skip to main content skip to footer

Construct Validity Study of the NTE Core Battery Using Confirmatory Factor Analysis NTE

Author(s):
Grandy, Jerilee E.
Publication Year:
1992
Report Number:
RR-92-03
Source:
ETS Research Report
Document Type:
Report
Page Count:
62
Subject/Key Words:
National Teacher Examinations Policy Council, Construct Validity, Factor Analysis, NTE Core Battery, National Teacher Examinations (NTE), Test Reliability, Test Validity

Abstract

This two-part study investigated the construct validity and population generalizability of the National Teacher Examination (NTE) Core Battery. The report explains the logic of construct validation, in general, and explores the construct validity of the NTE Core Battery in particular. The study used confirmatory factor analysis to model the structure of test scores in relation to the knowledge and abilities (i.e., the constructs) they purport to measure. According to its specifications, the Core Battery measures achievement in three broad areas: Communication Skills, General Knowledge, and Professional Knowledge. Communication Skills and General Knowledge each consists of four subtests that should be somewhat related in content but still be different from one another. Statistically, each of the four sets of subtests should be moderately correlated but not so highly correlated that their scores are redundant. In Part I, a nine-factor model (one factor for each subtest) with data from the November 1982 administration was tested. Results indicated that the factors correlated too highly to be different constructs. Furthermore, the subtest scores did not group into the three factors defined by test specifications. The nine-factor model with various other models was then compared and it was found that the simplest model to fit the data was a three-factor model consisting of general academic skills, mathematics, and essay factors. The simpler model fit the data very well, accounting for nearly all of the variance explained by the nine-factor one. Part II, which used data from a later test version, entered tetrachoric correlations among items into an exploratory factor analysis. Results were a general factor, a mathematics factor, and a factor with a few items from the Social Studies test. Confirmatory factor analysis yielded the same results obtained in Part I. Implications for construct validity were that the Core Battery, contrary to design, measures three constructs: general academic skills, mathematics, and essay writing. Aside from the fact that the Mathematics test and a few Social Studies items were different from Communication Skills, there was no other construct validity evident. To investigate population generalizability, the same model was tested simultaneously across four populations: White males, White females, Black American males and Black American females. No evidence was found that the test was biased in the assessment of any of these groups. (62

Read More