European Journal of Psychology of Education

, Volume 28, Issue 2, pp 275–295

Teacher judgment, student motivation, and the mediating effect of attributions

Authors

    • Psychology of Excellence in Business and EducationLudwig-Maximilians-University of Munich
  • Detlef Urhahne
    • Department of Educational PsychologyUniversity of Halle
Article

DOI: 10.1007/s10212-012-0114-9

Cite this article as:
Zhou, J. & Urhahne, D. Eur J Psychol Educ (2013) 28: 275. doi:10.1007/s10212-012-0114-9

Abstract

Based on Weiner’s attributional theory of intrapersonal motivation, the mediating effect of attributions between teacher judgment and student motivation was examined. In two studies, 144 German and 272 Chinese fourth-grade elementary school students were tested on their mathematical achievement, causal ascriptions for success and failure, expectancy for success, self-concept, and test anxiety. Mathematics teachers were asked to estimate students’ performances on the applied mathematics test. Discrepancies between teacher judgment and student performance led to groups of underestimated and overestimated students. One year later, Chinese students were retested on their mathematical achievement. Results show that the attributional pattern of underestimated students was maladaptive compared with overestimated students. Attributions mediated the effect of teacher judgment on students’ expectancy for success, self-concept, test anxiety, and, in case of the Chinese sample, mathematics achievement of the next year. The results indicate the important role of student attributions as a function of teacher judgment and imply attribution retraining as a possible intervention.

Keywords

Teacher judgmentAttributionMotivationSelf-conceptTest anxiety

Introduction

Teachers’ judgments of students’ academic achievements play a key role in the educational system. Teachers are responsible for giving fair grades and feedback to students (Shavelson and Stern 1981), making pedagogical decisions about student enrolment and curriculum changes (Borko and Cadwell 1982; Demaray and Elliott 1988), and preparing assessment reports for parents (Sharpley and Edgar 1986). Teachers’ judgments have a decisive influence on students’ motivation to learn and their willingness to put effort into tasks. They enable students to make better estimations of their abilities (Spinath and Spinath 2005), which in turn affect their motivation level.

However, there is no simple, direct relation between teacher judgment and student motivation. Each student perceives teacher judgment differently, and students’ explanations vary for why a teacher has judged a performance in a certain way (Weiner 1986). The differences in students’ causal attributions have an impact on psychological consequences such as expectancy for success, self-efficacy perceptions, and affective reactions (Weiner 1992). Accordingly, it can be assumed that students’ attributions might work as mediators between teacher judgment and student motivation. In the following, an introduction is given to the fields of teacher judgment, attribution research, and their relation to student motivation.

Theoretical background

Teacher judgment and student motivation

Acknowledging the importance of teacher judgment for students’ school and vocational career, questions of the accuracy of teacher judgment and the influence of inaccurate judgments on students’ perceptions and behavior have aroused much attention (Artelt et al. 2001; Bates and Nettelbeck 2001; Demaray and Elliott 1988; Feinberg and Shapiro 2009; Hoge 1983; Hoge and Butcher 1984). Teachers are in a certain way accurate judges of students’ academic performance. The mean correlation between teacher judgment of student achievement and their actual performance is, according to meta-analyses, between 0.61 and 0.66 (Hoge and Coladarci 1989; Südkamp et al., submitted for publication). Although teachers in general seem to be fair and objective (Hoge and Coladarci 1989; Südkamp et al., submitted for publication), differential teacher treatments, even sometimes implicit or unintentional, will be sensitively perceived by students and impact their self-concept (Blöte 1995). Teacher’s influence might be more pronounced for some special groups, e.g., students reporting higher differential teacher treatment (McKown and Weinstein 2007). In an investigation of fourth-grade elementary students in Austria, a group of underestimated students was contrasted with a group of accurately estimated and overestimated students (Urhahne et al. 2011). Research results revealed that underestimated students had the same test performance as non-underestimated students but showed lower expectancy for success, displayed lower academic self-concept, and felt more test anxiety. Two other investigations on motivation and emotions of under- and overestimated students could confirm these findings (Urhahne et al. 2010). Again, the pattern emerged that underestimated students had lower expectancy for success, a lower self-concept, and more test anxiety than overestimated students.

Brophy (1983) has provided a plausible explanation why teacher’s misjudgment of student achievement can influence students’ self-perceptions. He proposed a model comprising the following stages: (a) the teacher forms differential expectations for student performance; (b) the teacher treats students differentially; (c) this behavior communicates teacher’s expectancies; (d) and leads to changes in students’ motivation, emotions, and self-concept. Moreover, he assumed that (e) these effects will reinforce the teacher’s expectations, and (f) ultimately, these changes will be reflected in student achievement, showing that teacher’s expectations can function as self-fulfilling prophecies (Rosenthal and Jacobson 1968). The current investigation is concentrating on the link between teacher’s differential expectations for student performance and the changes in students’ motivation, emotions, self-concept, and later achievement. While the teacher is seen in Brophy’s (1983) model as the key factor of these changes in students’ thinking and performance, it is worthwhile knowing how students perceive and explain messages from the teacher.

