Purchase this article with an account.
Erin M Harvey, Tina K Green, Kathleen M. Mohan, Marjean T Kulp, Amy Davis, Joseph M Miller, John Daniel Twelker, Irene Campus; Inter-Scorer and Test-Retest Reliability of the Beery-Buktenica Developmental Test of Visual-Motor Integration in School-Age Children. Invest. Ophthalmol. Vis. Sci. 2016;57(12):1550.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Assess inter-scorer and test-retest reliability of the Beery-Buktenica Developmental Test of Visual-Motor Integration, 6th Edition (VMI) (Pearson Clinical Assessment, Bloomington MN).
VMI was administered to 163 3rd–8th grade students twice on separate days. All tests were scored by two research assistants who received extensive training in scoring, and 50 tests were also scored by a more experienced scorer. Reliability analyses for inter-scorer and test-retest differences using raw scores included 95% limits of agreement (LOA) (mean difference +/- 1.96 SD), Bland-Altman plots, Wilcoxon Signed Rank Tests comparing median differences to 0 (assessment of bias) and Spearman correlation.
Summary results are provided in Table 1 and Figures 1 and 2. The mean inter-test interval was 63.38 days, SD 33.98. The 95% LOA indicated that most scores agreed within approximately +/- 3 for the experienced scorer and the two trained scorers with no significant bias in scoring (median difference not significantly different from 0). For comparisons between the two scorers, most scores agreed within approximately +/- 4 points, with one scorer tending to give higher scores. For test-retest comparisons, 95% LOA indicated that most scores agreed within approximately +/- 6 points, with no significant bias. Correlations were strong for inter-scorer agreement and moderate for test-retest agreement (all p values <0.001).
Slightly wider 95% LOA for the two trained scorers than in comparisons between the trained scorers and the experience scorer suggest that precision of results may vary with scorer experience. The 95% LOA for test-retest results provide an expected range of change in scores between test and retest that may be useful for determining the amount of change considered to be meaningful for this test. Most previous research has assessed VMI reliability using correlation, which assesses the relation between two measures but does not assess measurement agreement. Further research on test-retest reliability using Bland-Altman methods on additional populations is recommended, particularly if the test is to be used to detect changes due to intervention.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.
This PDF is available to Subscribers Only