LDS Church Growth, Member Activity, and Convert Retention:
Review and Analysis By David Stewart

Return to Table of Contents

Chapter III-05: Power and Confidence Interval

Adequate numbers of respondents are necessary to generate the statistical power to identify relevant trends or findings. Larger samples produce narrower "confidence intervals" (defined by two standard deviations on either side of a result) and lower standard errors, providing higher confidence that the result obtained is representative.

Studies vary widely in their statistical power. A national census which lists the number of individuals identifying the LDS Church as their faith of reference with six significant figures provides a much higher degree of precision than off-the-cuff statements of church leaders offering estimates for church attendance with only one or at best two significant figures. With national census is an independent sociologic studies, the methodology is published along with the findings. This provides an essential tool in defining terminology as well as in assessing strengths or limitations of methodology.