Attributional theory

The link between teacher judgment and student motivation can be better understood by Weiner’s attributional theory of intrapersonal motivation (Graham and Williams 2009; Weiner 1983, 2005, 2010). A modified version of Weiner’s attributional theory (2005, 2010), which offers a general framework for our research, is depicted in Fig. 1. The beginning of the model is a discrepancy between teacher judgment and student achievement. Students who are misjudged by the teacher can be assigned to two different groups, labeled as overestimation and underestimation. The overestimated group has achieved a positive outcome because of their positively judged achievements among teachers, while the underestimated group has a negative outcome due to their low-assessed achievements. Misjudged students can attribute their outcomes along three causal dimensions: locus of causality, stability, and controllability (Weiner 1986). On the locus of causality dimension, the individual attributes the outcome either to internal factors like ability, effort, or mood or to external factors like task difficulty, chance, or significant others. On the stability dimension, the person attributes either on stable causes such as ability, task difficulty, or temporal causes such as effort, mood, or chance. On the controllability dimension, the individual characterizes the outcome as either controllable like effort or other persons, or uncontrollable like task difficulty, ability, mood, or chance (Schunk et al. 2007). A maladaptive attributional pattern lowering student’s motivation would be if a positive outcome is attributed to external and variable causes and if a negative outcome is attributed to internal and stable causes (Weiner 1985).
https://static-content.springer.com/image/art%3A10.1007%2Fs10212-012-0114-9/MediaObjects/10212_2012_114_Fig1_HTML.gif
Fig. 1

Attributional model of teacher judgment and student motivation

Depending on the ascription of the outcome to a certain cause, different psychological consequences can occur. The attributional model in Fig. 1 proposes three psychological consequences, namely expectancy for success, self-concept, and test anxiety. These consequences are related to teacher’s misjudgment of student achievement (Urhahne et al. 2010, 2011) and to certain attributions. Among them, expectancy for success is directly taken from Weiner’s (2005, 2010) attributional model and was empirically tested by McMahan (1973) and Tapasak (1990). The other two psychological consequences are not included in Weiner’s model, but their connection with causal attributions was supported by different empirical studies. For example, Ames and Felker (1979) found that students with a high self-concept attribute success more to ability and less to luck in comparison to students with a low self-concept. Marsh (1984, 1986a) has shown that students with a higher academic self-concept attribute success more to ability and effort and failure less to these causes. Weiner (2010) used self-esteem instead of self-concept because his focus is more on the emotional component of the self-construct. However, these two concepts are highly related (Marsh 1986b) and sometimes used interchangeably (Ehrlich and DeBruhl 1986). For test anxiety, Weiner (2010) stated that an unfair teacher could cause feelings of hopelessness, which is one of the emotional symptoms of test anxiety. Other studies pointed to the importance of the worry component of test anxiety for attribution research (Bandalos et al. 1995) and found evidences for the relationship between test anxiety and causal attributions (Hunsley 1987; Leppin et al. 1987).

The relationship between teacher judgment and student attributions has not been intensively researched in the past. However, beliefs about justice in the classroom, which are related to teacher judgment, have shown to be of importance for students’ causal attributions (Weiner 2006). Teacher judgment is a frequent event in students’ school life, and misjudgment of student performance due to the fallibility of the human mind and senses is likely. Thus, a mediation model incorporating the relationship between teacher judgment, student attributions, and psychological consequences was hypothesized (see Fig. 1) and tested with two samples from different cultures.

The role of culture

In the field of attribution research, culture has been considered to be an important influencing factor. There are many cross-cultural studies showing significant differences between individualistic and collectivistic cultures on attributions (e.g., Choi et al. 1999; Mezulis et al. 2004). Former research has shown that Americans emphasize ability and effort equally (Weiner 1986), while Chinese give more credit to effort (Chang 1985). However, studies of Urhahne et al. (2010, 2011) with samples from Austria, China, and Germany indicated consistency across cultures by showing that teacher’s underestimation of student performance was constantly associated with a decrease of student motivation. In combination, the consistent results of influences of teacher judgment on student motivation (Urhahne et al. 2010, 2011) and the different attribution tendencies across cultures lead to the hypothesis that in different cultures teacher judgment has similar influence on student motivation, but the mediating attributional processes might vary.

In this research project, special attention is paid to two student groups: underestimated and overestimated students, as identified by the difference between teacher judgment and their actual performance. It should be tested whether teacher’s biased judgment would lead to certain student attribution patterns and what the further psychological and behavioral consequences will be. The effect of teacher judgment on student motivation will be retested with a sample from China, which is thought to be a collectivistic country (e.g., Choi et al. 1999; Mezulis et al. 2004).

Research hypotheses

According to Weiner’s model of interpersonal motivation, it is assumed that student attributions will work as mediators between the relationship of teacher judgment and student motivation. This does not exclude that there would also be direct paths from teacher judgment to student motivation. The research hypotheses of the two studies are the following:
  1. 1.

    Underestimated students show different attribution patterns for success and failure in comparison to overestimated students. They will be more likely to attribute success to external and variable causes and failure to internal and stable causes, which can be regarded as maladaptive.

     
  2. 2.

    Students’ attributions work as mediators between teacher judgment of student performance and psychological as well as behavioral consequences for the students.

     

In the first study, hypotheses 1 and 2 will be tested among a sample of German students. The second study will test hypotheses 1 and 2 among a sample of Chinese students and will extend hypothesis 2 by adding a longitudinal measure of mathematical achievement as an additional dependent variable. Furthermore, it is assumed that different casual ascriptions work as mediators in the two countries. This additional hypothesis will be tested simply by comparing the pattern of results of the first and second study.

Study 1

The first study aimed at testing the attributional model of teacher judgment and student motivation (see Fig. 1).

Method

Sample

The participants were 144 fourth-grade students (78 male) from four elementary schools in Munich, Germany. The students came from eight different classes with 12 to 23 students in each class. The average age of the fourth graders was 9.93 years old (SD = 0.61). All eight classes had female teachers. The teachers were on average 42.13 years old (SD = 10.99) and had considerable teaching experience of an average of 14.38 years (SD = 9.43). In the German school system, elementary school teachers are responsible for teaching different subjects like mathematics, science, music, sports, arts, and German. On average, the teachers instructed the children for 19.63 h per week (SD = 3.11), mainly in German and mathematics.

