Limitations of analysis of covariance designs in aging research.

ABSTRACT Control of peripheral hearing loss represents a significant design problem in research examining age-related changes in central auditory processing. Statistical control procedures (e.g., partial correlation or analysis of covariance) represent one means of achieving such experimental control. Although these techniques have been advocated for this purpose by some researchers, potential confounding of the data due to statistical dependence between age and peripheral hearing loss may introduce significant bias into the data distorting the reported results.