The Achievement Emotions Questionnaire-Argentine (AEQ-AR): internal and external validity, reliability, gender differences and norm-referenced interpretation of test scores

This article reports on the reliability, internal validity, external validity, gender differences, and norms of a Spanish version of the Achievement Emotions Questionnaire (Pekrun et al., 2011) adapted for Argentinean university students (namely AEQ-AR). The AEQ-AR contains 24 scales measuring enjoyment, hope, pride, relief, anger, anxiety, shame, hopelessness, and boredom during class, while studying, and when taking tests and exams. Argentinean undergraduates studying at the National University of Córdoba participated in the study. An estimation sample (N = 400) and a validation sample (N = 266) were formed to examine internal validity and reliability. The total sample (N = 666) was used to analyze external validity, gender differences and to obtain norms for the scales. Results indicate that the scales are reliable, internally valid as demonstrated by exploratory and confirmatory factor analysis, and externally valid in terms of relationships with task value, social academic self-efficacy, achievement goals, avoidance of help seeking, and academic performance. In addition, partial support for the gender differences hypothesis of prospective and retrospective emotions related to negative results was found. The obtained norms for male and female students will allow interpret the scores obtained for practical purposes. Finally, instructions and scales of the AEQ-AR are presented in the appendix.


INTRODUCTION
Development and adaptation of instruments to measure test anxiety has been constant and has covered languages such as English (Cassady & Johnson, 2002), German (Hoddap, 1996), Japanese (Kondo, 1997) and Spanish (Ferrando Varea & Lorenzo, 2002;Heredia, Piemontesi, Furlan, & Hodapp, 2008).While the measurement of test anxiety has made systematic progress in the last sixty years, there is still a lack of measures for addressing other relevant academic emotions and situations.An exception to this is Pekrun´s work on the development of the Achievement Emotions Questionnaire (for a review of the control-value theory of achievement emotions on which this instrument is based, see Pekrun, 2006;Pekrun & Perry, 2014).
In this research, the Achievement Emotions Questionnaire was translated into Spanish for Argentinean college students and their psychometrics properties were analyzed.

The Achievement Emotions Questionnaire (AEQ)
Pekrun and colleagues developed the AEQ (Pekrun et al., 2002(Pekrun et al., , 2005(Pekrun et al., , 2011)), a selfreported instrument assessing college students' achievement emotions.The AEQ consists of 232 items and measures eight different class-related emotions, eight learning-related emotions, and eight test emotions.The class-related emotion scales include 80 items and instruct students to report how they feel with regard to class-related enjoyment, hope, pride, anger, anxiety, shame, hopelessness, and boredom.The learning-related emotion scales include 75 items and instruct students to report how they feel with regard to studying in terms of the same eight emotions as above.Finally, the test-related emotion scales include 77 items and instruct students to indicate how they feel with regard to test-related enjoyment, hope, pride, relief, anger, anxiety, shame, and hopelessness.Furthermore, by varying the instructions accordingly, the AEQ is able to assess students' general emotional reactions in academic situations (trait achievement emotions), emotional reactions in a specific course or domain (course/domain-specific achievement emotions), or emotions at a specific time point (state achievement emotions).
Reliabilities of the original AEQ scales range from adequate to excellent (e.g., Alpha = .75to .93 in Pekrun et al., 2011).The overall structural validity of the instrument has been tested in confirmatory factor analyses (Pekrun et al., 2011).A two-facet approach best represented the data, with different emotions (enjoyment, pride, hope, etc.) represented as separate latent factors and the three settings (class, learning, and tests) represented as correlated uniquenesses.The results confirmed that the measurement of achievement emotions should attend both to the differences between discrete emotions and between the various academic settings in which these emotions take place.With regard to external validity, the AEQ has been shown to predict students' academic achievement, course enrollment, and dropout rates.Also, achievement emotions as assessed by the AEQ relate to variables of students' learning such as study interest, academic control, self-efficacy, task value, achievement goals, motivation to learn, cognitive and metacognitive learning strategies, investment of study effort, irrelevant thoughts, perceived competence, and selfregulation of learning (see Artino & Jones, 2012;Daniels et al., 2009;Mouratidis, Vansteenkiste, Lens & Auweele, 2009;Pekrun et al. 2002Pekrun et al. , 2009Pekrun et al. , 2011Pekrun et al. , 2014;;Pekrun & Perry, 2014;Spangler et al. 2002).In summary, findings indicate that the scales are reliable, internally valid as demonstrated by confirmatory factor analysis, and externally valid in terms of relationships with students' control-value appraisals, learning, and academic performance.
Interestingly, although the accumulated evidence shows that women experience more test anxiety (Zeidner, 1998), few studies have directly explored gender differences in achievement emotions, with the clearly exception of test anxiety.However, some studies have consistently shown that women experience frequently more anxiety, shame and hopelessness related to class attendance (González, Donolo & Rinaudo, 2009;Pekrun et al., 2006;Sánchez Rosas, 2013).Sánchez Rosas (2013) hypothesized that women would experience more often prospective and retrospective emotions related to obtaining negative results in class (hopelessness, anxiety, shame).Similarly, one might think the same pattern could be extended to situations of study and examination.
To date, the instrument has mainly been employed for research purposes, but Pekrun et al. (2011) claimed that it also may be well-suited to serve practical purposes for assessment in counseling and evaluation.Moreover, they stated that given the overall length of the instrument, this may require further research to tailor the scales to the specific purposes within given diagnostic settings.In consequence, shorter versions of the AEQ would be beneficial for practitioners' purposes, but research would be needed to norm the scales to ease interpretation In summary, an instrument is needed that fulfils quality requirements and that assesses a broad spectrum of academic emotions.In addition, is needed that such instrument provides the information needed to plan and evaluate interventions while it should be easy to administer and interpret.In the present study, the psychometric properties of a Spanish version of the AEQ (Pekrun et al., 2011)

