The Psychometric properties of test items in a Junior High School mathematics entrance-placement examination : implications for test item banking

Date of Award

2017

Document Type

Thesis

Degree Name

Master of Arts in Education, major in Basic Education Teaching (Option 1: Thesis)

Department

Education

First Advisor

Soto, Cornelia C., Ph.D.

Abstract

This study used correlation, factor analysis, and item analysis to look at the psychometric properties of a Junior High School Mathematics Entrance-Placement Examination (MEPE) and their implications for test item banking. It examined the difficulty index, discrimination index, reliability, and validity of the test items. Based on the results of item analyses, 11 out of 60 (18.33%) of the test items were very difficult; 27 out of 60 (45.00%) questions were difficult; 13 out of 60 (21.67%) questions were average; 11 out of 60 (13.33%) were easy; and 1 out of 60 (1.67%) question was very easy. Moreover, in terms of discrimination index, 20 out of 60 (33.33%) of the items have very good, 6 out of 60 (10%) are good, 13 out of 60 (21.67%) are fair, and 21 out 60 (35%) are poor. The results of the reliability test indicated a KR-20 coefficient of 0.81524, indicating that the MEPE is very reliable. The findings of peer evaluation for content validity showed that the MEPE covers basic topics in elementary mathematics, the allowed time for the examination is ideal, the examination consists of definite and clear problems, the examination questions are answerable by examinees, the examination reveals no pattern for the correct answers, the examination choices are plausible, and the examination can still be made more challenging. The outcomes of the Pearson Product-Moment Correlation revealed a moderate correlation between the results of the MEPE and the fourth quarter grades (or equivalent) from the previous school year for concurrent validity and a significant correlation between the results of the MEPE and the guidance test for predictive validity. The results of factor analysis for construct validity showed that most of the questions were leaning towards application questions.

Comments

The E3.C87 2017

Link to Full Text

Share

COinS