Test anxiety and intelligence testing: A closer examination of the stage-fright hypothesis and the influence of stressful instruction

Abstract

The influence of test anxiety and the content of instruction (stressful versus reassuring) on measurements of intelligence were investigated. It was expected that components of test anxiety would show differential effects on test performance. A Latin Square design was used to unravel the effects of test type and test order. Furthermore, effects of type of instruction, stressful versus reassuring, were studied by means of a within-subjects design. Test anxiety was measured with the Revised Worry-Emotionality Questionnaire. Measurements for verbal ability, reasoning, and memory were administered. Performance on memory tests showed less vulnerability to test anxiety compared with the other tests, with a picture recall test being insensitive. The negative effect of test anxiety was mostly confined to the beginning of a test session, independent of the type of test. Partial support for the so-called stage-fright hypothesis was found. The effect of instructional content was equivocal.

Keywords:

• Test anxiety
• intelligence
• education
• memory
• verbal ability
• reasoning ability

Acknowledgments

The authors wish to thank professor Charles D. Spielberger for his constructive and helpful comments.

Notes

1. Since the two-factor model is a special case of the three-factor model (worry and self-confidence are collapsed), the models are nested. The unexpected high values of RMSEA may be due to the relatively small sample size and a suboptimal ratio of sample size to the number of variables. Hu and Bentler (1999) contend that in small samples (i.e., less than 250), RMSEA performs poorly in the model. That is to say, using RMSEA as a criterion for model acceptance may lead to over rejection of true-population models.