Participants
Argentinean undergraduates studying in thirteen departments at the National University of Córdoba participated in the study (N = 666; 85 % female, 15 % male; M = 25.09years, SD = 6.79), with predominance of psychology (58%) and languages (23%) students.Randomly, an estimation sample (N = 400) and a validation sample (N = 266) was formed to examine internal validity and reliability.The total sample (N = 666) was used to analyze external validity, gender differences and to obtain norms for the scales.

Measures
Achievement Emotions Questionnaire (Pekrun et al., 2011).As mentioned above, this instrument measures different achievement emotions that take place when students attending class, studying, and taking tests.Students rated their emotional experiences on a five point Likert-type scale from (1) Never, to (5) Always.
Task value.The unidimensional Task Value Scale by Pintrich, Smith, Garcia and McKeachie (1993) was used.This scale evaluates perceived interest, importance and utility regarding learning materials and contents, and consists of six items (e.g., I think what I learn in this course will be useful in others, original α = .90).The items were answered using a Likert scale, expressing the degree of agreement, from (1) Strongly disagree to (5) Strongly agree.This scale demonstrated criterion validity regarding achievement emotions, in a sample of university students from the same population (Sánchez Rosas, Piotti, Sánchez, Pereira, & Debat, 2011).Unidimensionality and internal consistency yielded acceptable results in this study (KMO = .86,58% variance accounted and factor loadings > .70,α = .85,N = 666).
Social academic self-efficacy.The unidimensional social academic self-efficacy scale by Olaz (2006) was used.This scale assesses students' beliefs regarding their interpersonal abilities in an academic context.It has six items (e.g., Ask questions to the teacher loudly and in front of your classmates) and the original internal consistency is good (α = .84).Participants responded on a scale from (1) I can't do it to (10) Totally sure I can do it, expressing confidence for each behavior.Unidimensionality and internal consistency were tested.Good results were obtained (KMO = .88,73% variance accounted and factor loadings > .80,α = .93,N = 666).
Help-seeking avoidance.The Avoidance of help-seeking was measured with a five items scale (e.g., I don't ask questions in class even if I don't understand the lesson, original α = .90)validated by Sánchez Rosas and Perez (2015) for Argentinian university students.The items are answered using a Likert scale, expressing the degree of agreement, from (1) Strongly disagree to (5) Strongly agree.Here, dimensionality and internal consistency were tested.Optimum results were obtained [χ² (5, N = 666) = 15.75, p = .008,CFI = .99,GFI = .99,RMSEA = 0.057, α = .90].

