This analysis examines test scores which have been derived from the correction-for-guessing formula. Studies referred to have shown that significant discrepancies usually exist between Corrected Scores and Absolute Scores (measures of “pure knowledge”) for group averages. New data are presented which show that the correlations between Corrected Scores and Absolute Scores are not high enough to interpret the two as measuring the same attribute. Correlations of Corrected Scores with Relative (total right) Scores are significantly higher than with Absolute Scores. The conclusion is that Corrected Scores are not good estimates of “pure knowledge,” either for group or individual data.
Get full access to this article
View all access options for this article.
References
1.
DavisF.Educational measurements and their interpretations. Belmont, Calif.: Wadsworth, 1964.
2.
LittleE.Overcorrection and undercorrection in multiple-choice test scoring. J. exp. Educ., 1966, 35, 44–47.
3.
LittleE.CreaserJ.Uncertain responses in multiple-choice examinations. Psychol. Rep., 1966, 18, 801–802.
4.
LittleE.CreaserJ.Undercorrection for chance success on multiple-choice examinations. Improving Coll. & Univer. Teaching, in press.
5.
McCallW.A new kind of school examination. J. educ. Res., 1920, 1, 32–46.
6.
McNemarQ.Psychological statistics. New York: Wiley, 1949.
7.
MeadA.SmithB. M.Does the true-false scoring formula work? Some new data on an old subject. J. educ. Res., 1957, 51, 47–53.
8.
RuchG.StoddardG.Comparative reliabilities of five types of objective examinations. J. educ. Psychol., 1925, 16, 89–103.
