In all instances ICCs for single measures were identical to the above Pearson correlations for the first two decimals. Therefore, differences in Learning scale were not substantial.
The presented version of the AMAS questionnaire was an adaptation of an already established scale. Therefore, construct validity was analyzed by means of confirmatory factor analysis.
We aimed at testing the structure of math anxiety and its components using structural equation modeling. The model was built in such a way that it matched the original factor structure of the AMAS also found in an exploratory factor analysis—Data Sheet 1. It involved two correlated latent variables representing the Learning and Testing Math anxiety components.
Items 1, 3, 6, 7, 9 were assumed to contribute to the Learning latent variable, whereas items 2, 4, 5, 8 were assumed to contribute to the Testing latent variable. The model, together with standardized path coefficients, is presented in Figure 2. All parameter estimates were found to be significantly different from zero. Figure 2. Indices of model fit are provided and discussed in the main text. The results of the confirmatory factor analysis show that the internal structure of the Polish adaptation of the AMAS is similar to the structure found in the US-American sample.
Standardized coefficients are provided for the structural equation model. Variables labeled with e1, e2 etc…denote the respective error terms. Squared multiple correlations between items and the respective subscales are reported in Table 1.
Apart from Item 1 using tables , all values are close to or above 0. Taking into consideration the loadings for item 5 homework on both scales, observed in exploratory factor analysis, an alternative structural model was tested with a path also from the Learning latent variable to this item.
This is also justified from a theoretical point of view. Being given difficult homework involves both a learning situation and elements of being tested when the work is checked, usually in front of the class.
In the modified model, paths to this item were 0. As was stated in the predictions section, several correlational analyses were conducted in order to examine the convergent and discriminant validity of the AMAS.
All respective data is presented in Table 2. For clarity of description we only present a simplified correlation matrix, in which only correlations between the AMAS scores with external measures are presented. The complete correlation matrix is presented in Data Sheet 2. As can be seen in Table 2 , the AMAS scores strongly correlated with self-assessed math skills: higher levels of math anxiety were associated with poorer self-assessed math competence.
Visual inspection of scatterplots representing the relationship between the AMAS total score and average school grade and the AMAS total score and self-assessed math skill showed no departures from a linear relationship. It was also corroborated by inspection of the Lowess curves superimposed over the scatterplots.
The same was true in case of both AMAS scales. This negative correlation is present for all fields of math included in the scale. The AMAS scores correlated negatively with self-reported typical math scores at all levels of education. Participants with a higher level of math anxiety achieved worse grades in the Polish system of school grades, numerically high grades correspond to good scores. Hence, math anxiety is more strongly related to self-assessed skill than to school grades but it correlates with both.
Moreover, participants showing higher levels of math anxiety reported getting discouraged faster when struggling with math problems. Interestingly, in the case of struggling with difficult essays, they perceived themselves to be more persistent. Here, the correlation with the AMAS was very small but positive. This correlation may be caused by several factors—highly math anxious participants prefer humanities because of better performance in the latter.
On the other hand, participants might simply have contrasted their persistence in math and humanities and the latter seemed much higher to them. Higher math anxiety was associated with more use of non-allowed aids when struggling with math problems, but did not correlate with it in the case of humanities all based on self-reports.
Contrarily, higher math anxiety was associated with more liking of humanities. Therefore, the AMAS score can neither be accounted for by general attitude toward school and school subjects nor by lack of persistence when struggling with problems.
Math anxiety is specifically negatively related to math skills objectively and self-assessed and to math attitudes. Contrarily it is not correlated or sometimes positively correlated with all these factors with regard to humanities.
The AMAS correlations with temperamental traits revealed an interesting pattern of results. Temperament as an elementary characteristic should be considered as primary to math anxiety and some attempts at explaining math anxiety may be based on temperamental traits.
Math anxiety did not correlate with Sensory Sensitivity or Activity. A high positive correlation with Emotional Reactivity may be interpreted as an indication that math anxiety may be a form of an exaggerated emotional response toward math problems.
On the other hand, a moderately positive correlation with Perseverance may suggest that math anxiety is increased by mentally elaborating too long about unsuccessful attempts to deal with the problem.
A moderately negative correlation with Endurance and with Briskness may indicate that math anxiety is low in participants whose behavior can be described as highly energetic and persistent. Interestingly, correlations with state and trait anxiety were moderate. This indicates that math anxiety cannot be accounted for by anxiety in general. Moreover, as predicted, correlations of math anxiety with state anxiety were numerically smaller than those with trait anxiety.
In the math group correlations with math-related items were smaller than 0. Only correlations with state and trait anxiety were significantly larger than zero. This effect was not caused by reduced variance e. To further explore gender differences in AMAS scores we tested its correlations with external measures for female and male participants separately. As regards MAAA, correlations were stronger for female participants.
In male participants correlations of Learning scale were null and non-significant. For total score and Testing scale correlations were smaller but significant. Reverse pattern of correlations was observed in case of state and trait anxiety measures. Its relation to AMAS scores were more pronounced in male participants. In order to further explore relations between math anxiety, trait anxiety, and math skills both grades and self-assessed skills we performed path analyses.