Procedure
First, items were translated from English to Spanish by a professional translator; paying special attention that items have a clear, accurate and simple formulation, trying to keep the original meaning of the construct they intend to assess; and changes were made to some expressions not commonly used in Spanish.In this process, the translator was guided on conceptual issues that could clarify the intentionality of each item, regarding the target population.A cognitive pretest was applied to a small group of university students, aiming to determine how they interpret the items.Specifically, it attempted to find out the meanings the students attributed to the particular words.Subsequently, difficulties and comments regarding the items were analyzed, and slight modifications were performed on those items (Karabenick et al., 2007).
A protocol was formulated comprising instruments, and questions about gender, age, academic unit, year of coursework and GPA (Grade Point Average) including failed marks.The protocol was administered to the sample through the online survey system LimeSurvey (Pérez, 2007).All participants were informed about the study objectives, and confidential data processing was guaranteed.Students voluntarily agreed to participate.

Data analysis
Prior to the central analysis, items were explored in order to find missing values, outliersboth univariate and multivariate, normal distribution and multicollinearity (George & Mallery, 2007).Univariate outliers were identified by calculating z scores for each variable, considering values of z > 3.29 as inappropriate and multivariate outliers were detected by applying Mahanalobis distance (p < .001).In order to check normality, values of skewness and kurtosis ranging between +2 and -2 were considered acceptable (George & Mallery, 2007).Items' multicollinearity was estimated using bivariate Pearson correlations, considering values of r < .80 as appropriate.
In order to analyze internal validity, an exploratory and confirmatory strategy was implemented by conducting exploratory and confirmatory factor analysis.The reason for this is that the exploratory analysis seeks to identify the items with the best factor loadings which subsequently will be evaluated through a confirmatory analysis.Thus, an exploratory factor analysis was performed with the estimation sample (N = 400) to assess the structure underlying the set of items (Pérez & Medrano, 2014).Specifically, the guidelines for factor analysis recommended by Fabrigar, Wegener, MacCallum and Strahan (1999) were followed.Maximum Likelihood method for factor extraction was used, since it produces the best parameter estimates (Pérez & Medrano, 2014).Multiple criteria were used for factor selection: (a) the eigenvalues-greater-than-one rule proposed by Kaiser (Kaiser, 1960), (b) the scree plot (Cattell, 1966), (c) parallel analysis (Horn, 1965), (d) the percentage of variance explained by the obtained factor structure (cumulative variance of the factors extracted together) is of at least 50% of the total variability of response to test (Merenda, 1997).Because all analysis suggested extracting a single factor, a one dimensional solution was specified for each scale.Finally, as an additional criterion it was decided to retain those items with item-factor correlations > .50.On the other hand, a confirmatory factor analysis was performed with the confirmation sample (N = 266), to contrast the unidimensional specified theoretical model which was based on results of the exploratory factor analysis (Arias, 2008).In addition, concerning the relations between emotions, correlational analysis and confirmatory factor analysis were used to document the distinctness of the emotion constructs assessed by the AEQ.It was expected that a confirmatory factor analysis model representing the two-facet structure of the instrument (i.e., nine different emotions nested within three different achievement settings) would best fit the data, as compared with alternative models.The alternative models included a onefactor model representing positive versus negative emotions as one bipolar factor, as well as two models differentiating between emotions only, or between different settings only (Figure 1).Following recommendations of Hoyle and Panter (1995), model's goodnes-of-fit was diagnosed with multiple criteria.Chi-square/degrees of freedom ratio values (χ2/df), comparative fit index (CFI), goodness-of-fit index (GFI), and root mean square error of approximation (RMSEA) were considered.Goodness-of-fit values were interpreted as following: CFI and GFI > .95,RMSEA < .06 was considered a good fit; χ2/df < 3, CFI and GFI > .90,RMSEA < .08 was acceptable; and RMSEA from .08 to .10 was mediocre.
In order to assess reliability, internal consistency was then estimated using Cronbach's alpha coefficient.An alpha coefficient of .70 was interpreted as acceptable, .80 as good, and .90 as excellent (George & Mallery, 2007).An item shall only be removed if values of internal consistency are improved as a result.
To provide evidence of external validity, relations between the achievement emotions scales and task value, social academic self-efficacy, achievement goals, avoidance of help seeking, and grade point average were explored.For this purpose, correlations between variables were calculated using Pearson's r coefficient.As evidenced by some studies, task value (Pekrun et al., 2011;Sánchez Rosas et al., 2011), social academic self-efficacy (Sánchez Rosas, 2013), mastery goals (Pekrun et al., 2009) were expected to correlate positively with positive emotions.Positive emotions were expected to correlate negatively with performance goals and (Pekrun et al., 2009) and avoidance of academic help seeking (Sánchez Rosas, 2013;Sánchez Rosas & Pérez, 2015).An opposite pattern of relationships is expected for negative emotions.
To allow future interpretation of individual scores of the AEQ-AR scales for practical purposes, each scale should be normed.In consequence, deciles for each scale were calculated by gender.

