Learning and Individual Differences 21 (2011) 728–735 Contents lists available at SciVerse ScienceDirect Learning and Individual Differences journal homepage: www.elsevier.com/locate/lindif Metacognitive beliefs, self-confidence and primary learning environment of sixth grade students Sabina Kleitman ⁎, Jennifer Gibson 1 School of Psychology, The University of Sydney, Australia a r t i c l e i n f o Article history: Received 20 December 2010 Received in revised form 17 August 2011 Accepted 20 August 2011 Keywords: Metacognition Self-confidence Confidence bias Metacognitive beliefs Self-efficacy Self-handicapping Achievement Learning environment a b s t r a c t Metacognition is an integral component of a self-regulated approach to learning. The present study examined the relationships between academic self-efficacy and perceptions of one's own competence in memory and reasoning abilities, and their role in predicting the Self-confidence trait. The study also aimed to determine the role of key classroom factors (goal orientation and self-efficacy with the teacher) in predicting selfbeliefs, the Self-confidence trait and academic achievement in Year 6 students (N = 177). EFA and Path analysis were used to determine these relationships. The hypothesised path model was tested in a simultaneous fashion of the entire system of variables to determine whether or not hypothesised relationships were consistent with data. The results suggest that academic self-efficacy and metacognitive competency beliefs define a broad factor—Metacognitive Beliefs—which serves as a key predictor of Self-confidence. Mastery goal-orientation and self-efficacy with teacher predicted Metacognitive Beliefs and, indirectly, Selfconfidence. Students with stronger Metacognitive Beliefs were less engaged in self-handicapping behaviours. Known common factors—intelligence, gender and a proxy for SES, school fees—were controlled for. The present study has important implications for both metacognitive theory and educational practice. Crown Copyright © 2011 Published by Elsevier Inc. All rights reserved. 1. Introduction Metacognition refers to the executive processes involved in reflecting on one's own thinking—‘thinking about thinking’ (Flavell, 1979). Metacognition, along with cognition and motivation, is one of the three fundamentals of self-regulated learning (Efklides, 2006; Schraw, Crippen, & Hartley, 2006). It enables students to navigate effectively and to take control of the learning process (Paris & Winograd, 1990; Stankov & Lee, 2008). Metacognition does not occur in isolation— there are contextual factors that contribute to its development and influence the extent of its use (Garner, 1990; Pressley & Gaskins, 2006). The present study aims to determine a link between key aspects of the typical classroom environment, such as goal-orientation and selfefficacy with a teacher, and metacognitive beliefs and self-confidence in children. Most theories distinguish between knowledge about cognition and metacognitive experiences (Brown, 1987; Efklides, 2008; Nelson & Narens, 1994; Schraw & Dennison, 1994; Veenman & Elshout, 1999). Knowledge of cognition refers to the information one has about tasks, as well as strategies, goals and beliefs that one has about one's own thinking (Flavell, 1979). A subset of such beliefs, ⁎ Corresponding author at: School of Psychology, The University of Sydney, Sydney, NSW, 2006, Australia. E-mail address: sabinak@psych.usyd.edu.au (S. Kleitman). 1 The study reported in this manuscript was conducted at the University of Sydney. The data can be obtained by writing to the first author. namely academic self-efficacy (individual's judgments of his or her competence to perform academic tasks; Schunk, 1991) and perceptions of one's own competence in memory and reasoning abilities (Kleitman & Stankov, 2007), is one focus of this research. Both academic self-efficacy and metacognitive self-concept are measures of beliefs that reflect knowledge of oneself as a learner, beliefs inherently linked to the broad ‘knowledge of cognition’ aspect of metacognition. Thus, we expect (Hypothesis 1) that these beliefs will converge to define a broad construct—metacognitive beliefs (see also Kleitman, 2008, for a review). Metacognitive experiences comprise judgments, feelings and thoughts people make during on-task performance (Efklides, 2001, 2006; Flavell, 1979). They include the feeling of confidence that is another focus of this research. In this research, we used immediate online markers of confidence level that students assigned to their performance (Efklides, 2001, 2006; Flavell, 1979) using three aptitude tests: mathematics, vocabulary, and reading. Immediately after answering a question, participants are asked to rate how confident they feel that they have answered a question correctly (see Fig. 1). In adults, these on-task confidence judgments have high internal consistency (Kleitman & Stankov, 2007) and robust test–retest estimates (Jonsson & Allwood, 2003). There is strong evidence showing individual differences in confidence ratings. That is, relationships between confidence ratings from a broad battery of cognitive tests reflecting diverse cognitive abilities (e.g., Gf, Gc, Gv, general opinions) are consistent and high enough to define a strong Self-confidence factor which is meaningfully related to, yet is independent of intelligence, relevant accuracy of 1041-6080/$ – see front matter. Crown Copyright © 2011 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.lindif.2011.08.003 S. Kleitman, J. Gibson / Learning and Individual Differences 21 (2011) 728–735 729 negative predictions for (d) self-handicapping (we outline a model below and in Fig. 2; we label our predictions as ‘a’–‘d’ accordingly). 