A Comparison of Pattern Matching Indices.

Area of Science:

• Multivariate statistics
• Psychometrics
• Data analysis

Background:

• Comparing pattern matrices from independent studies is a common challenge in multivariate applications.
• Evaluating the performance of different pattern matching indices is essential for reliable data interpretation.

Purpose of the Study:

• To compare the performance of four pattern matching indices: coefficient of congruence (c), s-statistic (s), Pearson's r (r), and kappa (k).
• To investigate how experimental conditions like saturation, sample size, and number of variables affect index performance.

Main Methods:

• Constructed population pattern matrices by systematically varying saturation, sample size, number of observed variables, and number of derived variables.
• Generated sample patterns and matched them to population patterns using each of the four indices.
• Analyzed the accuracy of each index under different experimental conditions.

Main Results:

• With the exception of Pearson's r, the four pattern matching indices showed similar performance.
• Increased saturation (loading size) generally led to more accurate index values.
• Increased sample size also generally resulted in more accurate index values.

Conclusions:

• Saturation and sample size are key factors influencing the accuracy of pattern matching indices in multivariate studies.
• Most tested indices (c, s, k) perform comparably, suggesting flexibility in their application.
• Researchers should consider saturation and sample size when interpreting results from pattern matrix comparisons.