Teachers’ accuracy in judging students’ giftedness

How accurately teachers can detect giftedness in students has been a frequent subject of research. Judgments about giftedness are mostly dichotomous decisions whether or not a student is gifted but can also be based on rating scales on which teachers rate facets of students’ giftedness (Hoge & Cudmore, 1986). To estimate the accuracy of giftedness judgments, it has to be considered which giftedness criterion teacher judgments are tested against. As outlined in Section 1.2, some scholars see giftedness as high potential, others as actual superior performance. Furthermore, some researchers use intelligence as the sole characteristic for explaining giftedness, whereasothers combine several factors like intelligence, creativity, motivation, and environmental variables (Preckel & Vock, 2013). Additionally, different methods are used to measure these characteristics, for example, with tests, work samples, and ratings and nominations by teachers, school committees, parents, counselors, and peers—as single methods or in

combination (Carman, 2013; Coleman & Galagher, 1995; National Association of Gifted Children, 2013; Ziegler & Raul, 2000). Some scholars like Renzulli and Delcourt (1986) have emphasized that students’ success in a gifted education program should be a criterion for the quality of teacher nominations. Based on this diversity, it is difficult to determine the “true” accuracy of teacher nominations (e.g., Hoge & Cudmore, 1986; McBee, 2006).

Moreover, different kinds of measures are used to estimate different aspects of the accuracy of judgments. Generally, three components of accuracy can be differentiated (Schrader & Helmke, 1987): the level component that allows statements about whether teachers over-, correctly or underestimated students, the differentiation component that indicates, for example, whether teachers over-, correctly or underestimated the variance in cognitive ability among students in a class, and the correlational component that shows, for example, how accurately teachers can sort students’ cognitive abilities into a rank order. In giftedness research, effectivity and efficiency measurements are often reported (see Acar et al., 2016). Effectivity is the percentage of students who are nominated by teachers and are gifted (e.g., as determined by an IQ score above 130) in relation to all gifted students in a sample. Efficiency is the percentage of gifted students in the group of nominated students. The use of effectivity-efficiency measures was seriously criticized by Hoge and Cudmore (1986). Although the measures are dependent on the base rates, base rates were rarely reported and statistical significance tests were missing. Furthermore, Gagné (1994) criticized that the two measurements are negatively correlated. He proposed using the phi (Φ) coefficient that contains the correlation of the two dichotomous variables nomination status and giftedness criterion.

A review of research results on the accuracy of teacher judgments indicates that teacher nominations seem to be more effective than efficient (Heller, Reimann, & Senfter, 2005; Neber, 2004; Wild, 1993). An often cited study is the one by Pegnato and Birch (1959) in which teacher nominations—in addition to, for example, honor roll listing and group intelligence and achievement tests with different cut-off levels—were compared with students’ scores in an individually administered intelligence test. If students had an IQ score of 136 or higher, they were identified as gifted (6.5% in this study). Teachers overlooked more than half of the gifted students (i.e., 45% effectivity). Furthermore, nearly three out of four of the nominated students were not gifted (i.e., 27% efficiency).

The authors concluded that teacher nominations should not be relied upon for the identification of gifted students. A review of 22 studies (Hoge & Cudmore, 1986) that

included the study by Pegnato and Birch reported high variance for both measures: 0-86% effectivity and 4-78% efficiency. Higher values than in Pegnato and Birch’s study but a similar ratio were reported in a recent meta-analysis (Acar et al., 2016) that examined two groups of methods for identifying gifted students. They distinguished non-performance methods like nominations by teachers, parents, and students themselves from performance methods that included tests of academic achievement, cognitive ability, and creativity. If the performance methods were used as giftedness criteria, nonperformance methods were interpreted as effective (59%) but less efficient (39%).

Teacher nominations did not differ significantly in their effectivity or efficiency from the other nonperformance methods.

Gagné (1994) reanalyzed Pegnato and Birch’s (1959) study with the phi coefficient.

Based on these analyses, the accuracy of teacher nomination was with Φ = .29 not significantly lower than the accuracy of the group achievement test and only in two out of four cases significantly lower than the group intelligence tests (if the cutoff IQ for giftedness was > 120 or > 125, but not if the cutoff IQ was > 115 or > 130). Similarly, in the already mentioned meta-analysis by Acar et al. (2016), nonperformance methods correlated moderately with performance methods (r = .30). Heller, Reimann, et al. (2005) concluded based on correlations with an ability test that elementary school teachers were satisfactorily to sufficiently accurate in identifying gifted students. A meta-analysis by Machts, Kaiser, Schmidt, and Mӧller (2016) compared, among other things, teacher giftedness nominations with intelligence tests and reported correlations that ranged from r = .04 to r = .52 across the seven included studies, again with a moderate mean correlation. Wild (1993) controlled for the fact that several students were rated by the same teacher und noted that the spectrum of correlations between teacher judgments and an intelligence test varied strongly from nearly perfect congruence to zero correlations between judgments and tests. If a more comprehensive giftedness criterion that included tests of achievement, cognitive ability, creativity, and motivation was used, high accuracy was noted in a study by McBee (2006). Teacher nominations of students as gifted were with Φ = .51 highly accurate but were outperformed by standardized tests (Φ = .68).

If students’ school success is viewed as criterion, Foreman and Gubbins (2014) showed that students whose teachers nominated them as having high learning potential benefitted more from a mathematics intervention than students who were not nominated.

After controlling for pretest and reasoning ability scores, they received higher posttest

scores. Also, Hunsaker, Finley, and Frank (1997) analyzed the success of students who were nominated by their teachers. Teacher judgments of facets of giftedness were positively associated with students’ performance in a gifted education program as rated by the program’s instructors.

Concerning different types of teacher judgments of giftedness, rating scales for judging facets of giftedness can be an effective aid for teacher judgments about giftedness when they provide a focus on relevant and observable characteristics (see Westberg, 2012, for a review). In Acar et al.’s (2016) meta-analyses, teacher judgments’ accuracy was higher if they used ratings scales instead of solely giving a dichotomous judgment about students’ giftedness. Furthermore, Ashman and Vukelich (1983) reported that it is preferable to use rating scales with items that have more than two response categories.

This result was also supported by Machts et al.’s (2016) meta-analysis.

Overall, teachers tend to misidentify more students as gifted than they overlook gifted students. They are on average moderately accurate in judging giftedness in students, but the level of accuracy measured differs greatly across studies and teachers.

Higher accuracy levels are reported when the giftedness criterion was multifaceted, more closely linked to student achievement, and based on rating scales.