Alpha Factor Analyses of Reading Measures at the Elementary, Secondary, and College Levels

Abstract

Correlation matrices consisting largely of variables related to reading were corrected for attentuation or errors of measurement, and were factor analyzed separately for fourth graders, senior high school, and college undergraduate students. There were 283 students and 28 variables at the elementary level, 400 students and 46 variables at the high school level, and 126 students and 19 variables at the college level. The major results indicated that speed and comprehension in reading were separate factors at the college level but were combined into one factor of general reading ability at the two younger age levels. A separate factor of word analysis or word attack skills appeared for the two older age groups but was incorporated into the general reading ability factor at the elementary level. The reading comprehension or general reading ability factor at all age levels consisted of a large number of variables or skills that need to be learned in order to read well.Since Holmes published his early studies of the "substrata factor theory of reading" with college students (1954), similar studies, employing somewhat different reading variables, have been conducted on secondary school (Holmes and Singer, 1961) and elementary school students (Singer, 1964(Singer, , 1965. The basic assumption of the substrata factor theory is that a person's ability to read is dependent upon many subabilities or factors, e.g., vocabulary, phonics, etc. How the person integrates and utilizes these factors, along with the complexity of the reading material, and his interests all determine to a large extent his general capacity to read. In order to determine which subabilities or factors underlie general capacity to read the studies by Holmes and by Singer involved administering tests, appropriate to the age level of the students, which on logical or empirical grounds were thought to correlate with either speed or power (comprehension) of reading. Correlation matrices among the test scores were computed separately for each age level. The major analyses of the data then consisted of computing multiple correlations (R) with either speed or power of reading separately as the dependent or criterion variable. The variable with the largest contribution to variance was then used to replace the former criterion variable and a substrata or second-order (R) analysis was then calculated. This procedure was then employed for a third-order substrata (R) analysis (Holmes, 1965).