Measures

Student data

During the course of the study, the fourth graders were asked to work on a standardized mathematics test and a student questionnaire. The materials used are described below in more detail. As elementary school teachers are responsible for different subjects, students’ causal attributions and self-concepts were tested on a general level, assuming that teachers have a broader influence on students’ self-perceptions, which is not necessarily restricted to mathematics.

Mathematics achievement

The German mathematics test for the fourth graders (DEMAT 4; Gölitz et al. 2006) was applied to measure a student’s mathematical achievement. The test is based on the mathematics curriculum for fourth graders in Germany. It consists of different tasks from the areas arithmetic (addition, subtraction, multiplication, and division), algebra (story problems and size comparisons), and geometry (spatial positions and mirror drawings). Two parallel forms of the test were applied to prevent students from cheating. Thereby, adjacent students received different test versions. The test, including an introduction and instructions, takes 45 min to complete. Students could reach a maximum of 36 points on the mathematics achievement test. The internal consistency of the achievement test in this study was given by Cronbach’s α = 0.89.

Attributions

Students’ attributions were tested with a simplified version of a questionnaire developed by Dresel et al. (2005). The original three conditions (accomplish a goal, get a good grade, experience success) were considered as very similar, so that only one related to mathematics grade was kept in the short version. The 12 items of the questionnaire describe six success and six failure situations and their possible causes. The causes stem from six attribution categories, namely ability, effort, mood, other persons, task characteristics, and chance. An example item for a success situation was “When you get a good grade, then it is because… of your abilities. (Examples: You knew the material, you are good in the subject, you understood the material, or you are talented).” In parallel, an example item for one of the six failure situations was “When you get a poor grade, then it is because… you did not put in effort. (Examples: You did not study or studied too little, you did not pay attention, you did not practice, or you did not concentrate).” Research results of Dresel et al. (2005) show that the attribution categories are relatively independent of each other with the highest correlation between categories of less than 0.31.

Expectancy for success

A student’s belief to succeed on a given task is called expectancy for success (Schunk et al. 2007). In our study, expectancy for success was measured in relation to mathematics performance by one item “What do you think: What grade will you get on your next mathematics test?” The item was taken from the Ulm Motivational Test Battery (Ziegler et al. 2005, 2008). Students could answer the question by giving grades from 1 (best possible grade) to 5 (worst possible grade). The item was re-coded for statistical analysis so that higher values indicate a higher expectancy for success.

Test anxiety

In the questionnaire, students were asked about their domain-specific test anxiety with items such as “When I’m called upon in mathematics, I am scared to say something wrong.” Four of the six items on test anxiety were taken from the Ulm Motivational Test Battery, whereas two items stem from PISA 2003 (PISA-Konsortium Deutschland 2006). The items were judged on a four-point Likert scale (1 = not at all true, 2 = somewhat true, 3 = rather true, 4 = totally true). The scale on test anxiety had an internal consistency of Cronbach’s α = 0.79.

Academic self-concept

Academic self-concept is a student’s perception of his/her competencies at school. According to the hierarchical self-concept model of Shavelson et al. (1976), academic self-concept is part of the global self-concept. Academic self-concept was measured with items of Rost and Lamsfuss (1992) like “I’m one of the best in school.” The five items on academic self-concept were judged on a four-point Likert scale and yielded a Cronbach’s α of 0.81.

Teacher data

The elementary school teachers were asked socio-demographic questions including age, gender, work experience, teacher education, and further qualifications before judging the students’ performances. In order to make accurate judgments on test performance, they received a copy of the DEMAT 4 and estimated the correctly solved test items for each student in their class. The difference between teacher judgment and student performance finally led to groups of underestimated and overestimated students.

Procedure

The investigation was conducted in the middle of the second half of the school year and lasted for two lessons. In the first lesson, the standardized mathematics achievement test for fourth graders (DEMAT 4) was given to the students. The students completed the test as mentioned in the instructions. In the second lesson, students completed the motivation questionnaire. An instructor read the items aloud in class and students could ask questions if they had difficulty understanding the content.

The mathematics teacher made a score prediction for each student in the class in the teacher’s room. Usually, the teachers finished this task in 45 min. After some weeks, the teachers received differentiated feedback about class performance in the standardized mathematics achievement test.

Mediation analysis

The method of hierarchical regression analysis and the Freedman–Schatzkin procedure were selected to test for multiple mediation effects (Baron and Kenny 1986; MacKinnon et al. 2002). According to Baron and Kenny (1986), the following conditions have to be met for mediation: (a) The independent variable (teacher judgment) has to correlate with the mediator (attributions) as well as with the dependent variable (psychological or behavioral consequences); (b) the mediator has to be correlated with the dependent variable; (c) when controlling for the effect of the mediator, the previously significant relation between the independent variable and the dependent variable should no longer be significant or will even become zero in case of complete mediation (Baron and Kenny 1986). Teacher judgment will be regressed on student motivation, and a significant regression coefficient should be the outcome. Then, after introducing attributions in the regression model, the effect of teacher judgment on student motivation should be reduced, and a lower regression coefficient should occur. Additionally, the Freedman–Schatzkin procedure was conducted to test the change of the regression coefficient before and after the introduction of attributions on the basis of a t-statistic.

Results

First, the difference between teacher judgment and student performance in the mathematics test was calculated. Elementary school teachers overestimated students’ mathematics performances on average by 3.07 points (SD = 5.43) (t(143) = 6.78, p < 0.001, d = 1.13). Thirty-three students, 16 boys and 17 girls, were underestimated in their test performance by M = 4.30 points (SD = 3.68). Ninety-seven students were overrated in their test performance by M = 6.02 points (SD = 3.25). The remaining 14 accurately estimated students were excluded from the statistical analysis. In reality, underestimated students (M = 20.94, SD = 6.50) and overestimated students (M = 21.02, SD = 5.01) had the same mathematical abilities and did not differ in their test achievement (t(128) = −0.07, ns, d = −0.01).

