Reliability of the Fry Sampling Procedure
Gisela G. Fitzgerald · Reading Research Quarterly · 1980
sampling reliability of the procedure recommended by Fry in estimating a given book's mean readability level. Two distinct but related data analyses were undertaken: 1) a computation of confidence intervals based on variable sample sizes, and 2) a computation of power functions based on variable sample sizes. A sample size of three was found to be inadequate in estimating population readability means because 1) the standard error among samples is frequently very high; 2) the confidence intervals on samples of three are often so broad as to be of little value in locating the true population mean; and 3) the probability of Type II errors is unacceptably high with this sample size.