Identification of underachievement with standardized tests, student, parental and teacher assessments. An empirical study on the agreement among various diagnostic sources

Parents, teachers and students themselves are close to irreplaceable as diagnostic sources in the identification of gifted students. The relevant research literature has, however, expressed skepticism concerning the accuracy of such assessments, in particular with regard to the recognition of underachievers. In an empirical investigation, a comparison was made between assessments made by parents, teachers and the students themselves and the results of an intelligence test regarding their efficiency in identifying underachieving and achieving students. Also, the success rates demonstrated by parents and teachers in assessing the motivation and ability self-confidence levels of their children were evaluated. The results demonstrate that these person groups are ill-suited as reliable sources of information. Identification of Underachievement 3 A great deal of interest can be generated for a reliable diagnosis of talent. Only on the basis of such a diagnosis can appropriate suggestions for scholastic planning be voiced and suitable promotional measures be executed. In particular, when it comes to selecting talented students for participation in limited promotional programs or contests, is such a diagnosis irreplaceable. Major diagnostic problems are often the result of limitations on the amount of resources which can be made available for investigations. This frequently leads to compromise, whereby reliable measuring instruments are replaced by more economic, although less reliable, sources. Examples of this practice are visible in the ENTER Model developed by Ziegler & Stoeger (2003) or the „Screening / Preliminary Assessment“ advanced by Feldhusen and Jarwan (2000). Central here are the efforts undertaken to document descriptions of behaviors, relevant for the type of talent in question, observed in everyday activities within the subject’s social environment. The most important sources of information are attachment figures from the social environment (parents, teachers or caretakers). Concrete diagnostic steps will only be taken when indications of a high degree of ability based on their observations have been submitted. An exacting model for such a system is integrated into the recently opened state run school for the highly gifted in the German State of Baden-Württemberg, for which, theoretically, about 1.5 million students are eligible to apply. Since this is from a practical standpoint impossible, all potential applicants who are invited to a comprehensive diagnostic assessment, must first be submitted to a preliminary selection process. A specific problem inherent in preliminary assessments is triggered by the existence of underachievers. Since ability assessments made by parents and teachers are strongly biased by scholastic achievement (Hanses & Rost, 1998; Schrader & Helmke, 1990), they often underestimate the actual ability levels of talented underachievers and consequently don’t register them for the ensuing diagnostic selection processes. Lamentably, underachievers Identification of Underachievement 4 represent exactly that segment of the population of gifted students who are the most in need of promotion, since they are insufficiently challenged by regular classroom instruction, as evidenced by the inadequate (with respect to ability) scholastic achievements they produce (see National Commission on Excellence in Education, 1984; Renzulli & Park, 2000; Richert, 1991; Robertson, 1991; Stephenson, 1985). For this reason, several relevant sources suggest that a generous criterion for admittance to gifted programs be in effect in order to effectively reduce the number of type 2 errors made (e.g. Gagné, 1991; Renzulli, 1994). It will remain questionable as to whether this generous admittance policy actually solves this problem, until a systematic investigation on the quality of ability assessments made by attachment figures from the social environment of underachievers has been published. Actually, according to evidence supplied by several investigations, these sources are very unreliable. Judgment accuracy of talents and personal traits Although ability assessments are better than chance (Murphy, Hall, & Colvin, 2003; Zebrowitz, Hall, Murphy, & Rhodes, 2002), a series of influences exist which can decisively sway the quality of such judgments. These include such factors as more gazing (Borkenau & Liebler, 1995; Wheeler, Baron, Michell, & Ginsburg, 1979), more open arms (Borkenau & Liebler, 1995), a less awkward, stiff or sluggish style of walking (Murphy, Hall, & Smith LeBeau, 2001), a voice that is more pleasant (Borkenau & Liebler, 1995; Murphy et al., 2001), louder (Reynolds & Gifford, 2001) or lower pitched (Borkenau & Liebler, 1995), being more „properly“, more neatly, or more desirably attired (Behling & Williams, 1991; Kwon, 1994), a narrower (Borkenau & Liebler, 1995) or a more symmetrical face (Zebrowitz et al., 2002); less stoutness