In the first research hypothesis, it was assumed that underestimated students differ from their overestimated counterparts by showing a maladaptive attributional pattern. The results of several analyses of variance (ANOVAs) in Table 1 reveal that some significant differences in attributional style emerged between under- and overestimated students. Underestimated students attribute success more on variable and external causes like chance but less to stable and internal causes like ability, which can be seen as maladaptive. They also attribute success less on positive mood and blame failure more on negative mood.
Table 1

Univariate analyses of variance for causal attributions of underestimated and overestimated students in the German sample

 

Success

Failure

Underestimated

Overestimated

  

Underestimated

Overestimated

  

Attribution

M

SD

M

SD

F(1, 120)

η2

M

SD

M

SD

F(1, 120)

η2

Mood

2.41

1.09

2.89

1.06

4.47*

0.036

2.73

1.08

2.17

1.04

6.38*

0.051

Others

2.79

1.15

2.32

1.19

3.53

0.029

1.87

1.04

1.92

1.08

0.53

0.000

Task

3.34

0.81

3.18

0.85

0.83

0.007

2.93

0.91

2.67

1.01

1.64

0.013

Effort

3.55

0.87

3.52

0.83

0.04

0.000

2.57

1.14

2.18

1.13

2.58

0.021

Ability

2.69

0.76

3.18

0.87

7.49**

0.059

2.23

0.97

1.93

1.02

2.05

0.017

Chance

2.83

0.85

2.13

1.08

10.06**

0.077

2.37

0.93

2.51

1.16

0.38

0.003

*p < 0.05, **p < 0.01

Furthermore, it was tested whether underestimated students show lower motivation than overestimated students. From Table 2, it can be seen that underestimated students had lower expectancy for success, a lower academic self-concept, and higher test anxiety than overestimated students. This result is remarkable since both student groups performed at the same level on the achievement test. Besides, it becomes obvious from Table 2 that certain attributions qualify as mediators between the relationship of teacher judgment and student motivation. In the case of success, attributions to ability and chance are significantly correlated with both teacher judgment and all psychological consequences. In the case of failure, attributions to negative mood have the potential to function as mediators.
Table 2

Descriptive statistics and correlations of teacher judgment, student attributions, and psychological consequences in the German sample

 

M

SD

Range

Teacher judgment

Expectancy for success

Test anxiety

Self-concept

Teacher judgment

1.75

0.44

1–2

   

Expectancy for success

2.34

0.92

1–5

0.29**

  

Test anxiety

2.27

0.90

1–4

−0.21*

−0.52***

 

Self-concept

2.70

0.74

1–4

0.41***

0.70***

−0.38***

Mood (success)

2.74

1.08

1–4

0.21*

0.09

0.08

0.17

Others (success)

2.44

1.17

1–4

−0.16

−0.08

0.11

−0.18*

Tasks (success)

3.16

0.86

1–4

−0.02

0.03

−0.01

0.18*

Effort (success)

3.52

0.81

1–4

0.02

0.12

−0.01

0.18*

Ability (success)

3.06

0.87

1–4

0.22*

0.48***

−0.19*

0.53***

Chance (success)

2.33

1.09

1–4

−0.26**

−0.21*

0.26**

−0.27**

Mood (failure)

2.32

1.09

1–4

−0.20*

−0.22**

0.52***

−0.19*

Others (failure)

1.96

1.11

1–4

0.02

0.04

−0.07

0.03

Tasks (failure)

2.73

1.01

1–4

−0.13

−0.30***

0.22*

−0.22**

Effort (failure)

2.33

1.15

1–4

−0.16

−0.18*

0.14

−0.14

Ability (failure)

2.00

1.01

1–4

−0.16

−0.34***

0.15

−0.27**

Chance (failure)

2.48

1.10

1–4

0.06

0.02

−0.09

0.15

Teacher judgment coding: 1 underestimated students, 2 overestimated students. Intercorrelations between attributions amount on average to 0.16 for success (range, 0.01–0.36) and to 0.18 for failure (range, 0.01–0.40)

*p < 0.05, **p < 0.01, ***p < 0.001

The second research hypothesis asks of a mediating effect of attributions between teacher judgment and psychological consequences. To answer this question, three hierarchical regression analyses for the dependent variables, expectancy for success, test anxiety, and academic self-concept, were conducted. The results are shown in Table 3. In the first step, teacher judgment was entered into the regression model. In the second step, the possible mediators indicated by correlations were added to the model. For expectancy for success, the formerly significant beta coefficient of teacher judgment becomes insignificant when attributions are taken into account. The effect of teacher judgment on expectancy for success is mediated by the ability attribution of success and mood attribution of failure. For test anxiety, the mood attribution of failure and ability attribution of success emerged as mediators as the standardized regression coefficient of teacher judgment is strongly reduced from step 1 to step 2. For academic self-concept, teacher judgment still has a direct effect on the dependent variable but also has indirect effects via the ability attribution of success and mood attribution of failure. All Freedman–Schatzkin tests in Table 3 are significant and corroborate the hypothesis of a mediating role of causal attributions.
Table 3

Hierarchical multiple regression analyses predicting psychological consequences in the German sample

 

Expectancy for success

Test anxiety

Self-concept

Predictor

B

SE B

ß

R2

B

SE B

ß

R2

B

SE B

ß

R2

Step 1

   

0.07**

   

0.04*

   

0.16***

Teacher judgment

0.53

0.17

0.27**

 

−0.42

0.19

−0.21*

 

0.65

0.14

0.40***

 

Step 2

   

0.21***

   

0.32***

   

0.22***

Teacher judgment

0.25

0.16

0.13

 

−0.09

0.17

−0.05

 

0.41

0.13

0.25**

 

Ability (success)

0.41

0.08

0.42***

 