1.2. Predictors Fig. 1. Confidence rating scale. Note: the participants could draw the line anywhere in the unshaded section of the rectangle. A mark near 25% indicated that they felt absolutely unsure, or guessed, and a mark closer to the other end of the rectangle (100%) indicated that they were absolutely sure of their answer. performance and personality factors when factor analytic methods are used (see Pallier et al., 2002; Stankov, 1999; Stankov & Kleitman, 2008 for reviews).2 Kleitman and Moscrop (2010) investigated the generality of the Self-confidence factor across different achievement domains (Maths and Vocabulary) and Raven's Progressive Matrices (RPM) Test with children in Grades 4 and 6 (N = 197). As with adults, reliability coefficients for confidence ratings for these tests were high (.84–.96). The results of the CFA determined that a distinct general Self-confidence factor emerged, exclusively defined by the high loadings of the confidence scores from all three cognitive tests. Thus, Self-confidence exists as a stable construct both within a test and across cognitive tests, in children as young as nine years of age. The Self-confidence factor is pertinent to test-taking behaviour and learning outcomes, even after controlling for intelligence (Kleitman & Moscrop, 2010) and other key factors (Stankov & Lee, 2008). That is, Self-confidence, although relating positively to intelligence, still makes a unique contribution to the prediction of standardised school grades, in addition to the child's fluid intelligence, gender, age and family dynamics (Kleitman & Moscrop, 2010). In adults, a small, yet statistically significant incremental validity of the confidence scores was apparent above and beyond the relevant accuracy scores in predicting performance on the TOEFL test (Stankov & Lee, 2008). These findings support a ‘mixed model’, which posits a predictive validity of metacognitive factors on school and learning achievement that is incremental to intelligence (Van der Stel & Veenman, 2010). 1.1. Outcome measures: different aspects of academic achievements In the present investigation, we bring together different strands of metacognitive, educational and differential psychology research, that have not yet been considered in combination, to enhance our understanding of two key metacognitive features—metacognitive beliefs and self-confidence—within the learning environment. In addition, we will also consider two other well-established measures of academic achievement, accuracy of performance on aptitude test (reading comprehension, vocabulary and mathematics), and the reported tendencies towards self-handicapping. These purposeful tendencies involve ‘creating impediments to successful performance on tasks that the individual considers important’ (Urdan, 2004, p. 251). Such behaviours include procrastination or staying up late before an exam. Most of our hypotheses will be explicitly formulated with respect to the two novel outcome measures of (a) metacognitive beliefs and (b) self-confidence. However, considering the nature of the constructs under investigation and research outcomes already available in the literature, similar hypotheses are assumed for the other two outcome variables—we expect positive predictions for (c) accuracy of academic performance (e.g., Kleitman & Moscrop, 2010), and 2 We will use the term confidence judgments/ratings/levels to index on-line confidence scores that people assign to indicate how confident they feel about their answers. These confidence ratings for all test items are averaged to give an overall confidence score on a test. We will refer to the Self-confidence construct to index a broad psychological trait that cuts across diverse cognitive domains. 1.2.1. Metacognitive beliefs There are strong theoretical links between metacognitive knowledge and metacognitive experiences (e.g., Efklides, 2008). Empirical results support this suggestion: the metacognitive beliefs and specific academic self-concept/-efficacy judgments predict a significant proportion of variance for self-confidence, after controlling for accuracy of performance (Efklides & Tsiora, 2002; Kleitman & Stankov, 2007; Kröner & Biermann, 2007). Similarly, metacognitive beliefs positively predict accuracy of performance on the different cognitive tests (Kleitman, 2008). Although never examined, metacognitive beliefs should provide a buffer against self-handicapping tendencies. Selfhandicapping strategies are purposeful and they are deliberately used to avoid the appearance of incompetence, as they enable students to deflect attention from their ability if they perform poorly (Urdan, 2004). Thus, children with high levels of self-beliefs should be less likely to self-handicap. Accordingly, we expect that higher metacognitive beliefs will correspond to higher levels of: b) confidence; c) accuracy of academic performance; and lower levels of d) self-handicapping tendencies, incrementally to other factors (Hypothesis 2 b, c, d; see Fig. 2 below). 1.2.2. Early social environment One important external influence in children is their early social environment, including their relationships with parents and teachers. The theoretical origins for this view stem from Vygotsky (1978), who proposed that development occurs through the gradual internalisation of social activities. Thus, children develop the capacity for self-regulation through interactions with others who are more knowledgeable than themselves. These others act as ‘metacognitive mentors’ (Stright, Neitzel, Sears, & Hoke-Sinex, 2001, p. 456) who initially provide metacognitive information for children, until the child can take responsibility for his/ her own executive functions. 1.2.2.1. Parents. A nurturing caregiver–child relationship allows the child to explore their world, leading to the development of internal working models that ‘promote exploratory competence, self-effectance, and self-esteem’ (Moss & St-Laurent, 2001, p. 863). In adults, retrospective reports of high levels of maternal overprotection during childhood negatively predicted Self-confidence (Want & Kleitman, 2006). In children, high levels of maternal care predict higher levels of Self-confidence (Kleitman & Moscrop, 2010). 1.2.2.2. Teachers. Similarly, a positive relationship with teachers has a positive influence on the learning habits and academic aspirations of children (Burchinal, Peisner-Feinberg, Pianta, & Howes, 2002; Murdock & Miller, 2003; Wentzel, 2002). Although the nature of the relational bond is different between parent–child and teacher–child interactions, the essence of the relationship is the same: caring, closeness, warmth and open communication (e.g. Crosnoe, Burchinal, Keating, Friedman, & Clarke-Stewart, 2010; Crosnoe, Johnson, & Elder, 2004). Self-efficacy beliefs with teacher refer to how competent a student feels about communicating with and relating to their teacher (Schunk, 1989). In Hypothesis 3 (a, b, c, d), we expect that feelings of competence in relating to the teacher will be positively related to three of the outcome measures—a) metacognitive beliefs, b) self-confidence, c) accuracy of academic performance, and negatively to d) self-handicapping (see Fig. 2 below). 