Anastasia Galkina

Experience Sampling Methodology and Its Methodological Advances

The work presented in this thesis aims at exploring, applying, and expanding the methodological foundations of Experience Sampling Method (ESM) research through systematic measurement development, advanced analytical techniques, and innovative solutions for common distributional challenges. ESM, also known as Ecological Momentary Assessment (EMA), has gained increasing prominence in psychological research by enabling the collection of real-time data on variables of interest as they naturally occur in participants’ daily lives. However, the methodology presents unique analytical challenges involving complex nested data structures, measurement development needs and a lack of common practice, and data characteristics that necessitate specialized approaches to ensure accurate parameter estimation and valid conclusions.

Following an introduction to the ESM methodology and its challenges, the thesis is divided into three main parts. Part I describes the systematic bottom-up development of Honesty-Humility measures across trait, motivation, and state levels. Part II explores interpersonal emotion dynamics in romantic relationships using advanced dyadic ESM approaches. Part III addresses fundamental methodological issues in ESM data analysis, particularly the problem of censored distributions and bounded psychological constructs.

Systematic Measurement Development

Chapters 2, 3, and 4 form an integrated sequence focused on developing and validating measures of Honesty-Humility suitable for both cross-sectional trait assessment and intensive longitudinal research. These chapters demonstrate a systematic, bottom-up approach that minimizes researcher bias while ensuring ecological validity and psychometric rigor.

Chapter 2 presents the development of the Adjective Checklist of Honesty (ACH), a comprehensive 22-item adjective-based measure capturing the four main facets of Honesty-Humility: Sincerity, Fairness, Modesty, and Greed-Avoidance. Using an extensive corpus of Italian adjectives refined through independent expert raters, this systematic approach yielded important theoretical insights, particularly that Truthfulness and Sincerity are conceptually indistinguishable in natural language use. The scale demonstrated good factorial validity and showed meaningful associations with theoretically relevant constructs, while revealing interesting overlap with conscientiousness in workplace ethics domains.

Chapter 3 extends this approach to uncover the motivational core of Honesty-Humility, developing the Goals for Honesty and Goals for Dishonesty (GH/GD) questionnaire. Through participant-generated goal elicitation from over 9,000 textual responses, classified by experts with the help of natural language processing algorithms, researchers identified 48 goal classes potentially related to Honesty-Humility. The final 78-item questionnaire revealed that Goals for Honesty (GH) and Goals for Dishonesty (GD) represent two distinct, only moderately negatively correlated motivational orientations. Critically, GD consistently predicted actual dishonest behavior in an incentivized cheating task beyond traditional trait measures, while GH related more strongly to authentic living and self-concept integrity.

Chapter 4 completes this measurement sequence by adapting the GH/GD questionnaire for intensive longitudinal research (GH/GD-EMA). In a 15-day ESM/EMA study involving 198 participants who received five daily prompts, multilevel confirmatory factor analyses confirmed the two-factor structure at both the within- and between-person levels. Most importantly, momentary goal importance robustly predicted Honesty-Humility states both contemporaneously and temporally. Using latent growth curve modeling, the study demonstrated that changes in goals during the ESM period predicted trait-level change from baseline to follow-up through their effects on state changes, providing evidence for bottom-up personality development processes.

Interpersonal Emotion Dynamics

Chapter 5 shifts focus to interpersonal processes, examining emotional interdependence in romantic relationships through an extended Longitudinal Actor-Partner Interdependence Model (L-APIM) framework. Using intensive longitudinal data from 76 couples completing assessments five times daily over 30 days, the study investigated both contemporaneous and temporal patterns of emotional interdependence between partners while controlling for autoregressive effects and cross-valence actor effects.

Results revealed significant emotional interdependence for positive affect both contemporaneously and temporally, indicating that partners’ positive emotions show a modest but consistent linkage in daily life. However, no significant partner effects emerged for negative affect, suggesting that emotional transmission patterns might differ by valence. Contact between partners showed a positive main effect on males’ positive affect, but did not significantly moderate the strength of emotional interdependence. Substantial between-couple variability in emotional baselines was observed, with partners showing greater similarity in negative affect than positive affect baselines. These findings highlight that emotional interdependence is more complex and heterogeneous than theoretical models often assume, requiring more nuanced, couple-specific approaches.

Methodological Advances

Chapter 6 addresses a fundamental challenge in ESM research: analyzing skewed data that exhibits floor or ceiling effects. Many variables of interest in ESM studies, particularly negative affect and psychopathological symptoms, show highly skewed distributions that can violate assumptions of commonly used statistical models, like linear mixed-effects models, and lead to biased parameter estimates.

The study proposes conceptualizing skewed observations as censored manifestations of underlying latent variables and demonstrates that mixed-effects tobit models provide superior parameter recovery compared to standard linear mixed-effects models. Through artificial censoring of initially normally distributed positive affect data, the results showed that tobit models maintained stable parameter estimates under progressive censoring conditions, whereas standard models exhibited deteriorating slope estimates that approached zero. When applied to naturally skewed negative affect data, tobit models consistently estimated larger effect sizes, greater between-person variability, and more plausible intercept-slope correlations across all participant groups. These findings suggest that standard approaches may introduce systematic bias in the relationship strengths and individual differences in bounded psychological constructs.

New Horizons

Chapter 7 discusses the main findings and their implications for the future of ESM research. The thesis demonstrates that assessing psychological variability through intensive longitudinal methods is crucial to understanding the dynamic processes that occur in daily life. The systematic measurement development procedures, advanced multilevel modeling techniques, and innovative analytical solutions provide a foundation for more accurate, nuanced, and replicable ESM research. Future developments will likely focus on multimodal data integration, personalized assessment approaches, and continued refinement of analytical techniques for increasingly complex psychological phenomena.