The Spurious Prospective Associations Model (SPAM): Explaining longitudinal associations due to statistical artifacts

Area of Science:

• Psychological science
• Statistical modeling
• Longitudinal data analysis

Background:

• Longitudinal data analysis frequently employs models susceptible to statistical artifacts.
• Previous research indicated that some prospective associations arise from measurement error and regression to the mean, not true effects.
• Existing models may misinterpret these artifacts as genuine temporal relationships.

Purpose of the Study:

• To formalize the analysis of statistical artifacts in longitudinal data.
• To introduce the Spurious Prospective Associations Model (SPAM) as an alternative.
• To demonstrate SPAM's superiority over adjusted cross-lagged effects models in explaining observed associations.

Main Methods:

• Formalization of statistical artifact analysis.
• Introduction and application of the Spurious Prospective Associations Model (SPAM).
• Comparison of SPAM with adjusted cross-lagged effects models using existing and new datasets.

Main Results:

• SPAM effectively explains prospective associations without assuming true changes in constructs over time.
• SPAM outperforms adjusted cross-lagged effects models in accounting for observed associations.
• SPAM accommodates paradoxical findings of simultaneous increasing and decreasing effects, which challenge other models.

Conclusions:

• The Spurious Prospective Associations Model (SPAM) provides a robust explanation for observed longitudinal associations.
• SPAM is better supported by data than competing models in the investigated cases.
• The findings highlight the importance of accounting for statistical artifacts in longitudinal research.