1.2.3. Perceived classroom environment Achievement goal theory defines goal orientation as the student's perception of their own classroom environment, as a single learning environment does not provide a common experience for all students 730 S. Kleitman, J. Gibson / Learning and Individual Differences 21 (2011) 728–735 Achievements Perceptions of Learning Environment H3 Self Efficacy: Teacher H4 IQ a. Metacognitive Beliefs - + Mastery GO Control Variables H5 H2 + b. Self-confidence + - Gender School Fees + c. Achievement - + d. Self-Handicapping H6 Fig. 2. Anticipated relationships among variables in the theoretical model. H = Hypothesis, with H2–H6 = graphical representation of Hypotheses 2 to 6. All predictions associated with Metacognitive Beliefs are marked ‘a’. All predictions associated with Self-confidence are marked ‘b’. All predictions associated with Achievement are marked ‘c’. All predictions associated with Self-Handicapping are marked ‘d’. All predictions associated with variables marked ‘a’, ‘b’, and ‘c’ are expected to be positive, while we expect all predictions associated with self-handicapping (marked ‘d’) to be negative. (Ames, 1992). Mastery-oriented classrooms encourage the attribution that effort leads to success and emphasise developing new skills, improving competence, reflect a higher preference for challenging work (Ames & Archer, 1988; Nicholls, 1984) and tendencies to modify strategy use (Garner, 1990), thus possibly fostering metacognitive skills. We expect (Hypothesis 4a, b, c, d) that perceptions of mastery goal-orientation will positively predict three outcome variables—a) metacognitive beliefs, b) self-confidence, c) accuracy of academic performance, and negatively predict d) self-handicapping tendencies (see Fig. 2 below). 3 1.2.4. Other variables 1.2.4.1. Intelligence. One of the established predictors of academic performance and self-confidence is intelligence (Kleitman & Moscrop, 2010). Thus, in the present investigation, intelligence is considered in two ways. Firstly, it is considered as an important positive predictor of academic outcomes and a negative predictor of self-handicapping tendencies (Hypothesis 5a, b, c and d). Secondly, it is considered as a common factor which needs to be controlled for, when the influences of other factors on self-confidence and academic achievements are examined (Kleitman & Stankov, 2007). 1.2.4.2. Gender. While some researchers have found no apparent gender differences in self-confidence (Kleitman & Moscrop, 2010; Stankov, 1999), other researchers have found a tendency towards higher confidence in males (Cooke-Simpson & Voyer, 2006; Pallier, 2003). We included gender as a possible predictor and a control variable, allowing for results examining predictive relationships between variables of interest to be interpreted irrespective of gender. 3 Our original model and results also included performance goal orientation, with expectations that it would have negative or no relationships with the outcome variables. The results supported the latter (no relationships), thus for the sake of simplicity, this variable is omitted from all further discussion. The original model is available from the first authors of this paper. 1.3. The present study Our theoretical model takes into account a number of variables which can be classified into three broad clusters: (1) students' perception of the learning environment (perceptions of competency to communicate with teachers and perceived goal-orientation of a classroom); (2) control variables (intelligence, gender and a proxy of the family's SES, school fees); (3) different aspects of academic achievement (a. students' academic and metacognitive beliefs, b. self-confidence observed through on-task performance on aptitude tests, c. accuracy of performance on aptitude tests, and d. self-handicapping tendencies). While the first two clusters of variables constitute our set of predictors, the last cluster is comprised of the outcome variables. The model is specified in Fig. 2. In this model, we examine predictive relations between key classroom factors and the four academic outcomes, while controlling for essential student characteristics. We will also consider the relationships between the four outcome measures. An important contribution of this study is to extend available research by including metacognitive variables, namely the trait of Self-confidence, and Metacognitive Beliefs, while examining the relationships among well-established constructs in individual differences and educational research. 1.4. Aims and hypotheses 4 This study has three broad aims: (1) examine the generality of academic self-efficacy and metacognitive beliefs variables, presumed to capture a Metacognitive Beliefs construct (Hypothesis 1); (2) determine the role of these beliefs in predicting self-confidence, academic performance and self-handicapping tendencies (Hypothesis 2b, c, and d); and (3) establish the role of key classroom factors (perceived mastery goal orientation and self-efficacy with the teacher) in predicting metacognitive beliefs, self-confidence levels, achievements and self-handicapping tendencies (Hypotheses 3 and 4a, b, c, and d). We also expect that our key control variable—intelligence—will predict four outcome variables (Hypothesis 5a, b, c, and d). All 4 Hypotheses are formulated within the relevant literature review sections of this report. S. Kleitman, J. Gibson / Learning and Individual Differences 21 (2011) 728–735 predictions marked ‘a’, ‘b’, and ‘c’ are expected to be positive, while we expect all predictions marked ‘d’ to be negative. Finally, we expect higher levels of self-handicapping tendencies to have a negative relationship with academic performance (Hypothesis 6). 5 1.5. Statistical analyses To determine the generality of the Self-beliefs factor, an Exploratory Factor Analysis (EFA, PFA method) was conducted using SPSS v17. To examine hypothesised predictive relationships (both direct and indirect) between specified variables of interest (exogenous and endogenous, and between endogenous variables), while controlling for the ‘effects’ of known common cases, path analysis Maximum Likelihood (ML method) was conducted using AMOS v7 (Arbuckle, 2006). 