and greater height (Borkenau & Liebler, 1995). Similar biases have been detected for the judgment of personal traits, such as motivation or anxiety. Although most research on the accuracy of interpersonal judgments has focused on the judgment of states (Hall & Bernieri, 2001), a large degree of research has been conducted on Identification of Underachievement 5 the judgment of traits (e.g., Ambady & Rosenthal, 1992). Although studies have been reported which show that greater acquaintanceship normally leads to higher accuracy in judging personal traits of others (Funder & Colvin, 1997), there are also studies which indicate that strangers with limited behavioral observation opportunities are just as accurate as friends in judging personality traits such as anxiety (Colvin & Funder, 1991). Even among the closest attachment figures of gifted students such as parents and teachers, one would not unconditionally expect to attain the level of accuracy expected in general from diagnostically reliable evaluations. Our own investigations (e.g. Ziegler & Schober, 1999; Heller, Finsterwald & Ziegler, 2000) with parents and teachers recount, therefore not unexpectedly, low to at the most moderate correlations among reliable measurements of motivation and ability self-confidence from parent and teacher assessments. Aims of the study The aims of our study were to examine the quality of ability assessments made by three groups of persons: Parents, teachers and the students themselves, whereby specific attention was to be paid to the assessments of the underachievers. Parents and teachers operate as gate keepers in the learning processes of their children and pupils. They decide, for example, which talents are to be supported and encouraged, provide appropriate learning material or seek professional assistance in talent development. Since self-nominations are in some cases also open options, the assessments the students made of their own ability levels were also taken into consideration. Our initial interest was concretely focused on the quality of the direct ability assessments made by these three groups. Examinations were undertaken to determine how well these assessments agreed with one another and the results of a standardized ability test. Since talent is a quality which cannot be observed directly but rather needs to be derived from behavior patterns and achievement, we were also interested in obtaining assessments of two further meaningful variables. When the ability assessments of specific persons are oriented on Identification of Underachievement 6 the achievements made by the person in question (see. Hanses & Rost, 1998; Schrader & Helmke, 1990), then a correct evaluation can only be made when, in particular, the motivation of this person and its contribution to achievement is correctly evaluated. Furthermore, the perceived achievement of a person is substantially influenced by how much confidence this person has in his/her own abilities and what is communicated to other persons (Wicklund & Gollwitzer, 1982). For this reason, in addition to the direct ability assessments made by parents, teachers and the students themselves, assessments of motivation and ability selfconfidence for the pupils made by these groups were also considered and compared to one another in our empirical study. Of particular interest to us were the quality of assessments made by parents, teachers and students attending the fourth grade, since the state supported school for the highly gifted mentioned above begins with the fifth grade. Method Participants The participants of our study were 317 pupils (152 male, 165 female) attending 26 different fourth grade classes, their parents and their teachers. The pupils took about one hour to work through a questionnaire and an ability test during normal classroom instruction periods. The parents and teachers also receive questionnaires with which they were to evaluate specific variables relating to their children/pupils individually. Measurement instruments The ability levels of the pupils were measured with the assistance of the Raven Test (Standard Progressive Matrices, SPM; Heller, Kratzmeier, & Lengfelder, 1998). The test was chosen for three reasons: First the Raven has very good psychometric properties. Second because it is a group test it is very economical. Third and most important the Raven test shows high correlations to other intelligence tests (cf. Heller et al., 1998), which is why the results can be Identification of Underachievement 7 considered as representative. In the following, pupils will be referred to as underachievers when their z-standardized intelligence quotient was calculated to be at least one standard deviation higher than the z-standardized grade average calculated for their most recent report cards. Additionally, the children, their parents and their teachers were to estimate the ability of the children along a percentage scale. To this end the parents were presented with a line of 100 circles, under which the percentages 0%, 25%, 50%, 75% and 100% were written. The statement “only