The test expectancy effect: Is it a question of format or the level of questions?

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)




Vernon Hall


depth of processing approach, encoding specificity effect

Subject Categories

Educational Psychology


The test expectancy effect (TEE) is a metacognitive phenomenon wherein students commonly request information about the format of an examination (i.e. multiple choice or essay). It is presumed that this information provides a cue about how or what to study and that students adjust their examination preparation accordingly.

In general, previous investigations of the phenomenon have found that students who prepare for a recall examination (essay or completion) do better on that type of examination than students who have prepared for a recognition (multiple choice or true-false) examination.

Two different theoretical formulations, a depth of processing approach (Craik & Lockhart, 1972) and an encoding specificity effect (Morris, Bransford, & Franks, 1977) have been offered to explain the phenomenon. The current study investigated an additional variable, cognitive level of demand of questions, which previous research had partially ignored.

Two studies are reported. In the first study college students were asked to write 3 MC and 2 essay questions over text material which they were studying in class. As predicted, questions written in an essay format reflected higher classifications according to Bloom's Taxonomy than MC items.

A second study was conducted to determine if knowledge of format would influence performance on items varying both in format and in level of question. In addition subjects were asked to complete a questionnaire concerning their experiences with, perceptions of, and preferences for MC or essay tests.

A 3-way, repeated measures analysis of variance with Expectancy (MC or Essay) as the between-subjects variable and Format (MC or SAE) and Level of Question (high or low) as the within-subjects variables. No main effect for Expectancy was found. The Expectancy by Format and the Expectancy by Level interactions were also nonsignificant. Likewise there was no significant difference between the two expectancy groups on the long essay questions.

Thus, even though students appear to recognize consistent differences in the demands of MC versus essay items, as shown by their responses to the format questionnaire, these differences did not translate into differential performance when expectancy was manipulated.