The first path analysis comprised relations between AMAS, trait anxiety and school grades. The path model together with standardized coefficients is presented in Figure 3A. All depicted coefficients were significantly different from zero. Figure 3. Path model of the relation between trait anxiety, AMAS score, and math ability.
Panel A depicts the relation between these two variables and the average math grade. Panel B depicts the analogous relation with self-assessed math skill.
Both models reached satisfactory fit only when the relation between trait anxiety and the math ability measure was set to zero. All other coefficients were significantly different from zero. The path model together with the standardized estimates is presented in Figure 3B.
All estimates were significantly different from zero. Henceforth, we can conclude that there is a specific relation between math anxiety and math performance, which cannot be accounted for by general anxiety.
In the last step of the analysis we examined whether the results obtained in our study resembled those reported in a study by Hopko et al. The respective data are presented in Table 3. Table 3. As far as descriptive statistics are concerned, the results in all countries are very similar. Unfortunately, psychometric properties and statistics were not provided in all studies.
In general, all other correlations are rather similar across the different language versions. Nevertheless, in case of the Polish sample, the correlation with trait anxiety was higher. Interestingly, the observed correlation between math anxiety and math achievement self-reported, based on typical school grades was stronger than the estimated population correlation between math anxiety and math achievement reported in a metaanalysis by Ma This may be due to differences in the measurement of math skills.
Because such high correlations between math achievement and math anxiety in Poland were already found in the PISA study see Lee, , the current study points to culture-specific variations of validity of the AMAS.
We observed few differences between cultures, but confirmed previously reported gender differences. Good psychometric properties both validity and reliability of the Polish version suggest the usefulness of the AMAS in another cultural and linguistic context that is somewhat different from those that were already tested, namely in an Eastern Europe culture.
The AMAS is characterized by very good reliability properties as assessed by both Cronbach alpha as well as test-retest correlation.
When the ordinal alpha coefficient, considered to be more suitable for the Likert scale response format see Zumbo et al. In our study, the 4-months period between initial testing and retest was quite long compared to typical test-retest reliability study designs, which usually encompass only a few weeks. Nevertheless, satisfactory test-retest estimates indicate that math anxiety is substantially stable over time. When interpreting the values, one must keep in mind that the retest sample was very homogeneous, comprising only psychology students.
Therefore, reliability might be even higher for the general population. The factor structure obtained in the Polish sample supports a two-factor solution, one factor referring to math learning anxiety and the other to math testing anxiety. Based on our analysis, the item concerning being given difficult math homework should not be included in the Learning scale in our Polish sample. Factor loadings for this item were very similar for both factors.
Double loadings are different from the original sample, but in our view not inconsistent with item content, because it refers both to learning and being exposed to evaluation afterwards. Normally, items with double loadings are excluded. However, this item is characterized by a strong item-total correlation and therefore, it would not be recommended to exclude it from the scale. The results of the convergent and discriminant validity analyses also revealed satisfactory results.
As expected, the AMAS scores correlated moderately with state and trait anxiety, a trait measure for Negative Emotionality , a trait measure of Perseverance , and trait measure for Endurance. Highly math anxious individuals are somewhat more state- and trait-anxious in general, are more likely to respond with negative emotions in a wide range of situations, and have lower general endurance. The latter correlation is in line with the observation of local avoidance observed in highly math anxious individuals.
When facing a math problem, these individuals tend to terminate the anxiety-evoking situation by impulsively providing the answer and not considering its accuracy Ashcraft and Ridley, The correlations with other temperament trait measures were null or did not significantly deviate from zero, which may be taken as evidence for discriminant validity of the AMAS.
Henceforth, we conclude that the AMAS is related to some general psychological characteristics. Nevertheless, the generally moderate correlations in a large sample suggest that math anxiety is a unique trait that cannot be reduced to or fully explained by those general traits discussed above. All these correlations hold irrespective of whether participants study math related or math unrelated subjects.
Indications for both, convergent and discriminant validity of the AMAS, were observed. The AMAS correlated negatively with self-assessed math skill, but the correlation between the AMAS score and self-assessed math skill in general was significantly larger than the correlation with self-assessed geometry skill. This is in line with results obtained in children by Vukovic et al.
The AMAS score also correlated with self-reported math grades at all levels of education. Furthermore, consistent with previous US-American studies, the correlation between self-assessed math skills was more pronounced than the correlation between math anxiety and self-reported school grades Ashcraft and Ridley, However, the relationship observed in our study is considerably stronger than the average correlation between math anxiety and math achievement see: Ma, One must also keep in mind that we used self-reported math grades instead of official school documentation.
What is more, the results of the PISA study suggest that in Poland the relationship between math anxiety and math achievement is above average Lee, This literature suggests that the stronger relationship between math anxiety and math achievement in Poland may be real and not an artifact of the self-assessment question.
Nevertheless, to be sure, this has to be examined in future studies. Highly math anxious participants also reported getting discouraged more easily when facing difficult math problems, but not when writing an essay.