−0.17

0.09

−0.16

 

0.36

0.07

0.44***

 

Chance (success)

−0.03

0.07

−0.04

 

0.10

0.07

0.12

 

−0.04

0.05

−0.06

 

Mood (failure)

−0.18

0.07

−0.23

 

0.42

0.07

0.51***

 

−0.13

0.05

−0.19*

 

Total R2

   

0.28***

   

0.36***

   

0.38***

Freedman–Schatzkin test

  

t(119) = 4.56***

  

t(119) = 4.87***

  

t(119) = 4.91***

Teacher judgment coding: 1 underestimated students, 2 overestimated students

*p < 0.05, **p < 0.01, ***p < 0.001

Discussion

The study of fourth-grade elementary students in Germany examined the relationship between teacher judgment and student motivation. It turned out that underestimated students in a way have maladaptive attributions and show lower motivation than overestimated students. Underestimated students ascribe success more to chance but less to their own ability and positive mood. Failure, however, is blamed on negative mood such as nervousness. In addition, the motivation of underestimated students is impaired. Underestimated students have a lower expectancy of success, have a lower self-concept, and report more test anxiety than overestimated students. Finally, it was found that causal attributions mediate and, thus, compensate for the effect of a teacher’s judgment on student motivation.

The results of this study support the previous findings of the relationships between the causal dimensions and psychological consequences. Relations were found between the controllability dimension and expectancy for success (McMahan 1973; Meyer 1980; Tapasak 1990), as well as between the locus of causality dimension and academic self-concept (Ames and Felker 1979; Marsh 1984). Test anxiety was highly related to causal attributions on mood. They can be classified like attributions to ability as internal and uncontrollable. Prior research has shown that students with high anxiety attribute failure more to ability (Leppin et al. 1987). In this study, high test-anxious students attribute failure more to mood, but the attribution dimensions involved are the same. However, the attribution to mood in contrast to ability is more optimistic as mood can change more easily.

Moreover, the study broadens the research focus by investigating the antecedent conditions of the attribution process. An additional element, teacher judgment about student achievement, was taken into consideration. It can be supposed that teacher’s judgment is reflected and is shown in the daily interactions between the teacher and students. Students perceive their teacher’s attitude towards them by observing the way they are approached and judged compared with their other classmates. Students’ self-concept is at least partly developed from their teacher’s perception of them. “In the presence of one whom we feel to be of importance, there is a tendency to enter into and adopt, by sympathy, his judgment of ourself” (Cooley 1902, p. 175). For this reason, underestimated students who are judged as weaker by their teachers might display lower expectancy for success, higher test anxiety, and a lower self-concept. However, the influence of teacher’s judgment on expectancy for success and test anxiety disappears for underestimated students who have adaptive attributional patterns, and this influence is attenuated for students’ self-concept. Thus, attributions are an important source for the maintenance of high motivation when teachers are misperceiving students’ achievement.

Study 2

The attributional model of teacher judgment and student motivation was tested again with a sample of Chinese elementary students. The different sample offered the opportunity to investigate cultural differences in the mediating role of attributions. In addition, a retest of students’ mathematics achievement allowed examination of the complete mediation model. Students’ retesting could meaningfully only be done in China as German students change after fourth grade of elementary school to different secondary schools of different achievement levels.

Method

Sample

The participants of the second study were 272 fourth-grade students (150 male) from Beijing, China. The Chinese sample consisted of eight classes with 26 to 46 students in each class. Students’ average age was 9.87 years (SD = .65). In the Chinese educational system, students go to elementary school for the first 6 years and are instructed by specialized teachers for each subject. The eight mathematics teachers (three male and five female) were on average 34.14 years old (SD = 8.55) and had a teaching experience of M = 14.38 years (SD = 6.43). The hours they spent every week with the classes were relatively high. Six teachers instructed the students for 5 h, and two teachers instructed in mathematics for 7 h per week.

Measures

The questionnaires used were nearly the same as in the first study so that only the differences between the studies are reported. All the questionnaires were translated from German into Chinese and back-translated into German by a native Chinese person, who has studied in Germany for several years. The back-translation revealed that the items were linguistically equivalent to the original ones.

Mathematical self-concept

As the Chinese teachers are only responsible for instructing the students in mathematics, the self-concept was measured on a domain-specific level. Five items of the Ulm Motivational Test Battery (Ziegler et al. 2005) were selected, which measured students’ mathematical self-concept with a reliability of Cronbach’s α = 0.81 on a four-point Likert scale. A sample item of the scale is “In mathematics, I understand even the most difficult problems.”

Expectancy for success

The Chinese grading system differs from the German one as students can receive grades from 0 (worst possible grade) to 100 (best possible grade) with 60 equaling passed. The Chinese grading scale was used to measure students’ expectancy for success in the next mathematics examination.

Mathematics achievement

Chinese students were tested twice on their mathematics achievement. The first test was conducted almost at the same time of the first study, and the second test was done 1 year after the first investigation. The first time, Chinese students worked on the DEMAT 4 (Cronbach’s α = 0.92) while, for the second time, 40 testing tasks from the Swiss calculation test for the fifth and sixth grade were selected (Lobeck et al. 1990). The testing tasks indicate students’ understanding of mathematical operations, symbols, relations, units, and quantities. After checking the tasks, Chinese mathematics teachers reassured the researchers that the standardized testing tasks do not demand too much from the Chinese fifth graders. The reliability of the Swiss calculation test was satisfactory with Cronbach’s α = 0.88. Students could reach a maximum of 40 points on the Swiss calculation test.