slightly talented” was paired with the 0% and the statement “extremely talented” with the 100%. Above this line were the following instructions: “How talented do you believe your child/pupil is? Please mark the circle, which best represents your opinion with a cross. You can orient this assessment on the percentage rankings provided.” The children were given a similar version, under their line however were a series of pictures of children, and the instructions read as follows: “ Look below and you will see a line with a lot of circles on it. Under these circles are pictures of five children. On the left is a child who is not very talented. On the right is a child who is extremely talented. We would like to know how talented you think you are. Mark the circle in this line where you would place yourself. So if you think you are not very talented then mark one of the circles on the left-hand side with a cross, if you think you are extremely talented then mark one of the circles on the righthand side with a cross. Important: be sure to place your cross on a circle and not on a child!” We chose to use such a scale because it corresponds to the method usually engaged in the preliminary assessments made for gifted programs. Usually it is required that the students attain a specific percentile on an intelligence test before being admitted to such programs (for example, the state run giftedness school is restricted to the top 2 % of the pupils). In order to evaluate motivation, we engaged a goal orientation scale which consisted of 30 items broken down into a learning goal orientation, an approach orientation and an avoidance orientation. All items were to be assessed along a six-point Likert type scale with Identification of Underachievement 8 the poles 1 (I disagree completely) and 6 (I agree completely). Sample item: In school I want to learn as much as I possibly can (see Ziegler & Stoeger, 2002). Ability self-confidence was measured with the help of a scale developed by Dweck (1999). This 4-item scale uses a structured alternative format. Items are scored on a 6-point scale, where a score of 1 indicates low confidence in ones own math abilities and a score of 6 reflects high confidence in ones own math abilities. Sample item: When I get new work in mathematics, I ́m usually sure I will be able to learn it. vs. When I get new work in mathematics, I often think I may not be able to learn it. The parents completed the same scales which were rewritten to refer to their children. All scales, both those completed by the parents as well as those completed by the children, proved to be reliable (in each case α > .78). For practical reasons it was not possible to have the teachers complete comparative scales for each of their pupils. They were, instead, requested to judge their students with respect to the variables evaluated in the pupil/parent questionnaires along a six-point scale. For example they were to determine whether the pupil demonstrates absolutely no (1) or a great deal of (6) confidence in his abilities. Results Talent assessments Table 1 illustrates how many of the 317 pupils were categorized as below average, slightly below average, average, slightly above average and above average with regard to the intelligence quotients obtained. Furthermore, this table indicates how many of the students placed themselves in these talent groups and how their teachers and parents resolved this inquiry. Insert Table 1 about here Identification of Underachievement 9 In general, ability self-assessments and ability assessments made by parents and teachers reveal a tendency to choose the middle. While according to the results of the SPM only 109 (34,4%) of the 317 pupils have an average degree of talent (85 < IQ < 115), this talent group was seen by 219 (69,1%) of the pupils as being appropriate. The parents classified 251 (79,2%) of their children, and the teachers 244 (77%) of their pupils as to be of average intelligence. An intelligence quotient in the slightly above average region (115 < IQ < 130) was confirmed for 74 (23,3%) of the pupils by the test results. While 76 (24%) of the pupils estimated their intelligence to be slightly above average, the parents only classified 55 (17,4%) and the teachers only 44 (13,9%) of the children into this talent group. A similarly deviant picture could be drawn for the ability assessments made for children with above average abilities (IQ > 130). Although test results indicated that 36 (11,4%) of the pupils had attained an above average ability level, only 10 (3,2%) of the pupils categorize themselves into the highest ability group in their self-assessments. The parents only placed 4 (1,3%) of their children, the teachers only 5 (1,6%) of their students in this group. Although the ability assessments made by the students, their parents and their teachers as well as the SPM results are highly significant, they are only moderately positively correlated (Table 2). This does not by any means secure that they are describing the same variable. Insert Table 2 about here Since the attainment of