RESULTS
As a result of the preliminary analysis, some items were discarded because the values of kurtosis were inadequate (> 2) (When I think about class, I get queasy; I'd rather not go to class since there is no hope of understanding the material anyway; It's pointless to prepare for class since I don't understand the material anyway; When my studies are going well, it gives me a rush; After extended studying, I'm so angry that I get tense; I get so angry, I start feeling hot and flushed).The values of skewness and kurtosis for the resultant set of items (< 2) were adequate (George & Mallery, 2007), showing a normal distribution of the items.In addition, it was noted that some items had very high correlations, showing an unnecessary overlap in the items' content.Because it is an assumption of exploratory factor analysis that items are related and showing no multicollinearity, these items were removed (I get embarrassed; I find this class fairly dull; The lecture bores me; For me the test is a challenge that is enjoyable; I am proud of myself; My hopelessness robs me of all my energy).Finally, it was found that one item did not correlate with the items in its own scale (I can finally laugh again) so it was removed.

Exploratory Factor Analysis and Internal Consistency
In a first attempt, twelve items showed item-factor correlations < .

Confirmatory Factor Analysis and Internal Consistency
A confirmatory factor analysis was performed in order to examine the one-factor model obtained by exploratory factor analysis.As in this case, the presence of a large number of items per scale often leads to difficulties in obtaining good model fit.As Bandalos (2002) recommends, four parcels per scale were conformed.This way, a onefactor model was evaluated for each scale, in which each factor explained the behavior of its specified four elements.
The models showed good fit to the data with high factor loadings (p ≤ .001,see Table 2).In addition, with this sample, all the scales showed acceptable levels of internal consistency.

Relationships between emotions: Correlational analysis
In accordance with Pekrun et al. (2011), it is useful to distinguish (a) between the different discrete emotions that occur within a given achievement setting (class-related, learning-related, test-related), and (b) between the emotions experienced in different achievement settings.As may be seen from Table 3, the positive emotions of enjoyment, hope, and pride correlated moderately high and positively in all three settings.Similarly, there were moderate to high and positive correlations between the negative emotions of anger, anxiety, shame, hopelessness, and boredom.The correlations between these positive emotions, on the one hand, and negative emotions, on the other hand, were moderately negative.

Relationships between emotions: Structural equation modeling of latent relationships
In this research, in order to more fully assess the relationships between achievement emotions, the same four models proposed by Pekrun et al. (2011) were tested competitively: The one-factor model, the nine-emotion factor model, the three-setting factor model, the two-facet, emotion x setting model (Figure 1).A more fully descriptions of the models can be seen in Pekrun et al. (2011).
Similarly, the three-setting factor model had a poor fit (χ2/df = 9.37, CFI = .59,GFI = .44,and RMSEA = 0.178).In marked contrast, the two-facet, emotion x setting model showed a reasonable fit, with χ2/df = 2.79, CFI = .95,CFI = .91,and RMSEA = 0.079.In consequence, these findings demonstrate that the relationships between different achievement emotions can be best explained by taking into account both the differences between discrete emotions and the differences between emotions that occur in different achievement settings.
Latent relationships between the nine emotions of the two-facet model (Table 4) were positive for enjoyment, hope, and pride; positive for anger, anxiety, shame, hopelessness, and boredom; and negative between these positive and negative emotions.