Procedure

The study in the Chinese classes was conducted in the same way as in Germany. Students completed the mathematics achievement test and the student questionnaire within two lessons under the guidance of an instructor. The teachers worked on the student assessment task while students were being tested and also received feedback about students’ performance on the mathematics test a few weeks after the completion of the study. For the second mathematics achievement test, another lesson was necessary where the students worked on the test items guided by an instructor. The overlap between the two samples was markedly high as 263 out of 272 fourth graders could also be reached in the fifth grade.

Results

Underestimated and overestimated students were identified in the same way as in first study by the difference between teacher judgment and student achievement. The Chinese teachers also showed a tendency to overestimate students’ potential performance. The estimated scores were 1.76 points (SD = 4.55) higher than the actual scores of the students (t(271) = 6.40, p < 0.001, d = 0.78). Eighty-five students, 52 boys and 33 girls, were underestimated by M = 3.46 points (SD = 2.53) whereas 172 students of the overestimated group were overrated by M = 4.50 points (SD = 2.88). Fifteen accurately estimated students were excluded from the statistical analysis. As in the previous study, the actual performance of the two student groups did not differ significantly (t(255) = 1.60, ns, d = 0.20). The average score for the underestimated students was 25.95 points (SD = 5.98) and for the overestimated students was 24.78 points (SD = 5.24).

The first research hypothesis regarding the attributional differences of underestimated and overestimated students was tested by several ANOVAs for success and failure situations. As shown in Table 4, underestimated students attribute success more to chance and failure more to mood, task difficulty, and effort than overestimated students. This pattern could be labeled as maladaptive because underestimated students attribute success to a variable and external cause and failure mainly to uncontrollable causes like mood or task difficulty. However, underestimated students’ stronger tendency of attributing failure to effort has been regarded as somehow adaptive and runs against the expectation.
Table 4

Univariate analyses of variance for causal attributions of underestimated and overestimated students in the Chinese sample

 

Success

Failure

Underestimated

Overestimated

  

Underestimated

Overestimated

  

Attribution

M

SD

M

SD

F(1, 247)

η2

M

SD

M

SD

F(1, 241)

η2

Mood

3.08

1.02

2.91

1.16

1.16

0.005

2.28

1.11

1.96

1.17

4.39*

0.018

Others

1.95

0.98

1.76

0.97

2.00

0.008

1.93

0.97

1.70

0.98

2.78

0.011

Task

3.11

0.96

3.04

0.96

0.35

0.001

2.49

1.05

2.20

1.10

3.88*

0.016

Effort

3.30

0.82

3.32

0.88

0.03

0.000

2.44

1.04

2.10

1.14

5.25*

0.021

Ability

3.03

0.84

3.00

0.96

0.04

0.000

2.02

1.06

1.77

1.00

3.49

0.014

Chance

2.50

1.04

2.06

1.08

9.26**

0.036

2.23

1.09

2.08

1.11

1.06

0.004

*p < 0.05, **p < 0.01

Like in the German sample, the underestimated students had lower expectancy for success, a lower self-concept, and higher test anxiety than the overestimated students as can be seen from the correlations in Table 5. The correlations in Table 5 also indicate the mediating role of attributions in the relationship between teacher judgment and student motivation. For situations of success, the attribution of chance emerged as a potential mediator. For situations of failure, possible mediators are mood, task, effort, and ability.
Table 5

Descriptive statistics and correlations of teacher judgment, student attributions, psychological consequences, and mathematics achievement in the following year in the Chinese sample

 

M

SD

Range

Teacher judgment

Mathematics achievement

Expectancy for success

Test anxiety

Self-concept

Teacher judgment

1.67

0.47

1–2

    

Mathematics achievement

25.02

7.46

2–40

0.16*

   

Expectancy for success

95.15

6.17

60–100

0.20**

0.53***

  

Test anxiety

2.27

0.80

1–4

−0.19**

−0.37***

−0.32***

 

Self-concept

3.18

0.68

1.4–4

0.17**

0.39***

0.32***

−0.59***

Mood (success)

2.95

1.14

1–4

−0.05

0.21**

0.15*

−0.29***

0.31***

Others (success)

1.80

0.96

1–4

−0.08

−0.13*

−0.21**

0.16**

−0.14*

Tasks (success)

3.05

0.98

1–4

−0.05

0.02

0.06

−0.04

0.16**

Effort (success)

3.33

0.86

1–4

0.00

0.10

0.19**

−0.17***

0.27***

Ability (success)

3.01

0.92

1–4

−0.01

0.25***

0.19**

−0.26***

0.45***

Chance (success)

2.17

1.08

1–4

−0.19**

−0.24***

−0.17**

0.24***

−0.22***

Mood (failure)

2.04

1.17

1–4

−0.15*

−0.22***

−0.20**

0.53***

−0.35***

Others (failure)

1.79

1.00

1–4

−0.09

−0.19**

−0.23***

0.21**

−0.17**

Tasks (failure)

2.29

1.10

1–4

−0.13*

−0.16*

−0.14*

0.32***

−0.38***

Effort (failure)

2.20

1.12

1–4

−0.15*

−0.14*

−0.22***

0.26***

−0.29***

Ability (failure)

1.84

1.02

1–4

−0.13*

−0.30***

−0.38***

0.52***

−0.51***

Chance (failure)

2.15

1.10

1–4

−0.06

−0.07

−0.16**

0.22***

−0.09

Teacher judgment coding: 1 underestimated students, 2 overestimated students. Intercorrelations between attributions amount on average to 0.14 for success (range, 0.03–0.31) and to 0.24 for failure (range, 0.03–0.39)

*p < 0.05, **p < 0.01, ***p < 0.001

The second research hypothesis regarding the mediation was tested by hierarchical regression analyses. As shown in Table 6, the effect of teacher judgment on student motivation in all three cases is largely reduced after introducing causal ascriptions into the regression models. In addition, the Freedman–Schatzkin test is getting each time significant. For expectancy for success, teacher judgment is partially mediated by ability attributions of failure. However, the direct effect of teacher judgment on the motivation construct still remains significant. For test anxiety, the strongest mediators are ability and mood attribution of failure, but chance attribution of success and task attribution of failure also become significant. For mathematical self-concept, the mediators are solely failure attributions. Teacher judgment is mediated by attributions on mood, task difficulty, effort, and ability.
Table 6

