Linear Regression of Sampling Distributions of the Mean

Area of Science:

• Statistics
• Bioinformatics
• Computational Biology

Background:

• Linear regression is a fundamental statistical method.
• Gene expression analysis often involves large datasets.
• Hierarchical clustering is used for grouping similar genes.

Purpose of the Study:

• To demonstrate the equivalence of regression coefficients and R-squared calculated from sampling distributions versus individual data.
• To show the reduction in the standard error of estimate when using sampling distributions.
• To explore applications in gene expression correlation and clustering.

Main Methods:

• Calculation of simple and multiple linear regression coefficients.
• Computation of the coefficient of determination (R-squared).
• Analysis of sampling distributions of the mean (with and without replacement).
• Application of Pearson correlation coefficient for gene expression analysis.

Main Results:

• Regression coefficients and R-squared are identical whether computed from sampling distributions or individual data.
• Standard error of estimate is reduced by the square root of the group size for sampling distributions.
• Pearson correlation can effectively measure the correlation between differential expression of two genes.

Conclusions:

• Statistical analyses using sampling distributions of the mean are equivalent to using individual data for regression.
• This finding simplifies analyses with grouped data and reduces the standard error.
• The methodology is applicable to gene expression studies, particularly in hierarchical clustering and correlation analysis.