| Abstract: | ABSTRACT Second, third, and fourth grade children were administered one of three different versions of a word identification test. All versions tested the same target words but used a different combination of two different sets of distractor words. One set was deemed “highly confusing” (i.e., all dis‐tractors were graphically very similar to target words), the other “easily discriminable” (i.e., distractors were dissimilar to target words). One test version contained all highly confusing distractors, a second used all easily discriminable distractors, and a third contained a combination of both kinds. Results showed that at all grade levels children performed significantly better on the easily discritninable version than on the highly confusing version. Implications of this finding related to the content validity of extant word identification tests; that is, if test authors do not use a set of criteria for selecting distractor words and apply these criteria consistently, the content validity of a test may be in serious jeopardy, for one would not know whether the test measures target word difficulty of distractor word difficulty |
