S-Tests and C-Tests: Measures of Content-Based Achievement at Grade Four of High Schools

Abstract

This study aims to explore the internal, empirical as well as content validity of C-Tests and S-Tests developed on the passages of the textbook Learning to Read English for Pre-University Students (Birjandi, Sarab, & Samimi, 2011) taught to grade four high school students in Dargaz, Iran. To this end, all the single and phrasal words of the textbook were treated as schemata and assigned to semantic, syntactic and parasyntactic domains, genera and species. By choosing a number of paragraphs from the passages of the textbook, a 100-item C-Test and 90-item S-Test were designed and administered to 283 students. The reliability estimates showed that the C-test was more reliable than the S-Test. When the two tests were subjected to correlational analyses they correlated very highly with each other (r=.99, p<.01), establishing themselves as empirically valid measures of English achievement. The analysis of types and tokens comprising the two tests, however, showed that the S-Test enjoyed higher content validity because it measured more schema types than the C-Test did and its difficulty level was, therefore, higher. The results are discussed and suggestions are made for future research.