The law of initial values: a four factor theory.

Area of Science:

• Psychometrics
• Statistical modeling
• Behavioral research methodology

Background:

• The Law of Initial Values (LIV) describes a negative correlation between an initial state and subsequent change.
• Understanding factors influencing this correlation is crucial for accurate data interpretation.
• Previous research has not comprehensively simulated the interplay of various statistical artifacts.

Purpose of the Study:

• To investigate the impact of measurement error, skewness, floor/ceiling effects, and reactivity on the correlation between initial level and change.
• To identify factors that produce negative (LIV) and positive (anti-LIV) correlations.
• To discuss methods for distinguishing between operating factors in empirical data.

Main Methods:

• Computer simulations were employed to model four key factors: measurement error, skewness, floor/ceiling effects, and reactivity.
• A total of 24 distinct conditions were simulated, varying these factors and the direction/magnitude of change.
• Random samples of 50 cases were generated using SPSSX for each condition.

Main Results:

• Measurement error, skewness, and floor/ceiling effects were found to produce negative correlations, supporting the LIV.
• Reactivity and skewness were identified as factors yielding positive correlations (anti-LIV effects).
• Skewness demonstrated the capacity to produce both negative and positive correlations depending on simulation parameters.

Conclusions:

• Statistical artifacts like measurement error, skewness, and floor/ceiling effects can significantly distort the observed relationship between initial levels and change.
• Reactivity introduces an anti-LIV effect, potentially masking or reversing true initial value relationships.
• Distinguishing the underlying cause of observed correlations is essential for valid scientific conclusions.