different criteria on standardized tests for admittance to promotional programs have been applied throughout the literature (see Renzulli, 1994; Renzulli & Reis, 1997; Stanley, 1991), we decided to apply a rather generous cut-off-point and further subdivided the group of considered students into groups of mildly gifted students (SPM percentile between 85 and 98) and gifted students (SPM percentile above 98). In further Identification of Underachievement 10 analyses we will only be looking at the findings made for these two groups of students, since primarily these students would come into question for inclusion in programs designed to promote talent. Out of the 110 students who attained a SPM percentile of 85 or above, 74 were mildly gifted and 36 gifted. In the following, pupils will be referred to as underachievers when their z-standardized intelligence quotient was calculated to be at least one standard deviation higher than the z-standardized grade average calculated for their most recent report cards. Among the 74 mildly gifted students 16 underachievers could be identified, among the 36 gifted students 9 underachievers were found. The mildly gifted students had an average percentile ranking of 88.70, (S = 3.24) on the SPM, for the gifted students M = 98.83 ( S = 0.51; F(1,105) = 187.39, p<.01). The SPM percentile rankings of the achievers and underachievers were not significantly different for either of the talent groups (ability level: F(1.105) = 0.30, p>.10; ability level by achievement level: F(1,105) = 0.19, p>.10). In Figures 1 through 3 the ability assessments made by the pupils, their parents and the teachers are graphically illustrated. Calculations were made for 2 (achievement level: achievers vs underachievers) x 2 (ability level: mildly gifted vs gifted) analyses of variance. The mildly gifted students assess their ability at lower levels than the gifted students do (F(1,105) = 5.04, p<.05). Although the gifted underachievers assess themselves in a similar manner as the gifted achievers do, the mildly gifted underachievers make significantly lower ability self-assessments than the mildly gifted achievers (achievement level: F(1,105) = 7.87, p<.01; interaction: F(1,105) = 7.39, p<.01). This may possibly be grounded in the fact that the ability self-estimations made by the mildly gifted students among both underachievers (Kendall ́s tau: r = .41) and achievers (Spearman: r = .24) were significantly correlated to their school grades, while this was not the case for the gifted students. Insert Figure 1 about here Identification of Underachievement 11 No differentiation in parental ability assessments could be isolated between assertions made for mildly gifted students and gifted students (F(1,105) = 2.33, p>.10). Assessments made for both talent groups erroneously accorded underachievers somewhat less talent than achievers (achievement level: F(1,105) = 3.54, p<.10, interaction: F(1,105) = 0.16, p>.10). However, none of the ability assessments made by parents for the groups here considered were significantly correlated to school grades. Insert Figure 2 about here The teachers correctly assessed the gifted students as being more talented than the mildly gifted students (F(1,97) = 4.30, p<05). However, they were also of the opinion that underachievers were significantly less talented than achievers (achievement level: F(1,97) = 15.41, p<.001; interaction: F(1,97) = 0.27, p>.10). The attributions of higher levels of talent to the gifted students in contrast to the mildly gifted students may be a result of a strong orientation on the school grades they themselves assigned, a point which is supported by the strong correlations found between their assessments and these grades (Spearman; mildly gifted students: r = .45, p<.001; gifted students: r = .57, p<.001). For the underachievers no significant correlation between school notes and ability assessments could be confirmed. Insert Figure 3 about here Subsequent to these evaluations, we would like to draw attention to two further findings: Students, parents and teachers do not significantly vary from one another in ability assessments for gifted students (z<1.80, p>.10). Among the mildly gifted students, the parents attribute higher levels of ability to underachieving pupils than the students themselves (z = Identification of Underachievement 12 2.39, p<.05) and the teachers (z = 2.33, p<.05). A comparison of the percentile ratings on the SPM, the ability self-assessments and the ability assessments made by the parents and the teachers indicates the existence of systematic underestimations throughout all three evaluative groups. This is confirmed to be highly significant by Wilcoxon Tests for both the underachievers as well as the achievers in both ability groups (in each case z>2.56, p<.01). Due to the generally too low assessments of talent, the phenomenon of underachievement fails to materialize in all three of the evaluator groups. Motivation and self-confidence assessments In the next step we would like to investigate how well parents and teachers can assess