battery designed for children and adolescents. The development followed a rigorous, multi-stage process involving an interdisciplinary team of experts (psychologists, psychiatrists, psychometricians, and addiction specialists).

• Item Generation: An initial pool of 327 items was generated based on Griffiths' components model of behavioral addiction and domains identified in prior youth research. This was refined through expert review and independent judge evaluation (content validity and clarity) to a pool of 187 items.
• Study Design: Validation was conducted across four sequential phases between October 2024 and August 2025 using a stratified random sample of the Polish adult population (N = 2,323, aged 18–71).
  • Study 1 (EFA): Used Sample 1 (n=1,016) to explore the latent structure and reduce the item pool.
  • Study 2 (CFA & Validity): Used Sample 2 (n=1,095) to confirm the factor structure, assess internal consistency, and evaluate convergent/divergent validity against established measures (e.g., PIUQ-9, BFAS, GAS, BSAS, PPUS, CPGI, PTUS, SWLS).
  • Study 3 (Reliability): Used Subsample 2a (n=319) for test-retest reliability assessment over a 3–4 week interval.
  • Study 4 (Observational Version): Used Subsample 2b (n=213) to validate the observational (informant-report) version (E-MOI D2) by having partners assess the target participant.
• Analytical Approach: Analyses included Exploratory Factor Analysis (EFA) with Promax rotation and Confirmatory Factor Analysis (CFA) using robust maximum likelihood estimation. Model fit was evaluated using RMSEA, SRMR, CFI, and TLI. Reliability was assessed via McDonald's omega and Cronbach's alpha. Norms (sten scores) were developed stratified by gender and age groups (18–34 and 35–65).

Key Contributions
The primary contribution is the E-MOI battery, a comprehensive, multidimensional instrument available in two complementary formats: a self-report version (D1) and an observational version (D2). The instrument is grounded in empirical evidence and contemporary theoretical frameworks, addressing the gap between general severity and specific domain manifestations.

The final instrument consists of 77 items organized into 11 subscales:

1. General Dimensions: E-media overuse, Virtualization, Insight into losses, Positive protective behaviors, and Signs of e-media addiction.
2. Domain-Specific Dimensions: Signs of social media addiction, digital gaming addiction, online shopping addiction, online pornography addiction, online gambling addiction, and dating platform addiction.

Results

• Factor Structure: The CFA supported a 11-factor structure. While CFI and TLI values indicated moderate fit (0.861 and 0.856, respectively), the RMSEA (0.046) indicated a close fit, and the SRMR (0.102) suggested some residual misfit, which the authors attribute to the complexity of the construct.
• Reliability: The instrument demonstrated high internal consistency. The total score showed excellent reliability (McDonald's ω = 0.976 for D1; α = 0.966 for D1 self-report; α = 0.969 for D2 observer-report). Ten of the 11 subscales showed very good reliability (ω > 0.80), with "Positive protective behaviors" showing satisfactory reliability (ω > 0.70).
• Temporal Stability: Test-retest correlations were strong (r = 0.920 for total score), indicating high stability over time.
• Validity:
  • Convergent: All problem-focused subscales correlated positively with corresponding established measures (e.g., the gaming subscale correlated r = 0.816 with the Game Addiction Scale).
  • Divergent: Problem-focused subscales correlated negatively with life satisfaction (SWLS), while the "Positive protective behaviors" subscale correlated positively.
  • Inter-rater: Significant agreement was found between self-report and observational versions, particularly for total scores and general overuse/addiction indicators.
• Sociodemographic Patterns: Results aligned with theoretical expectations: men reported higher levels of gaming, gambling, pornography, and dating platform addiction, while women reported higher social media addiction. Younger participants exhibited higher levels of maladaptive use, whereas older participants showed higher protective behaviors.

Significance and Claims
The authors claim that the E-MOI addresses critical methodological gaps in the assessment of PTU by providing a standardized, theoretically grounded tool that captures both general and specific manifestations of problematic use in adults. By offering both self-report and observational formats, the instrument enhances assessment validity, particularly for individuals who may underreport behaviors due to social desirability or lack of insight.

The inclusion of population-based norms allows for the meaningful interpretation of individual scores, distinguishing between low, average, and high levels of risk. The authors posit that the E-MOI can replace the need for multiple single-scale instruments, thereby facilitating more precise screening, diagnosis, and the planning of targeted preventive and therapeutic interventions. The study notes that the prevalence rates of high-risk behaviors identified by the E-MOI (e.g., ~10% for general internet addiction, ~16% for social media) are consistent with recent meta-analytic findings, supporting the instrument's external validity.

The authors acknowledge limitations, including the CFI/TLI values falling slightly below optimal thresholds and the exclusion of emerging domains like cryptocurrency trading and artificial intelligence use, which were deemed to have low prevalence at the time of data collection. They emphasize that the E-MOI is a step toward a more comprehensive framework but encourage future cross-cultural adaptations and expansions to cover evolving digital activities.