2. Method 2.1. Participants 177 mainstream Year 6 students participated in this study (88 females, Age mean = 11.73 years, SD = 4.68 months). 6 Due to the range restriction, age was excluded from all following analyses. The sample comprised students from three Catholic schools in the Parramatta Diocese and two Independent schools (1 low and 1 high fee-paying) in the Western region of Sydney, Australia. These schools agreed to take part in this research after receiving a recruitment letter. Participating children received lollies on the days of testing as a token form of compensation. Ethics approvals were gained from the Ethics Committees of the University of Sydney, the State Education Research Approval Process and the Catholic Education Office in the Parramatta Diocese, Sydney. 2.2. Measures 731 3. Results 3.1. Descriptive statistics, reliabilities and correlations Descriptive statistics and reliability estimates (Cronbach's alpha) for self-report measures employed are outlined in Tables 1, 2 and 4. While the mean for the Memory component of MARCI is somewhat higher than the mean reported for adults (e.g., 3.87; Kleitman & Stankov, 2007), the mean for the Reasoning component is similar (e.g., 4.16; Kleitman & Stankov, 2007). The high reliability coefficients for both components (αs = .85) are similar to those found in an adult sample (αs = .88), suggesting that the scale is appropriate to be used with the Grade 6 children. Descriptive statistics for the other measures were comparable to those in other studies and their brief description is provided in Table 1. 3.1.1. Metacognitive beliefs factor There were consistent positive correlations between the two components of the MARCI and academic self-efficacy (see Table 2). Importantly, supporting Hypothesis 1, the results of the EFA (PFA method) suggest the existence of one broad factor which captures these beliefs (see Table 2). 3.2. Correlations between different measures employed Descriptive statistics and Pearson correlation coefficients for accuracy and confidence measures employed are outlined in Table 3. Strong reliability estimates and adequate spread of confidence and accuracy scores on the PAT test suggests that the test procedure was adequate to capture accuracy and confidence levels of 11yo children. The differences between confidence and accuracy scores (bias scores) are presented and described in Fig. 3. A pattern of correlations is typical for these types of measures8 (see Table 4, see also Kleitman & Stankov, 2007). Pearson correlation coefficients for all measures employed, including the estimated composites, are outlined in Table 4. As these correlations are best summarised by the forthcoming path model, we omit their description here. A list of measures and relevant psychometrics is outlined in Table 1. 3.3. Path analysis 9 2.3. Procedure Testing was conducted over three separate sessions (45 min each), at each school. All of the tests were group-administered using three booklets containing the relevant instruments and the unique identification number (groups of 25–56 students).7 The first day of testing consisted of the self-efficacy questionnaires, both of the PAT-R tests and the selfhandicapping questionnaire. In the second session, the MARCI was administered first, followed by the teaching style and goal-orientation questionnaires, and the PAT Maths test. The third session consisted of the intelligence tests. This order was utilised to ensure that students' beliefs in their academic and metacognitive competencies were not influenced by subsequent cognitive/aptitude testing. Standardised instructions were given to all participants. 5 This prediction reflects previous research which has identified self-handicapping strategies as a negative predictor of academic achievement (Thomas & Gadbois, 2007; Urdan, 2004). Thus, we expect to replicate this relationship (see Fig. 2). 6 Initial sample consisted of 190 students. However, thirteen students were excluded on account of having a high number of missing values (about 30%); thus, the total N was reduced to 177. Any other missing data (b5%) within tests was imputed using the Expectation Maximisation method. 7 Children were tested in their typical classrooms and/or in designated school areas, such as school halls. Full details of the testing protocol are available on request. The independent (exogenous) variables were mastery classroom goal-orientation, self-efficacy with the teacher, intelligence, gender and school fees variables. The dependent (endogenous) variables were metacognitive beliefs, self-confidence, achievement and selfhandicapping (see Figs. 2 and 4). Consistent with our hypotheses, we also investigated the relationships between endogenous variables: metacognitive beliefs were modelled to predict self-confidence, achievement and self-handicapping scores (Hypothesis 2b, c, d) and self-handicapping was modelled to predict academic achievement (Hypothesis 6). Consistent with these and previous results, Selfconfidence and Achievement scores (resulting from performance on the same cognitive tests) were permitted to correlate (r = .45). First, correlations between independent variables that were not statistically significant (pN .05) were fixed to zero. Then, the relationships between independent and dependent variables were determined. All 8 That is, there are clusters of consistent (p b .01) positive correlations between the accuracy scores from achievement and IQ tests (.37–.70), and between confidence scores (.52–.79). There are also positive correlations between accuracy and confidence scores, and their magnitude is higher for the same test (.49–.61, p b .01), and is lessened otherwise (ranging between .18, p b .05 and .55, p b .01). 9 The purpose of this study was not to investigate causality or reciprocal relationships. The words ‘effect’ and ‘influence’ here are used only for the sake of simplicity, and refer only to the predictive nature of the relationships between the constructs. 732 S. Kleitman, J. Gibson / Learning and Individual Differences 21 (2011) 728–735 Table 1 A list of measures used in this study and relevant psychometrics. List of measures Results from this and other studies Academic Achievements 1. Academic Self-efficacy was measured using six items from PALS (Midgley et al., 1995); The academic self-efficacy scale was sufficiently reliable in this study (α = .78), e.g., ‘I can do even the hardest schoolwork if I try.’ Items were measured on a 5 point consistent with the results found in other samples (Midgley et al., 1995). A distribution of scores in this study (M = 2.70, range is between 1.50 and 3.83) is also Likert scale ranging from 1 = False and 5 = True. consistent with that found in previous investigations (e.g., Midgley et al., 1995). 2. Memory and Reasoning Competence Inventory (MARCI; Kleitman & Stankov, 2007). The In adults, the MARCI has high internal consistency: αs N .80 for both components questionnaire consists of 16 items: eight items assessing beliefs about memory and (Kleitman & Stankov, 2007). The high reliability coefficients evident in this study for reasoning facets respectively, for example ‘My memory is above average’ and ‘I can both components (both αs = .85) are similar to those found in an adult sample. Both reason better than the average person.’ Responses were measured on a 6-point Likert scales also had good distribution of the scores: Ms = 4.58 and 4.29, ranges are scale with response options ranging from 1 = False to 6 = True. between 1.38 and 6 and 1.25 and 6 for the Memory and Reasoning composites respectively. This suggests that the MARCI is suitable for use with children of this age group. 3. Confidence Judgments. Confidence ratings were incorporated into the three achievement Distribution of the confidence scores within each test was good (see Table 3). The tests, thus producing three markers of self-confidence. The procedure and instructions distribution for the overall Self-confidence composite was excellent (M = 78.06, were based on the ‘line scale’ of Allwood et al. (2006; Fig. 1). Students were instructed to range is between 38.91 and 99.67), suggesting adequacy of the scale to capture the draw a vertical line on a 10 cm horizontal rectangle to indicate how confident they were feeling of confidence in children. The high reliability estimates for confidence ratings in Reading, Vocabulary and Mathematics revealed in this study (α = .94, .95 and .90 that their answer was correct. respectively) correspond to their consistency found in adults (e.g., Kleitman & Stankov, 2001, 2007; Pallier et al., 2002; Stankov, 1999; Stankov & Lee, 2008). They also provide clear support for the finding of Kleitman and Moscrop (2010) that, in children, confidence ratings have internal-consistency, just as they do in adults. 4. Achievement Scores. Progressive Achievement Test in Reading (PAT-R; ACER, 2001) and Distribution of the accuracy scores within each test was good (see Table 3). The Mathematics (PAT-Maths; ACER, 2005). Reading comprehension and Vocabulary distribution for the overall Achievement composite was excellent (M = 62.92, range components of the PAT-R were used to measure reading achievement. Three comprehen- is between 21.92 and 97.78). Good reliability estimates for Reading, Vocabulary and sion passages with 14 questions and 19 vocabulary items were administered. Mathematics Mathematics accuracy scores evident in this study (α = .70, .75 and .71 respectively), were congruent to the consistency of the accuracy scores found in other samples achievement was assessed by 15 items measuring numerical skills. (above .80; ACER, 2001, 2005). The overall Achievement composite had a good reliability estimate of .77. 5. Self-handicapping tendencies: 6 items (Midgley et al., 1996); e.g., ‘Some students waste The high reliability of the Self-handicapping (SH) Questionnaire (α = .87) and detime the night before a test so that if they don't do well they can say that is the reason. scriptive statistics evident in this study (M = 1.99, range between 1 and 4.33) correspond to the previous research of Midgley et al. (1996). How true is this for you?’ (response options 1 = False to 5 = True). Students' perception of the learning environment 6. Self-efficacy with teacher: 4 items from Patrick et al. (1997; e.g., ‘I can explain my point The reported Cronbach's α for these five items was .84 (Walker, 2008). In this study, of view to my teacher’ and 1 item from Bandura (1986): ‘I can get my teacher to help however, this measure had a somewhat low, yet still acceptable reliability estimate of .65. The mean of reported self-efficacy with teacher revealed in this study (M = 3.24, me when I have problems with other students.’ (as used in Walker, 2008). range between 1.8 and 4.6) corresponds closely to the previous research (Walker, 2008). 7. Mastery orientation: 6 items from the Patterns of Adaptive Learning Survey (PALS, Midgley The mean score obtained in this study (4.19, with a range of 1–5), is somewhat higher than those found in other samples (Walker, 2008), indicating that students from the et al., 1995). E.g., ‘My teacher makes sure students understand the work.’ present sample perceived high mastery orientation in their classrooms. The reliability estimate of the scale in this study was high (α = .84) and consistent with the other studies (Roeser et al., 1996). Control Variables 8. Intelligence: Verbal Comprehension (acquired knowledge; Gc) and Concept Formation (fluid reasoning; Gf) subtests of the Woodcock Johnson III Brief Intellectual Ability (Woodcock et al., 2001). Items were administered from the age appropriate starting points. A ceiling was set at 2 SDs above average. Although typically administered individually, due to sampling constraints, it was group-administered using standardised instructions and a PowerPoint presentation. A pilot study was conducted prior to testing to ensure that this method was appropriate. possible paths between exogenous and endogenous were built into the model, making it redundant to test any alternative models. This was done to ensure that the effects of each variable on self-confidence, achievement and self-handicapping were calculated, whilst statistically controlling for known common causes (intelligence, gender and school fees). The focus is on the discussion of significant direct effects. As path analysis also allows for calculation and interpretation of indirect effects, this is the best statistical model to investigate a complex network of relationships occurring within a school learning environment. For the sake of clarity, only significant regression coefficients (p b .05) are displayed in Fig. 4. 