both the motivation of their children/pupils and how much confidence these children have in their scholastic abilities. In Table 3 the correlations of assessments of motivation and selfconfidence among students, teachers and parents are keyed according to achievement level and ability level. For the mildly gifted underachievers the assessments of motivation among the three groups of persons are uncorrelated. For the mildly gifted achievers a slightly positive correlation could only be found between parental assessments of motivation and pupil test results. In the group of gifted underachievers indications of a positive correlation could only be confirmed between the motivational statements made by the students and the assessments of this trait by their teachers, this correlation turned out to be quite high. For the group of gifted achievers the motivational assessments made by all three groups turned out to be uncorrelated. With respect to confidence assessments for the group of mildly gifted underachievers, the only moderate correlation which could be found exists between parent and teacher evaluations. For the mildly gifted achievers low to moderately positive correlations can be isolated among all three of the groups. In the group of gifted underachievers one highly positive correlation was recorded between self-confidence assessments made by the pupils Identification of Underachievement 13 themselves and the teacher evaluations of this characteristic. For the gifted achievers moderate correlations were detected between the confidence assessments made by the students and the evaluations made by their parents, as well as between the students and their teachers. When scholastic achievement is factored out, no significant changes can be observed in the correlations. Based on these results one can presume that, for ability – as already seen by ability assessments – self-confidence and even more so for motivation, these three groups of persons are not assessing the same variables. Insert Table 3 about here In Figure 4 the motivational assessments made by the pupils are displayed according to ability level and achievement level. Motivational differences could neither be confirmed between mildly gifted and gifted students (F(1,105) = 0.03, p>.10), nor underachievers and achievers for either of the two talent groups (achievement level: F(1,105) = 0.15, p>.10; interaction: F(1,105) = 0.05, p>.10). Insert Figure 4 about here Figures 5 and 6 present the motivational assessments made by the parents and the teachers. The parents ascribe matching levels – similar to those made by the children themselves – of motivation to mildly gifted students and gifted students (F(1,105) = 2.15, p>.10) and also see the motivation of underachievers and achievers to be of similar magnitudes (achievement level: F(1,105) = 0.63, p>.10; interaction: F(1,105) = 0.3, p>.10). In contrast, teacher estimations of motivation vary with regard to both talent group and achievement level. They appraise the motivation of the gifted students to be higher than the motivation of the mildly gifted students (F(1,105) = 5.09, p<.05) and assign higher levels of Identification of Underachievement 14 motivation to achievers than to underachievers (F(1,105) = 18.43, p<.001; interaction: (F(1,105) = 0.43, p>.10). Insert Figures 5 and 6 about here Conspicuous here is that the parents substantially and across groups – with the exception of the mildly gifted underachievers – underestimated the motivation of their children (Friedman; mildly gifted students: underachievers: Chi2(1)=2.25, p>.10; achievers: t(56)=3.74, p<.001; gifted students: underachievers: Chi2(1)=2.78, p<.10; achievers: Chi2(1)=6.26, p<.05). The teachers underestimated the motivation of the mildly gifted underachievers (Friedman; Chi2(1)=6.25, p<.05) and correctly evaluated the motivation of the gifted underachievers (Friedman; Chi2(1)=1.00, p>.10). They overestimated the motivation of both the mildly and gifted achievers (mildly gifted achievers: t(56)=2.71, p<.01; gifted achievers: Friedman: Chi2(1)=4.81, p<0.05). According to the information displayed in Figure 7, the confidence mildly gifted students have in their scholastic abilities does not significantly differ from that held by the gifted students (F(1,105) = 1.12, p>.10). In contrast, the underachievers in both groups demonstrated less confidence in their scholastic abilities than the corresponding achievers (F(1,105) = 15.54, p<.001; interaction: F(1,105) = 0.02, p>.10). Insert Figure 7 about here Illustrated in Figures 8 and 9 are the confidence assessments made by the parents and the teachers. The parents assessed the confidence held by the mildly gifted students to be significantly lower than that of the gifted students (F(1,105) = 11.47, p<.01). They appraised the confidence the underachievers had in their scholastic abilities to be somewhat lower than Identification of Underachievement 15 the same quality among achievers (F(1,105) = 3.44, p<.10; interaction: F(1,105) = 1.33, p>.10). The teachers also correctly deemed the confidence underachievers