Hierarchical multiple regression analyses predicting psychological consequences in the Chinese sample

 

Expectancy for success

Test anxiety

Self-concept

Predictor

B

SE B

ß

R2

B

SE B

ß

R2

B

SE B

ß

R2

Step 1

   

0.04**

   

0.03**

   

0.03**

Teacher judgment

2.84

0.85

0.21**

 

−0.29

0.11

−0.17**

 

0.24

0.09

0.17**

 

Step 2

   

0.12***

   

0.39***

   

0.30***

Teacher judgment

1.89

0.83

0.14*

 

−0.07

0.09

−0.04

 

0.07

0.80

0.05

 

Chance (success)

−0.62

0.37

−0.11

 

0.09

0.04

0.13*

 

−0.05

0.04

−0.08

 

Mood (failure)

−0.28

0.36

−0.05

 

0.24

0.04

0.36***

 

−0.09

0.03

−0.15**

 

Task (failure)

0.08

0.37

0.01

 

0.06

0.04

0.09

 

−0.12

0.04

−0.19**

 

Effort (failure)

−0.56

0.37

−0.10

 

0.06

0.04

0.09

 

−0.07

0.04

−0.12*

 

Ability (failure)

−1.51

0.43

−0.24***

 

0.24

0.04

0.31***

 

−0.22

0.04

−0.33***

 

Total R2

   

0.16***

   

0.42***

   

0.33***

Freedman-Schatzkin test

  

t(231) = 4.32***

  

t(231) = 6.68***

  

t(231) = 6.77***

Teacher judgment coding: 1 underestimated students, 2 overestimated students’

*p < 0.05, **p < 0.01, ***p < 0.001

Using the mathematics achievement data of the follow-up test, the complete attribution model of teacher judgment and student motivation (see Fig. 1) could be tested. While underestimated students performed on the same level as overestimated students in the fourth grade, a t test for independent samples reveals that underestimated students in the fifth grade (M = 23.25, SD = 7.99) compared with overestimated students (M = 25.73, SD = 6.96) are significantly weaker in their mathematics achievement (t(247) = −2.53, p < 0.05, d = −0.33). The significant difference between student groups is also reflected in the first step of the hierarchical regression analysis in Table 7 where teacher judgment is regressed on mathematics achievement. In the second step, causal attributions were introduced into the regression model and chance attribution of success and ability attribution of failure emerged as mediators. In the third step, when psychological consequences were also taken into account, the effect of teacher judgment on mathematics achievement completely disappeared. Teacher judgment of student performance was fully mediated by chance attribution of success, expectancy for success, and mathematical self-concept.
Table 7

Hierarchical multiple regression analysis predicting mathematics achievement in the following year in the Chinese sample

 

Mathematics achievement

Predictor

B

SE B

β

ΔR2

Step 1

   

0.02*

 Teacher judgment

2.06

1.01

0.14*

 

Step 2

   

0.12***

 Teacher judgment

0.83

1.00

0.06

 

 Chance (success)

−1.31

0.44

−0.20**

 

 Mood (failure)

−0.50

0.42

−0.08

 

 Task (failure)

−0.17

0.44

−0.03

 

 Effort (failure)

−0.29

0.43

−0.05

 

 Ability (failure)

−1.29

0.51

−0.18*

 

Step 3

   

0.18***

 Teacher judgment

0.04

0.90

0.00

 

 Chance (success)

−0.98

0.40

−0.15*

 

 Mood (failure)

−0.11

0.41

−0.02

 

 Task (failure)

0.08

0.40

0.01

 

 Effort (failure)

0.26

0.40

0.04

 

 Ability (failure)

0.03

0.50

0.00

 

 Expectancy for success

0.52

0.09

0.35***

 

 Test anxiety

−0.44

0.75

−0.05

 

 Self-concept

2.57

0.80

0.25**

 

Total R2

   

0.32***

Freedman–Schatzkin test

  

t(220) = 4.26***

Teacher judgment coding: 1 underestimated students, 2 overestimated students’

*p < 0.05, **p < 0.01, ***p < 0.001

Discussion

The second study shows differences between underestimated and overestimated students in a sample of Chinese elementary school students. The differences are less remarkable in situations of success than in situations of failure, where group differences occur for four of the six measured causes. Chinese underestimated students show the same maladaptive attributions as German underestimated students when they attribute success to chance. Chinese students’ attribution of failure to effort might be due to the credit that Chinese give to effort (Chang 1985), but whether this is adaptive is still worth discussing. Some researchers have considered effort attribution of failure a double-edged sword (Covington and Omelich 1979). Attributing failure to effort could also be maladaptive when underestimated students make efforts, but it does not help them to improve themselves.

Furthermore, Chinese underestimated students showed equal achievement in the fourth grade but lower motivation than Chinese overestimated students. Especially in exam situations, their self-perceptions seem to be barred by a lower domain-specific self-concept accompanied by a lower expectancy for success and more test anxiety. It can be assumed that such an unfavorable motivation pattern has contributed to their inferior mathematics performance in the fifth grade.