3.3.1. Direct effects (see Fig. 4) As predicted (Hypotheses 3–5a) teacher's self-efficacy, perceived mastery goal-orientation, and intelligence each had positive direct effects on metacognitive beliefs. Moreover, as expected (Hypothesis 2b, d), metacognitive beliefs positively predicted self-confidence and Given differences in the presentation of this test, it is difficult to compare the results obtained to the results of the other studies. However, there was a good distribution of scores and the alpha reliability of the total score received in this study was high (.91), consistent with the reported reliabilities of .95 (Mather & Woodcock, 2001). Table 2 Reliability estimates, descriptive statistics, correlations and results from the EFA for academic self-efficacy and metacognitive beliefs. Metacognitive beliefs 1 Academic self-efficacy 2 Memory abilities 3 Reasoning abilities Alpha Mean SD 2 .78 .85 .85 2.70 4.58 4.29 .44 .81 .84 .46⁎⁎ 1 3 .50⁎⁎ .67⁎⁎ 1 % of variance Factor loadings h2 .58 .79 .85 56.37% .34 .63 .73 Academic Self-efficacy, Memory and Reasoning abilities measures were formed as a mean of the relevant responses. Given the results of EFA, a mean composite of three scores—labelled Metacognitive beliefs—was formed and used in the following analyses, similar to that in Kleitman and Stankov (2007). ⁎ p b .05. ⁎⁎ p b .01. 733 S. Kleitman, J. Gibson / Learning and Individual Differences 21 (2011) 728–735 Table 3 Reliability estimates, descriptive statistics and correlations for self-confidence and accuracy scores. Alpha Mean (Range) SD 2 Self-confidence 1 Reading 2 Vocabulary 3 Mathematics .94 .95 .90 73.80(28–100) 73.99 (24.6–100) 86.38 (26.9–100) 17.03 17.59 13.59 1 Achievement 4 Reading 5 Vocabulary 6 Mathematics .70 .75 .71 62.58 (14.3–100) 51.84 (10.5–100) 74.34(6.7–100) 20.02 19.84 18.15 Intelligence 7 Gc 8 Gf 9 g(IQ) .71 .92 .91 20.56 (11–28) 24.18 (0–34) 44.73 (11–61) 3 .79⁎⁎ .56⁎⁎ .52⁎⁎ 1 4 .52⁎⁎ .35⁎⁎ .30 1 3.45 7.17 9.47 5 6 7 8 9 .55⁎⁎ .49⁎⁎ .37⁎⁎ .32⁎⁎ .18⁎ .61⁎⁎ .53⁎⁎ .35⁎⁎ .46⁎⁎ .18⁎ .06 .42⁎⁎ .33⁎⁎ .17⁎ .49⁎⁎ .69⁎⁎ 1 .42⁎⁎ .47⁎⁎ 1 .59⁎⁎ .70⁎⁎ .57⁎⁎ .37⁎⁎ .43⁎⁎ .60⁎⁎ .49⁎⁎ .58⁎⁎ .66⁎⁎ .54⁎⁎ .77⁎⁎ .95⁎⁎ 1 1 1 Gc = Crystallised Intelligence, Gf = Fluid Intelligence, g = general intelligence composite. Relevant Self-confidence and Achievement scores were formed as an average of the relevant responses (presented as percentages out of 100). These composites are based on a robust pattern of the relevant positive correlations (the results of EFA, available from the first author of the ms) and the results from numerous previous studies (e.g., see Kleitman & Moscrop, 2010; Stankov & Kleitman, 2008 for reviews). Intelligence scores were formed as a sum of the relevant responses (Woodcock et al., 2001). Given the high correlation between Gf and Gc measures (.54), they were combined into a singlecomposite, g or IQ. ⁎ p b .05. ⁎⁎ p b .01. negatively predicted self-handicapping tendencies. Consistent with Hypothesis 5b, c, d intelligence positively predicted confidence and achievement composites and negatively self-handicapping tendencies. Gender and school fees each had a small, marginally significant positive prediction on confidence, such that boys exhibited greater confidence than girls, and children attending more expensive schools had marginally higher confidence levels. Also, consistent with Hypothesis 6, students who tended to self-handicap performed more poorly on achievement tests than those who did not. All these effects were significant after controlling for the other variables in the model. Bias_Reading Bias_Voc Bias_Maths 30.00 Error bars: 95% Cl Mean 20.00 10.00 0.00 f m SEX Fig. 3. Mean bias and error bar (95%) scores for reading, vocabulary and mathematics tests for female and male participants. Note: Bias Score (in percentages) = Mean Confidence − Accuracy on the relevant test. Positive bias scores reflect degrees of overconfidence, while negative bias scores indicate degrees of underconfidence. Bias scores close to zero indicate good correspondence between confidence and accuracy of performance. F = Females, M = Males. There were no statistical differences in bias scores between the two genders. Overall, three bias scores for the three aptitude tests demonstrate different degrees of overconfidence, with the largest overconfidence present for the Vocabulary test (Mean Bias = 22.15, SD = 19.02), followed by 12.04% of overconfidence (SD = 14.06) for the test of Mathematics and 11.21% of overconfidence (SD = 18.32) for the Reading test. These estimates are considerably larger than bias scores in an adult population for a selection of not identical, but similar cognitive tests (e.g., Kleitman & Stankov, 2007; Stankov & Lee, 2008) indicating that children tend to be more overconfident than adults; however, this statement requires further exploration. The model had an excellent fit and it accounted for notable amount of variance in outcome variables (see Fig. 4). 3.3.2. Indirect and total effects Partially supporting Hypotheses 3–4b, c, d, mastery orientation, and self-efficacy beliefs with teacher (as well as IQ) hold a number of meaningful indirect effects on self-confidence and self-handicapping via its significant effects with other variables in the model—for example, their positive relationship with metacognitive beliefs (see Fig. 2). Metacognitive beliefs also had a positive indirect effect on academic performance, via its relationships with self-confidence. 4. Discussion Although encountering some important limitations (small sample size, a small selection of variables which index different constructs as well as control variables), this study carries with it a number of important implications for educational and metacognitive theory and practice. Theoretically, it has created an avenue for new directions into metacognition research, by examining individual differences in metacognitive knowledge (reflected by metacognitive beliefs of academic and cognitive competence) and metacognitive experiences (reflected by the confidence ratings), mapping the key connections between these constructs and their role in learning. Metacognition is posited to be a key factor in self-regulated learning (Schraw et al., 2006). The results of this study support this claim, demonstrating that a broad set of beliefs (which encompass students' perceptions of competency of their cognitive abilities and academic self-efficacy) converged to define the Metacognitive