had in their scholastic abilities to be lower than that for achievers (F(1,105) = 14.03, p<.001). The two ability groups could not be differentiated with respect to scholastic self-confidence (F(1,105) = 0.13, p>.10; interaction: F(1,105) = 0.57, p>.10). Insert Figures 8 and 9 about here The parents overestimated how much confidence their gifted students had in their scholastic abilities, independent of whether they were underachievers or achievers (Friedman; underachievers: Chi2(1)=5.44, p<.05; achievers: Chi2(1)=9.78, p<0.01). In contrast, the parents of mildly gifted students were able to correctly assess how much confidence their students had in both of the achievement groups (Friedman; underachievers: Chi2(1)=0.07, p>.10; achievers: t(56)=0.56, p>.10). The teachers slightly underestimated the confidence their gifted underachievers had in their scholastic abilities (Friedman; Chi2(1)=2.78, p<0.10) and were able to correctly appraise self-confidence for the other groups (mildly gifted students: underachievers: Friedman; Chi2(1)=1.00, p>0.10; achievers: t(56)=0.37, p>.10; gifted students: achievers: Friedman; Chi2(1)=2.67, p>.10). Summary and discussion Highly gifted underachievers are exactly those who profit most from special promotional measures, such as a state run school for gifted students, since they cannot be optimally fostered by regular classroom instruction (see Renzulli & Park, 2000; Richert, 1991; Robertson, 1991). However, in practice, it is often the case that these students are neglected by special promotional programs. Resource problems are primarily responsible for Identification of Underachievement 16 this. Due to the great demand on many promotional programs, only a minimal proportion of the applicants will be able to participate in the actual diagnostic, a preliminary selection process must often be employed. In this preliminary selection process the opinions of various attachment figures are often taken into consideration, to ascertain whether the child in question has a high degree of talent or not (see Feldhusen & Jarwan, 2000; Ziegler & Stoeger, 2003). Should, on the basis of the opinions of these attachment figures, giftedness appear to be improbable – which is often the case among gifted underachievers – the applicant will be turned down, without having had the benefit of further diagnostic measures. In our study, therefore, an examination was made to determine how well ability estimations made by parents, teachers and the students themselves agree with one another and with the results of a standardized intelligence test. Particular attention was paid to underachievers. Since persons making ability assessments are to some extent orientated on (scholastic) achievements (see Rost & Hanses, 1997; Schrader & Helmke, 1990) and since these are not only stipulated by ability but also by motivation, we were also interested to determine how well parents and teachers were able to assess the motivation of their children and pupils. Finally, the effectiveness of parents and teachers in assessing the self-confidence their children/pupils had in their scholastic abilities was put to the test. This interested us fundamentally because persons who, due to their high levels of self-confidence in their own abilities, communicate higher achievement rates, are often judged to be gifted (Wicklund & Gollwitzer, 1982). In general, ability assessments by all three groups of persons were dominated by a tendency to move to the middle. Most of the pupils judged themselves to be of average intelligence and were also placed in this category by their parents and teachers. The group of mildly gifted students attracted just about as many students as were to be expected, based on the results of the intelligence test, in comparison the parents and teachers held far too few students for mildly gifted with respect to the actual test results. According to the ability selfIdentification of Underachievement 17 assessments, as well as the parent and teacher ability assessments, a considerably small number of students were to be classified as gifted students, notably fewer than the actual number confirmed by the results of the intelligence test. Between the results of the intelligence test and the ability assessments made by the parents, teachers and the students themselves, one can only confirm moderate relationships of statistical significance, which may indicate that these variables are not referring to the same entity. If one chooses to inspect – as is usually the case for single factor models of giftedness – a measure of discrepancy between cognitive abilities (IQ) and goal driven achievements (grades), in our sample about one quarter (21,62%) of the mildly gifted students and exactly one fourth of the gifted students turned out to be underachievers. This proportion is relatively small in comparison to the findings of other investigations (see Richert, 1991). When one considers the fact that our investigation was conducted in fourth grade classrooms, then this rather low rate is not very remarkable. Causal factors for