Chinese students’ attributions played an essential role in explaining the effects of teacher judgment on student motivation. The effects of teacher judgment on test anxiety and self-concept were nearly completely mediated by students’ causal ascriptions. For expectancy for success, a partial mediation of attributions could be substantiated. Moreover, a test of the complete attribution model revealed that teacher judgment and mathematics achievement are mediated by student attributions and student motivation. The effects, however, are small and shed light on the fact that mathematics achievement is only slightly dependent on the fairness of teacher’s judgment. Teachers’ expectations, on the other hand, indeed have the potential to influence students’ academic performance as research on the Pygmalion effect has impressively demonstrated (Jussim et al. 2009). The study results underline that mediating motivational processes should be taken into consideration to provide a full explanation of teacher expectancy effects (cf. Harris and Rosenthal 1985). In summary, the second study shows that how students perceive their academic success and failure as well as what causes students attribute to their achievements are of vital importance.

General discussion

In two studies in elementary schools, the effects of teacher judgment on student motivation were closely examined, and the hypothesis that the relationship is mediated by student attributions was tested. A common result of both studies is that underestimation of student performance is related to lower expectancy for success, more test anxiety, and a lower self-concept. Underestimated students are less confident about their performance potential than overestimated students, even though both groups achieved at the same level in the standardized mathematics test. A second outcome of both studies is that underestimation of student performance is associated with maladaptive attributional patterns. Underestimated students attribute success more to external causes and failure more to internal causes in comparison to overestimated students. A third result of both studies is that ability and mood attributions play a peculiar role in the explanation of the psychological consequences of teachers’ misperceptions of student performance. In both studies and for most of the psychological consequences, ability and mood attributions have qualified as mediators.

Most of these outcomes were theoretically predicted. Weiner (1985) proposed that the stability dimension is bound to the expectancy for success and that the locus of causality dimension is relevant for promoting the self-concept. The attribution of ability, as a stable and internal causal ascription, is therefore expected to relate to both expectancy for success as well as academic self-concept. Another important mediator, mood attribution of failure, was also observed in both studies. Weiner (1986) stated that the attribution of failure to nervousness or excitement depends upon personal controllability. If the cause of the event is seen as uncontrollable, strong emotions like shame, guilt, or anxiety can occur. For underestimated students, it means that the causal attribution of failure to negative mood causes more negative emotions to occur.

Ability and mood attributions are of importance in understanding the process of how teacher judgment impacts student motivation and point to different intervention options. Unfavorable ability attributions and negative mood are linked with lower expectancy for success, a lower self-concept, and more test anxiety. Students with detrimental ability and mood attributions lack the necessary confidence to become successful. As attributional patterns are malleable and can be changed by attribution retraining (Carr and Borkowski 1989; Haynes et al. 2009), underestimated students can be encouraged to reconsider maladaptive attributions and to adopt more adaptive attributions instead. They can argue on issues such as whether ability is stable or could be developed in order to change their entity view of ability into an incremental view, or whether emotions could be controlled. The support can be given immediately after test results are informed to the students and can even be a part of the explanation of grades. This kind of communication could also allow teachers to learn more about students’ attributions and motivation, which might help to reduce inaccurate judgments. Teachers might inform students that abilities are malleable or that emotional reactions depend upon cognitive evaluations and are under personal control. Changing students’ perceptions of the controllability of the causes could prevent that motivation and performance decrease in view of unfavorable teacher judgments.

A direct cultural comparison was not intended for this study, as the focus was on measuring the consistency of the results across cultures. The effect of teacher judgment on the three psychological consequences and the mediating of attributions can be evaluated as consistent between the two countries. One cultural difference is that, in Germany, under- and overestimated students differ the most in terms of success attributions whereas, in China, they differ the most in terms of failure attributions. The stronger focus on failure in China might have contributed to the attributional differences between under- and overestimated students. More significant, however, is that the attribution model of teacher judgment and student motivation can be empirically supported for both countries.

Both studies were conducted in elementary schools in a real-life learning environment. In contrast to many studies on attribution research, which are experimental studies with fictitious scenarios, these studies capture the real impressions of the teachers and students. Moreover—despite the cultural differences between Germany and China and the use of sometimes general and sometimes domain-specific measures—the studies led to quite similar outcomes. In both cases, underestimated students had lower motivation and worse attribution patterns than overestimated students. Ability and mood attributions appeared in both studies as the main mediators between teacher judgment and student motivation. It also became obvious that motivational consequences can be pursued by behavioral consequences and that the motivation deficit of underestimated students in the fourth grade had a negative effect on their future performance as predicted by Brophy’s (1983) model of expectancy effects. The performance effect now could only be shown for Chinese students and a longitudinal study with German students should be a future goal.

There were also some inconsistencies of the mediating role of attributions between samples. It could be a sign of cultural differences but might also indicate that the assumption of a well-structured attribution schema in the mind of a fourth-grade student is misleading. It should be pointed out that some information is still needed to understand the differences of the attributional patterns. For example, it needs to be clarified how teacher’s judgment is communicated to the students and how the students gain awareness of it. One way could be that underestimated and overestimated students are treated differently, which is reflected in their motivation (Blöte 1995). This process, however, has to be studied in more detail.

This study focused on only one direction, using teacher judgment to predict student characteristics. However, the effect could also be in the opposite direction, and student characteristics can be taken to study their impact on teacher judgment (Dompnier et al. 2006). For subsequent studies, this interaction should be kept mind. Another shortcoming of the study might be seen in a missing comparison between cultures. However, this was not the primary goal of the study. Special focus was set on the relationship between teacher judgment, student attributions, and the motivational variables across cultures. A direct comparison of attribution patterns can be suggested for future studies.

All in all, the two studies elucidate that teacher judgment and student motivation are related to each other. Whether teacher judgment has a positive or negative influence on student motivation depends upon the causal ascriptions they make. This calls for interventions such as attribution retraining, especially for underestimated students, targeting the enhancement of their motivation and performance. A positive attribution style can compensate for teacher’s misjudgment and enhances the likelihood of high student motivation in school.

Copyright information

© Instituto Superior de Psicologia Aplicada, Lisboa, Portugal and Springer Science+Business Media BV 2012