Effect of number of response options on the psychometric properties of Likert-type scales used with children


ALAN Ü., ATALAY KABASAKAL K.

STUDIES IN EDUCATIONAL EVALUATION, vol.66, 2020 (SSCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 66
  • Publication Date: 2020
  • Doi Number: 10.1016/j.stueduc.2020.100895
  • Journal Name: STUDIES IN EDUCATIONAL EVALUATION
  • Journal Indexes: Social Sciences Citation Index (SSCI), Scopus, Academic Search Premier, ASSIA, EBSCO Education Source, Education Abstracts, Educational research abstracts (ERA)
  • Keywords: Likert-type scales, Number of response options, Children, Psychometric properties, Validity, Reliability, SELF-REPORT MEASURES, R PACKAGE, RELIABILITY, CATEGORIES, ELEMENTARY, STUDENTS, 4-POINT, LABELS, ALPHA
  • Anadolu University Affiliated: Yes

Abstract

Although Likert-type rating scales are used in a great number of early childhood studies, knowledge of how the number of response options affects the psychometric properties of scales used with children is limited. The purpose of this study is to contribute to this knowledge. Data were collected from second grade students and third grade students. Accordingly, 1,092 second-and third-graders completed a 2-point, 3-point, and 4-point version of the School Attachment Scale for Children and Adolescents. Participants came from 11 schools, different in terms of socioeconomic status. The children received the versions approximately three weeks apart. Results revealed that as the number of response options increased, the means tended to decrease and the distribution to be normal. For the 2-point version, most items were below the cut-off point in terms of discrimination indexes. Compared to the 2-point version, there was a significant increase in discrimination indexes for the 3-and 4-point versions, and the items' discrimination indexes were high. It was concluded that the reliability coefficient increased with an increasing number of response options for all subdimensions of the scale. When the validity estimations of the three subdimensions were examined for the three versions of the scale, it was found that the 3-and 4-point versions were appropriate for the validity and that the validity of the 2-point version was weak. It was observed that using 2-point Likert-type scales with children negatively affected the psychometric properties and that these properties improved with an increased number of response options.