underachievement, such as low levels of motivation or self-confidence (see Butler-Por, 1993), occur rather infrequently at this age level. In contrast to older students, forth grade school children can be characterized by high levels of motivation and a general optimism with respect to their learning and intellectual capabilities (Nicholls, 1992), which could possibly work as a protective shield against underachievement. Regardless of talent group (mildly and gifted students) and whether underachievement was at hand or not, the assessments made by the students, their parents and their teachers were all lower than the actual results obtained with the intelligence tests. This systematic underestimation of ability had the curious repercussion that, from the point of view of the students, the parents and the teachers the phenomenon of underachievement could not be discerned at all. Although the pupils themselves and the teachers – regardless of the generally too low assessments – attribute higher levels of ability to the gifted students than to the mildly gifted Identification of Underachievement 18 students, the ability assessments made by the parents for these two groups cannot be differentiated from one another. Teachers and parents from both talent groups judge the underachievers to be significantly resp. somewhat less talented than the achievers. In contrast, only for the group of mildly gifted students do the underachievers see themselves as being less talented than the achievers do. The underachievers and achievers in the group of gifted students assess themselves as being equivalently talented. This can possibly be traced back to the significantly positive correlation found between the ability assessments of the mildly gifted students and school grades, which could not be duplicated for the gifted students. A comparison made among the ability assessments made by parents, teachers and students reveals no indication of differences in opinion among the gifted students. In the group of mildly gifted students these evaluations made for achievers are also very similar to one another among these three groups of persons. The ability of the underachievers in this group is judged by the parents to be higher than evidenced by judgments made by the teachers and the students themselves. In summation, the usefulness of ability assessments made by parents teachers and the students themselves appears to be rather poor. Their evaluations only moderately correlated with one another and with the results of the intelligence test, which indicates that they may not have been assessing the same variable. In addition, all three groups of persons made ability estimations which were too low in terms of the results of the test, which makes it practically impossible to recognize the phenomenon of underachievement. A similarly negative picture is drawn from the information gathered regarding motivation assessments. Only two correlations could be substantiated; between the mildly gifted achievers and their parents, and between the parents and teachers of the gifted underachievers. Here we are confronted with the same verdict maintained for assessments of ability, in that we cannot be certain that the same variable is being measured by the parties in question. Identification of Underachievement 19 All the same, the parents of the mildly gifted and gifted students correctly judged the motivation of the achievers and underachievers to be equivalent within each of the two ability groups. However, their evaluations of motivation for all groups – with the one exception of mildly gifted underachievers were significantly lower than the test results obtained from the students. In stark contrast to the parents, the teachers assigned higher motivation levels to the gifted students than they did to the mildly gifted students. Furthermore, they judged the achievers in both ability groups to be more motivated than the underachievers. This result can very possibly be due to a strong influence of scholastic achievement being exercised on the judgments made here by the teachers (see also Hanses & Rost, 1998; Hany, 1993; Schrader & Helmke, 1990). In accord with this synopsis, the teachers overestimated the motivation levels of the achievers in both ability groups. The motivational drive of the mildly gifted underachievers was underestimated by the teachers, and the gifted underachievers were accurately judged. The statements gathered from the pupils as to the confidence they had in their scholastic ability and the assessments made by their parents and teachers on this variable did result in more correlations than the assessments of ability and motivation, they do not however indicate that all three groups of persons are referring to the same variable. Correlations were predominantly isolated for estimations made by achievers, their parents and their teachers. The test results obtained for the mildly gifted achievers showed moderately positive correlations between the assessments made by parents and the students, and between the teachers and their students. The parent