This kind of behavior resembles the mechanism of local avoidance already described above. Highly math anxious participants also reported using more non-allowed aids than low anxious participants when solving math problems, but not when solving problems in humanities.
Furthermore, highly anxious individuals also reported liking math less than low anxious individuals. This was more pronounced than the relationship between the AMAS score and liking science. In sum, highly math anxious individuals report worse math performance and more specific negative attitudes toward math.
However, this correlation pattern was present only in individuals from non-math group i. In the math group we did not observe correlations between math related measures and math anxiety. This result deserves more attention in future studies.
Results of Polish version fall between results from other versions as regards average scores, correlations and reliability estimates. The only substantial difference was a lower test-retest reliability estimate of the Learning scale in Polish than in the US American sample which is the only for which such reliability estimates are available.
This is not necessarily due to a cross-cultural difference, because the Polish retest sample was very homogeneous and the test-retest interval was much longer—therefore lower reliability scores are to be expected.
Furthermore, in the US American sample the subscales were more strongly correlated than in Polish sample, unfortunately such estimates were not provided in Italian and Iranian studies. In our study we found a significant mean difference in math anxiety between male and female participants. This is in line with several studies conducted up to date. Such discrepancy in estimated effect size of the gender difference may originate from the fact that participants of PISA study were adolescents year-olds whereas we tested students.
The other reason may be that different instruments were used to measure math anxiety. Interestingly, the observed gender difference was largely driven by Testing scale. However, it requires further investigation whether larger effect size in case of Testing scale originates from the fact that strong floor effect was observed in case of Learning scale. The other explanation could be that Testing scale may be more strongly related with test anxiety.
It was shown that, contrarily to male, female individuals perceive testing situation as threat rather than challenge Zeidner, Suinn Mathematics anxiety rating scale - psychometric data. Journal of Counselling Psychology, 6 , Introduction Mathematics anxiety is a negative emotion associated to mathematics related activities, such as mathematics homework. There are many self-help videos about mathematics anxiety on YouTube.
Click here to see a list of them. A way around this is just using the word "mathematics", as has been chosen for the PsyToolkit demonstration. Students generally found the tasks maths and reading tests and filling in questionnaires within the realm of what they normally do in the school day. AD and DS made substantial contributions to the conception and design of the work. EC and DS were involved in analysis of the data. EC drafted the work with contributions from FH.
AD and DS were involved in critical revisions and discussion of intellectual content. The project also received funding from the James S. McDonnel Foundation The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Ashcraft, M. Math anxiety: personal, educational, and cognitive consequences.
Bentler, P. Significance tests and goodness of fit in the analysis of covariance structures. Brown, L. A randomized controlled trial of acceptance-based behavior therapy and cognitive therapy for test anxiety: a pilot study. Carey, E. The chicken or the egg? The direction of the relationship between mathematics anxiety and mathematics performance. Cipora, K. Math anxiety assessment with the abbreviated math anxiety scale: applicability and usefulness: insights from the polish adaptation.
Costello, A. W Best practices in exploratory factor analysis: four recommendations for getting the most from your analysis. Department for Education a. Department for Education b. School and College Performance Tables. DeVellis, R. Scale Development: Theory and Applications. Devine, A. Gender differences in mathematics anxiety and the relation to mathematics performance while controlling for test anxiety.
Brain Funct. Finch, H. Comparison of the performance of varimax and promax rotations: factor structure recovery for dichotomous items.
Fox, J. Gadermann, A. Estimating ordinal reliability for likert-type and ordinal item response data: a conceptual, empirical, and practical guide. Garrido, L. Performance of velicer's minimum average partial factor retention method with categorical variables. A new look at horn's parallel analysis with ordinal variables. Methods 18, — Hembree, R. Correlates, causes, effects, and treatment of test anxiety. The nature, effects and relief of mathematics anxiety.
Hopko, D. The abbreviated math anxiety scale AMAS. Assessment 10, Horn, J. A rationale and test for the number of factors in factor analysis.
Psychometrika 30, — Hu, L. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. A Multidiscip. Kazelskis, R. Mathematics anxiety and test anxiety: separate constructs? Kohavi, R. A study of cross-validation and bootstrap for accuracy estimation and model selection. Joint Conf. Google Scholar. Maloney, E.
Math anxiety: who has it, why it develops, and how to guard against it. Trends Cogn. Marsh, H. In search of golden rules: comment on hypothesis-testing approaches to setting cutoff values for fit indexes and dangers in overgeneralising Hu and Bentler's findings.
Mplus User's Guide, 7th Edn. Nussbeck, F. Analysing multitrait-multimethod data with structural equation models for ordinal variables applying the WLSMV estimator: what sample size is needed for valid results?
Osborne, J. Best Practices in Exploratory Factor Analysis. Primi, C. Putwain, D. Is the relationship between competence beliefs and test anxiety influenced by goal orientation? Raiche, G. Ramirez, G. Level C required. Please log in to download sample materials. Select Pricing Below. In stock SKU amas-adult-manifest-anxiety-scale. Customize and Add to Cart. Add to Favorites. Includes Questionnaire and Evaluation Form. Your Customization.
0コメント