Relationships with students' appraisals, motivation, strategy, and performance
Table 5 shows the relations between the achievement emotions scales and task value, social academic self-efficacy, achievement goals, avoidance of help seeking, and academic performance.Task value, social academic self-efficacy, mastery goals, and academic performance correlated generally positively with the positive emotions and negatively with the negative emotions.On the other hand, performance goals and avoidance of academic help seeking were found to correlate negatively with positive emotions, and positively with negative emotions.

The AEQ-AR: Gender Differences
As presented in Table 6, eight emotions showed gender differences.Female students reported more test-related relief, class-related anxiety, learning-related anxiety, test-related anxiety, class-related shame, learning-related shame, test-related shame, and test-related hopelessness than male students.In interpreting these gender differences, it should be noted that the effect sizes of the differences were generally small (all ds < .56).Interesting, anxiety and shame differences were generalized across settings.As hypothesized, testrelated hopelessness was higher in female students, but there were no significant mean differences for class and learning-related hopelessness.The AEQ-AR: Norm-Referenced Interpretation of Test Scores In Table 7 norms for male and female students of AEQ-AR scales are reported.
Note that although gender differences were only found for some of the emotions, also gender-differentiated norms were established for all scales.

DISCUSSION
The last decade showed a growing and sustained interest in research of different discrete emotions in educational contexts using self-reports as the Achievement Emotions Questionnaire (Pekrun & Bühner, 2014).This instrument proved to be reliable and valid for measuring a set of emotions prevalent in typical academic situations in different cultures (Ismail, 2015;Kim & Lee, 2014;King, 2010;Molfenter, 1999;Peixoto et al., 2015;Titz, 2001).However, psychometric emphasizes the need to adapt psychological assessment instruments developed in other cultural contexts and rigorously assess compliance with the psychometric standards.In Argentina, previous publications referring to this instrument have reported data using preliminary versions or selected scales only (Sánchez Rosas, 2011, 2013, 2015;Sánchez Rosas & Bedis, 2015;Sánchez Rosas & Pérez, 2015;Sánchez Rosas et al., 2016, in press), but no studies had evaluated the psychometric properties of the overall instrument.The aim of this study was to obtain a Spanish version of the AEQ (Pekrun et al., 2011) adapted for Argentinean university students, namely AEQ-AR (see Appendix for the complete instrument).In this research, reported results provide evidence of reliability and validity of the AEQ-AR.In addition, gender differences were in line with expectations.Lastly, the norms will allow interpret the scores obtained for practical purposes.

The AEQ-AR: Internal Validity and Reliability
The validity and reliability of the AEQ-AR scales have been accomplished by exploratory and confirmatory factor analysis, alpha coefficients and item-total correlations (see Appendix).The validity and reliability studies were conducted with two different samples.The first sample was used for exploratory factor analysis and reliability analysis.
The second sample was used for confirmatory factor analysis.The results of both studies provided acceptable evidence for reliability and validity of the AEQ-AR.The exploratory factor analysis showed that a one dimension factor solution for each scale presented adequate average values of explained variance, high factor loadings, and excellent internal consistency.The same one-dimensional factor solutions obtained with the estimation sample were assessed by confirmatory factor analysis, obtaining good fit to the data with high factor loadings, and good levels of internal consistency.
Positive emotions (enjoyment, hope, and pride) correlated moderately highly and positively in all three settings, except for relief that showed moderate but positive correlations (Table 3).Similarly, there were moderate to high and positive correlations between the negative emotions (anger, anxiety, shame, hopelessness, and boredom).
Correlations were high for the negative activity-related achievement emotions of anger and boredom, and for the negative outcome-related achievement emotions of anxiety, shame and hopelessness.When object focus of these negative emotions changed (like anger and shame) correlations were moderate.Correlations between these positive emotions, on the one hand, and negative emotions, on the other hand, were moderately negative.These correlations were much higher when the object focus was the same, as in the case of enjoyment with boredom or anger, and pride and hope with anxiety, shame and hopelessness.Finally, correlations of emotions experienced in different settings (e.g., classenjoyment, learning-enjoyment, and test-enjoyment) were high but not so high as to indicate overlap.Taken together and in accordance with Pekrun et al. (2011), it is useful to distinguish between different discrete emotions that occur within a given achievement setting, and between emotions experienced in different achievement settings.Even more, it is useful to distinguish the emotions according to the valence and object focus (Pekrun, 2006).
Additionally, findings demonstrated that relationships between different achievement emotions can be best explained by taking into account both the differences between discrete emotions and the differences between emotions that occur in different achievement settings.When four models were tested competitively, the two-facet, emotion x setting model showed a good fit compared with others models (one-factor model, nineemotion factor model, and three-setting factor model).Latent relationships between the nine emotions of the two-facet model were similar with the manifest correlations (Table 3).