and teacher assessments made for this group were also positively correlated. While the teachers, in harmony with the statements made by the pupils, saw no need to differentiate between levels of self-confidence displayed by mildly gifted and gifted students, the parents attributed lower degrees of self-confidence to the mildly gifted students than to the gifted students. The parents as well as the teachers correctly Identification of Underachievement 20 assigned lower levels of self-confidence in scholastic ability to underachievers than to achievers. However, the parents of gifted students overestimated the confidence these students have in their abilities. The teachers slightly underestimated the self-confidence their gifted underachievers have in their scholastic ability. All things considered, our study leads us to a rather skeptical opinion on how appropriate assessments made by parents, teachers and the students themselves can be as a diagnostic source for the evaluation of giftedness or underachievement. Particularly remarkable and thought provoking is the fact that the phenomenon of underachievement does not materialize at all due to a systematic underestimation of ability levels in general. If an attempt is being made to recruit suitable candidates for a promotional program specifically designed for gifted students, then the opinions held by the student candidates, their teachers or their parents cannot be confirmed as reliable sources. Since in many cases limitations on investigative resources do not permit one to forgo these assessments, it is essential to contemplate measures which can be taken to improve the quality of assessments made by these groups of persons. Whether a special seminar for parents and teachers could lead to better identification of gifted students, and in particular of gifted underachievers, needs to be clarified in the framework of subsequent studies. For example, one could supply parents and teachers with detailed information regarding the problem of underachievement, and explain to them that they should rely more on the competence of the child in question and less so on his/her scholastic achievements or learning behavior when making ability assessments. However, experiences made by previous studies (see Gear, 1978) generate apprehension that this method may not lead to an improvement in the quality of identification, but rather solely to an increase in the nomination rate – irregardless of whether underachievement is at hand or not. Implications for research and practice Identification of Underachievement 21 The study at hand was able to uncover two central findings. First, the combination of psychometric tests and assessments made by parents, teachers and the students themselves provide us with very contradictory images. In particular the attribution of general ability levels led to highly differential assessments. Second, an answering tendency was identified which was extremely significant not only for the investigation of talent, but essentially for the identification of talents: Assessors refrain from assigning pupils to either of the extreme talent categories. Due to this bias, hardly any students were designated to be highly gifted, therefore teachers, parents and the students themselves appear to be unsuitable informants of giftedness. Even when this talent criteria is relaxed for example when teachers, parents and pupils are to categorize talented individuals into the top 20% the results are disappointing and a significant proportion of gifted students remain unrecognized. It must be mentioned that these findings, which are surely unexpected for those involved with giftedness research and promotion, are concordant with findings found by studies on the assessment abilities of persons. These two findings, which could be replicated in other studies, have varied and meaningful implications for both giftedness research as well as the identification of talents in the praxis. In retrospect, the results of several research studies in which the identification of talented persons were based entirely on assessments made by individuals, must now be placed in question. Also when a combination of assessments and psychometric tests are applied, a critical examination must be made to determine, which source and to what degree this source contributed to the final evaluation. Concerns regarding the quality of these identifications are also valid with respect to diagnoses made in practical environments insofar as they were primarily based on assessments. For the time being, in practice as in research, there does not seem to be a viable alternative to psychometric testing, whereby our investigation was able to demonstrate their superiority in comparison to assessments, but not their validity. We would like to recommend that future studies take a closer look at questions of identification, due to Identification of Underachievement 22 the fact that it is a ground premise of every relevant research area, and it permits the object of the investigation to be identified with sufficient confidence.