Beliefs factor, which acted as an important positive predictor of Self-confidence and achievement. Moreover, stronger metacognitive beliefs were linked to reports of lower self-handicapping tendencies, incrementally to the child's intelligence, gender, school fees, classroom environment and relationship with the teacher. This is particularly important, considering the disadvantages these strategies have for learning processes and outcomes (including the results of this study). Mastery goal orientation, and relationship with teacher had a number of indirect effects on the outcome variables through their relationship with metacognitive beliefs and intelligence constructs. Although these results require replication and extension (especially using a longitudinal design), the study also emphasises the important role played by the typical learning environment, the primary school classroom—through classroom goal orientation and relationships with teacher. In particular, communicating effectively with the 734 S. Kleitman, J. Gibson / Learning and Individual Differences 21 (2011) 728–735 Table 4 Reliability estimates, descriptive statistics and pearson correlations between the measures of study (N = 177). 1 2 3 4 5 6 7 8 9 Met. Beliefs Self-confidence Achievement SH tendencies SE Teacher Mastery GO IQ Sex School Fees Mean SD 33.09 78.06 62.92 1.99 3.24 4.19 44.73 4.56 14.01 16.02 .84 .53 .74 9.47 4111 4555 .78# 2 3 4 5 6 7 .37⁎⁎ .83 .31⁎⁎ .57⁎⁎ −.33⁎⁎ −.23⁎⁎ −.33⁎⁎ .35⁎⁎ .18⁎ .18⁎ −.07 .65 .26⁎⁎ .11 .05 −.18⁎ .26⁎⁎ .84 .32⁎⁎ .36⁎⁎ .69⁎⁎ −.18⁎ .16 −.12 .91 .77 .87 8 9 −.06 .13 .01 .06 .02 −.06 −.03 n/a −.01 .20⁎⁎ .27⁎⁎ .10 −.10 −.12 .30⁎⁎ .08 n/a Met. Beliefs = Metacognitive Beliefs; SH = self-handicapping; SE = self-efficacy; GO = goal orientation; IQ = intelligence composite. Cronbach's Alpha estimates (#if based on standardised items) are reported on the diagonal. The reliability estimates for the Metacognitive Beliefs, Self-confidence and Achievement composites are each based on the three relevant contributing scores. School fees are given in Australian dollars. ⁎ p b .05. ⁎⁎ p b .01. 15.63 Metacognitive Beliefs .23 ** Self Efficacy: Teacher .28 .27** -.28 ** .23 ** 14.57 Mastery .22 .33** Self-confidence .24** .15 * IQ .14* .44 .31 .63** Gender .11* Achievement 26.77 School Fees -.18** -.16 * 15.5 Self Handicapping Fig. 4. Final path model. Although all possible paths between exogenous and endogenous (betas) were built into the model, only significant regression coefficients (p b .05) are displayed in Fig. 4. * p b .05 (marked by dotted lines); ** p b .01 (marked by solid lines). The Metacognitive Beliefs, Self-confidence and Achievement composites are the sums of the relevant means of confidence and accuracy scores for the achievement tests. Fit Indices: χ28 = 10.62, p = .22, χ2/df = 1.33, RMSEA = .043 (.001 b 90%CI b .10), GFI and CFI = .99, TLI = .96. The model accounted for 26.1% of the variance in metacognitive beliefs, 27.3% of the variance in self-confidence, 56.1% of the variance in achievement and 15.5% of the variance in self-handicapping. teacher, and focusing learning on mastering tasks, creates a learning environment conducive to developing strong metacognitive beliefs— the key aspect of metacognition that has profound implications for both student confidence and performance. References Allwood, C. M., Granhag, P. A., & Jonsson, A. C. (2006). Child witnesses' metamemory realism. Scandinavian Journal of Psychology, 47, 461–470. Ames, C. (1992). Goals, structures, and student motivation. Journal of Educational Psychology, 84, 261–271. Ames, C., & Archer, J. (1988). Achievement goals in the classroom: Students' learning strategies and motivation processes. Journal of Educational Psychology, 80(3), 260–267. Arbuckle, J. L. (2006). AMOS 7.0 Users guide. Chicago: SPSS Inc. Australian Council for Education Research (2001). Progressive achievement tests in reading. Melbourne: ACER Press. Australian Council for Education Research (2005). Progressive achievement tests in mathematics (3rd ed.). Melbourne: ACER Press. Bandura, A. (1986). Social foundations of thought and action. Englewood Cliffs, NJ: PrenticeHall. Brown, A. (1987). Metacognition, executive control, self-regulation and other more mysterious mechanisms. In F. E. Weinert, & R. H. Kluwe (Eds.), Metacognition, motivation and understanding (pp. 65–116). Hillsdale, NJ: Lawrence Erlbaum Associates. Burchinal, M. R., Peisner-Feinberg, E., Pianta, R., & Howes, C. (2002). Development of academic skills from preschool through second grade: Family and classroom predictors of developmental trajectories. Journal of School Psychology, 40(5), 415–436. Cooke-Simpson, A., & Voyer, D. (2006). Confidence and gender differences on the Mental Rotations Test. Learning and Individual Differences, 17, 181–186. S. Kleitman, J. Gibson / Learning and Individual Differences 21 (2011) 728–735 Crosnoe, R., Burchinal, M., Keating, D., Friedman, S. L., & Clarke-Stewart, K. A. (2010). Instruction, teacher–student relations, and math achievement trajectories in elementary school. Journal of Educational Psychology, 102(2), 407–417. Crosnoe, R., Johnson, M. K., & Elder, G. H., Jr. (2004). Intergenerational bonding in school: The behavioral and contextual correlates of student–teacher relationships. Sociology of Education, 77, 60–81. Efklides, A. (2001). Metacognitive experiences in problem solving: Metacognition, motivation, and self-regulation. In A. Efklides, J. Kuhl, & R. M. Sorrentino (Eds.), Trends and prospects in motivation research (pp. 297–323). Dordrecht, The Netherlands: Kluwer. Efklides, A. (2006). Metacognitive experiences: The missing link in the self-regulated learning process. Educational Psychology Review, 18, 287–291. Efklides, A. (2008). Metacognition: Defining its facets and levels of functioning in relation to self-regulation and co-regulation. European Psychologist, 13, 277–287. Efklides, A., & Tsiora, A. (2002). Metacognitive experiences, self-concept, and selfregulation. Psychologia: An International Journal of Psychology in the Orient, 45, 222–236. Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitivedevelopmental inquiry. The American Psychologist, 34(10), 906–911. Garner, R. (1990). When children and adults do not use learning strategies: Toward a theory of settings. Review of Educational Research, 60(4), 517–529. Jonsson, A., & Allwood, C. M. (2003). Stability and variability in the realism of confidence judgments over time, content domain, and gender. Personality and Individual Differences, 34(4), 559–574. Kleitman, S. (2008). Metacognition in the rationality debate. Self-confidence and its calibration. Germany: VDM Verlag Dr. Müeller Inc., Publishers. Kleitman, S., & Moscrop, T. (2010). Self-confidence and academic achievements in primary-school children: their relationships and links to parental bonds, intelligence, age, and gender. In A. Efklides, & P. Misailidi (Eds.), Trends and Prospects in Metacognition Research (Chapter 13). US: Springer. Kleitman, S., & Stankov, L. (2001). Ecological and person-oriented aspects of metacognitive processes in test taking. Applied Cognitive Psychology, 15, 321–341. Kleitman, S., & Stankov, L. (2007). Self confidence and metacognitive processes. Learning and Individual Differences, 17, 161–173. Kröner, S., & Biermann, A. (2007). The relationship between confidence and selfconcept—Towards a model of response confidence. Intelligence, 35(6), 580–590. Mather, N., & Woodcock, R. W. (2001). Examiner's manual. Woodcock-Johnson III. Itasca, IL: Riverside. Midgley, C., Arunkumar, R., & Urdan, T. (1996). ‘If I don't do well tomorrow, there's a reason’: Predictors of adolescents' use of self-handicapping strategies. Journal of Educational Psychology, 88(3), 423–434. Midgley, C., Maehr, M. L., Hicks, L., Urdan, T. U., Roeser, R. W., Anderman, E., et al. (1995). Patterns of Adaptive Learning Survey (PALS) manual. Ann Arbor: University of Michigan. Moss, E., & St-Laurent, D. (2001). Attachment at school age and academic performance. Developmental Psychology, 37, 863–874. Murdock, T. B., & Miller, A. (2003). Teachers as sources of middle school students' motivational identity: Variable-centered and person-centered analytic approaches. The Elementary School Journal, 103, 383–399. Nelson, T. O., & Narens, L. (1994). Why investigate metacognition? In J. Metcalfe, & A. P. Shimamura (Eds.), Metacognition: Knowing about knowing (pp. 1–25). Cambridge Massachusetts: The MIT Press. Nicholls, J. G. (1984). Achievement motivation: Conceptions of ability, subjective experience, task choice and performance. Psychological Review, 91(3), 328–346. Pallier, G. (2003). Gender differences in the self-assessment of accuracy in cognitive tasks. Sex Roles, 48(5 & 6), 265–276. Pallier, G., Wilkinson, R., Danthir, V., Kleitman, S., Knezevic, G., Stankov, L., et al. (2002). The role of individual differences in the accuracy of confidence judgments. The Journal of General Psychology, 129(3), 257–299. 735 Paris, S. G., & Winograd, P. (1990). How metacognition can promote academic learning and instruction. In B. F. Jones, & L. Idol (Eds.), Dimensions of thinking and cognitive instruction (pp. 15–51). Hillsdale, NJ, England: Lawrence Erlbaum Associates, Inc. Patrick, H., Hicks, L., & Ryan, A. M. (1997). Relations of perceived social efficacy and social goal pursuit to self-efficacy for academic work. Journal of Early Adolescence, 17, 109–128. Pressley, M., & Gaskins, I. W. (2006). Metacognitively competent reading comprehension is constructively responsive reading: How can such reading be developed in students? Metacognition Learning, 1, 99–113. Roeser, R., Midgley, C., & Urdan, T. (1996). Perceptions of the school psychological environment and early adolescents' psychological and behavioural functioning in school: The mediating role of goals and belonging. Journal of Educational Psychology, 88(3), 408–422. Schraw, G., Crippen, K. J., & Hartley, K. (2006). Promoting self-regulation in science education: Metacognition as part of a broader perspective on learning. Research in Science Education, 36, 111–139. Schraw, G., & Dennison, R. S. (1994). Assessing metacognitive awareness. Contemporary Educational Psychology, 19(4), 460–475. Schunk, D. H. (1989). Self-efficacy and achievement behaviours. Educational Psychology Review, 1(3), 173–208. Schunk, D. H. (1991). Self-efficacy and academic motivation. Educational Psychologist, 26(3 & 4), 207–231. Stankov, L. (1999). Mining on the ‘no man's land’ between intelligence and personality. In P. L. Ackerman, P. C. Kyllonen, & R. D. Roberts (Eds.), Learning and individual differences: Process, trait, and content determinants (pp. 315–367). Washington, DC, USA: American Psychological Association. Stankov, L., & Kleitman, S. (2008). Processes on the borderline between cognitive abilities and personality: Confidence and its realism. In G. J. Boyle, G. Matthews, & D. Saklofske (Eds.), The handbook of personality theory and testing (pp. 541–555). Sage Publications. Stankov, L., & Lee, J. (2008). Confidence and cognitive test performance. Journal of Educational Psychology, 100, 961–976. Stright, A. D., Neitzel, C., Sears, K. G., & Hoke-Sinex, L. (2001). Instruction begins in the home: Relations between parental instruction and children's self-regulation in the classroom. Journal of Educational Psychology, 93(3), 456–466. Thomas, C. R., & Gadbois, S. A. (2007). Academic self-handicapping: The role of selfconcept clarity and students' learning strategies. The British Journal of Educational Psychology, 77, 101–119. Urdan, T. (2004). Predictors of academic self-handicapping and achievement: Examining achievement goals, classroom goal structures and culture. Journal of Educational Psychology, 96(2), 251–264. Van der Stel, M., & Veenman, M. V. J. (2010). Development of metacognitive skilfulness: A longitudinal study. Learning and Individual Differences, 20(3), 220–224. Veenman, M. V. J., & Elshout, J. J. (1999). Changes in the relation between cognitive and metacognitive skills during the acquisition of expertise. European Journal of Psychology of Education, XIV, 509–523. Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge: Harvard University Press. Walker, J. M. T. (2008). Looking at teacher practices through the lens of parenting style. The Journal of Experimental Education, 76(2), 218–240. Want, J., & Kleitman, S. (2006). Imposter phenomenon and self-handicapping: Links with parenting styles and self-confidence. Personality and Individual Differences, 40(5), 961–971. Wentzel, K. (2002). Are effective teachers like good parents? Teaching styles and student adjustment in early adolescence. Child Development, 73(1), 287–301. Woodcock, R. W., McGrew, K. S., & Mather, N. (2001). Woodcock-Johnson III. Itasca, IL: Riverside.