Mother-Daughter Attachment and Family Cohesion

Abstract

Summary: This contribution investigates the psychometric properties of two scales for measuring family cohesion and mother-daughter attachment. The scales were administered to 206 adult female subjects on three occasions of measurement 9 months apart. Various single-construct latent state-trait models with and without method factors were tested against the data and compared to each other. The most parsimonious version of a latent state-trait model with method factors was found to fit the data best for both constructs. The parameter estimates of this model reveal (1) that both scales are very reliable, (2) that they measure stable traits, (3) that test halves (formed by randomly splitting items) are not strictly parallel, but have their own test half-specific (method) factors, and (4) that the test scores vary across time due to systematic effects of the situation at the occasion of measurement. The last result indicates that the scales measure not only cohesion and attachment traits, but also cohesion and attachment states. In a second series of analyses, the two single-construct models were combined to a multi-construct latent state-trait model in order to determine the correlation of the latent traits and the correlations between the latent state residuals of the two constructs within the same occasion of measurement. The correlation between the two traits amounted to .61 and was considerably higher than the correlations between the corresponding manifest variables. Furthermore, the within-occasion correlations of the latent state residuals are substantial, indicating that the occasion-specific effects influenced the measures of both constructs in the same direction, i. e., attachment and cohesion states fluctuate synchronously across time.