The AEQ-AR: External Validity
In accordance with Pekrun's (2006) control-value theory, relations between achievement emotions and control-value appraisals, namely social academic self-efficacy and task value, attested external validity of the scales (Table 5; Pekrun et al., 2011;Sánchez Rosas, 2013;Sánchez Rosas et al., 2011).Social academic self-efficacy and task value correlated positively with positive emotions and negatively with negative emotions.On the one hand, task value correlated higher with activity-related emotions than with outcomerelated emotions.On the other hand, social academic self-efficacy correlated higher with outcome-related emotions than with activity-related emotions.This is because task value refers to task-related appraisals, while social academic self-efficacy refers to the ability to obtain certain outcomes.A more complex pattern of relations was found for achievement emotions scales and motivation, evaluated here as achievement goals.As predicted by Pekrun et al. (2009), mastery-approach and mastery-avoidance correlated positively with positive emotions (higher with activity-related emotions, such as enjoyment, than with outcome-related emotions, such as hope or pride) and negatively with negative activityrelated emotions (anger and boredom).Performance-approach and performance-avoidance correlated positively with negative outcome-related emotions (anxiety, shame, and hopelessness), although performance-approach also correlated positively with enjoyment and pride.These results demonstrate the detrimental effects of performance goals on outcome-related emotions, but showing at the same time beneficial effects of the approach component of performance-approach goals on some positive emotions (Pekrun et al., 2009).
In addition, as predicted by Pekrun's (2006) control-value theory and informed by some studies (Sánchez Rosas, 2013;Sánchez Rosas & Pérez, 2015), there were clear linkages between emotions and avoidance of academic help seeking as a learning strategy.Positive emotions decreased the avoidance of help seeking and an opposite pattern of relationships was found for negative emotions.Finally, positive emotions (enjoyment, pride, hope, and relief) and negative outcome-related emotions (anxiety, shame, and hopelessness) were facilitator and inhibitors of the academic performance (GPA), respectively.

The AEQ-AR: Gender Differences
This work progressed providing evidence for the gender differences hypothesis stated by Sánchez Rosas ( 2013) which proposed that women would experience prospective and retrospective emotions related to obtaining negative results in class more frequently (hopelessness, anxiety, shame).In this research, the same pattern was assumed and explored in other achievement situations (class, learning, and test situations).As expected, anxiety and shame differences were generalized across settings, and test-related hopelessness was higher in female students, but there were no significant mean differences for class and learning-related hopelessness (Table 6).In consequence, partial support was found for the gender differences hypothesis of prospective and retrospective emotions related to negative results.

The AEQ-AR: Norm-Referenced Interpretation of Test Scores
At last, Table 7 presented the norms for male and female students of AEQ-AR scales.In this way, the AEQ-AR represents a useful tool that could be employed by researchers and counselors.For example, experimental studies that evaluate the effects of some interventions could perform measurements to classify people according to the level of emotions.Also, teachers and educational psychologists could identify and guide students with positive and negative emotional experiences, comparing the individual's scores with the reference group.

Conclusions and Directions for Future Research
Taken together, the results of this study provided satisfactory evidence that the AEQ-AR is reliable and valid for the university population from Argentina, but further research is needed in order to extend the scope of this study.
Currently, there is a strong interest in educational research of achievement emotions in the context of careers related to science, technology, engineering and mathematics (STEM).Therefore, researchers interested in investigating achievement emotions in Spanish-speaking populations, particularly in Argentina, would benefit using the AEQ -AR scales in STEM careers.Even, further research could adapt and analyze the psychometric properties of the AEQ-AR for assessing domain-specific achievement emotions, such as the Achievement Emotions Questionnaire-Mathematics (AEQ-M; Frenzel, Thrash, et al., 2007).
Although the current study was conducted with university-level students, it would be helpful to adapt the AEQ-AR to assess achievement emotions in populations of other levels of education, such as primary or secondary level.
The challenge of getting a shorter version of the AEQ-AR with good psychometric properties could also be addressed.In this way, the scales could be used in experimental research, which are often limited when trying to make measurements with large scales such as those reported here.In the same way, it could be used in traditional or virtual contexts of teaching, as well as to provide immediate feedback in these contexts.
This research analyzed the bivariate relationship of achievement emotions to task value, self-efficacy, achievement goals, and academic help seeking.However, further research could explore other self-regulated learning strategies (cognitive, emotional or motivational) or coping strategies (coping with boredom), and different control-value appraisals.In addition, more sophisticated statistic method, such as path analysis (Pérez, Medrano, & Sánchez Rosas, 2013), could be employed in analyzing relationships between achievement emotion and their antecedent and outcome variables.
Although population norms for the instrument were not known, the norms developed in this study represent an advance and would have important implications for educational practice.However, these standards were prepared considering a population of university students distinguished only by gender.Further research could develop norms for specific domains such as mathematics or statistics.The level of career advancement could also be considered, as well as a distinction between general areas of knowledge such as health sciences, natural sciences or social sciences and humanities.

Escalas
adapted for Argentinean university students are assessed (namely AEQ-AR).The purposes are: (a) to examine internal validity and to obtain data of reliability from each scale, (b) to analyze external validity, (c) to test gender differences, and (d) to obtain norms for the scales.

Figure 1 .
Figure 1.SEM models for relationships between emotions.Upper left part: One emotion-factor model.Upper right part: Eight emotion-factors model.Lower left part: Three setting-factors model.Lower right part: Emotion x setting-factors model.C, I, and T denote class-related, learning-related, and test-related emotions, respectively.Jo = enjoyment, Ho = hope, Pr = pride, Re = relief, An = anger, Ax = anxiety, Hl = hopelessness, Bo = boredom.

Table 1 .
Exploratory Factor Analysis and Internal Consistency for AEQ-AR scales 50 (My enjoyment of this class makes me want to participate; I enjoy participating so much that I get energized; I am hopeful that I will make good contributions in class; I am proud that I do better than the others in this course; I feel anger welling up in me; Because I'm angry I get restless in class; I get physically excited when my studies are going well; When I solve a difficult problem in my studying, my heart beats with pride; When I excel at my work, I swell with pride; Because I look forward to being successful, I study hard; I get angry over time pressures which don't leave enough time to prepare; I get angry about the amount of material I need to know).These items were removed and a new exploratory factor analysis was performed.Table1shows the results of the exploratory factor analysis with values of KMO, variance percentage, mean of factor loadings, internal consistency, and number of items per scale.In addition, average values indicate adequate values of KMO (.89), variance percentage (59%), high factor loadings (.73), excellent internal consistency (α = .90),and enough number of items per scale (9).

Table 2 .
Confirmatory Factor Analysis and Internal Consistency for AEQ-AR scales

Table 5 .
Correlations of achievement emotions with appraisals, motivation, strategy, and performance Within each block, upper/middle/lower coefficients are for class-, learning-, and test-related emotions, respectively.For relief, test-related relief was assessed only.For boredom, class-related and learning-related boredom were assessed only.N = 666.* p < .05;** p < .01.

Table 6 .
Gender differences of AEQ-AR scales

Table 7 .
Norms for male and female students of AEQ-AR scales