European Journal of Psychology of Education 2007, Vol. XXII, nº 2, 201-216 © 2007, I.S.P.A.

On the relationship between value orientation, valences, and academic achievement Stefan Fries Sebastian Schmid Manfred Hofer University of Mannheim, Germany

Value orientations are believed to influence learning in school. We assume that this influence is mediated by the valences attached to specific school subjects. In a questionnaire study (704 students from 36 classes) achievement and well-being value orientations were measured. Students also rated valence scales for the school subjects German and Mathematics and reported their respective grades. In order to take into consideration the nested data structure, the mediation hypotheses were tested using the Hierarchical Linear Model in a series of intercept only models. School grades were significantly predicted by value orientation. A mediation analysis indicated that this relation was completely mediated by the valences of the different school subjects.

Despite a well established tradition of value research in sociology, cross-cultural psychology, and social psychology (cf. Greenfield, Keller, Fuligni, & Maynard, 2003; Hitlin & Piliavin, 2004; Seligman, Olson, & Zanna, 1996), individual values and value orientations are rarely taken into account in analyses of learning behavior. In reaction to this situation, Hofer and Peetsma (2005) recently assembled a special issue of the European Journal of Psychology of Education, in which papers mainly dealt with societal values and school motivation. Building on research presented in this special issue, the current article focuses on the relationship of value orientations and academic achievement. Furthermore, we ask how this relationship is to be explained. We will argue that valences might be a potential mediating factor of the effect of abstract value orientations on concrete learning behaviors. Defining values Since individual values are rarely considered in Educational Psychology, we briefly discuss the issues of value definition and of differentiating them from similar constructs. Values can be defined as generalized beliefs about the desirability of behaviors and events. Freedom, self-direction, power, security, etc., are typical examples of values. In contrast to The study presented in this paper was supported by the research Grant HO 649/17-1 by the German Research Foundation.

202

S. FRIES, S. SCHMID, M. HOFER

similar constructs like goals, attitudes, or norms, values have no explicit reference to specific behaviors, events, or objects. However, they allow people to decide what to prefer and what to avoid, because behaviors, events, and objects can be judged on the basis of their match or mismatch to an individual’s value system (Schwartz, 1992). Concepts from which values should be differentiated are goals, attitudes, norms, and needs. Values versus goals. Whereas goals refer to states an individual tries to achieve, a definition of an end-state is usually not part of a value. Therefore, while one can say that someone has reached a goal, it would sound odd to say that someone has “reached” a value. This distinction is still correct when considering long-term goals and life goals (cf. Cantor, 1994). However, values have motivational power, because they allow the individual to decide which goals to choose and to pursue by reflecting their convergence with the personal values held. Values versus attitudes. In a similar fashion, values can be distinguished from attitudes (cf. Maio, Olson, Bernard, & Luke, 2003). Attitudes are tendencies concerning the evaluation of events and objects. In comparison to values, attitudes are more concrete as they refer to specific events and objects. Furthermore, attitudes and values are represented on different scales. Whereas attitudes are typically scaled on a dimension of approval vs. disapproval (Himmelfarb, 1993), values are scaled with respect to a dimension of importance (Schwartz, 1992). Despite these differences, there is an important connection between attitudes and values: Attitudes are rooted in values. One function of attitudes lies in the fact that some of them are expressing an underlying value (Maio & Olson, 2000). Values versus norms. Values must also be distinguished from norms. Values are about desirability, whereas norms are about oughtness (Marini, 2000). For example, taking off your hat when entering a church is a norm, while passionate prayer may be an expression of the value spirituality. A second difference between values and norms is that norms are also more concrete than values. Values versus needs. Finally, values have to be distinguished from needs. Needs refer to innate or learned generalized desires that must be fulfilled in order to survive and/or to experience psychological growth (cf. Deci & Ryan, 2002). The major difference between needs and values lies in the fact that the latter are cognitively represented beliefs about desirability, whereas the former are psychological powers representing innate or learned desires which may or may not be cognitively represented. However, as Rokeach (1973) already noted, values and needs are related to each other. The universal value system developed by humans can be seen in part as a result of universal needs. According to Schwartz (1992), values evolved from three basic human requirements: (a) needs of the individual deriving from its organism, (b) requirements for interpersonal coordination, and (c) needs for group survival and welfare. However, values and needs are not linked by an isomorphic relationship. Indeed, values have an important regulatory function with respect to needs, since in many situations the individuals’ needs must be transcended in order to prevent conflict with the needs of other individuals. Values, valences, and behavior In Educational Psychology, individual values are usually not part of the motivational analysis of learning behavior. However, the term “value” is applied quite frequently with a different meaning (cf. Eccles & Wigfield, 2002). Within prominent expectancy-value models the term “value” refers to the evaluation of specific tasks (or domains of tasks) with respect to their desirability. For example, Wigfield and Eccles (2000) distinguish different components of subjective task value and typically operationalize them by asking students for their evaluation of specific domains (e.g., Mathematics). Such research is in line with a general trend in Educational Psychology relating constructs to specific domains. A major advantage of this trend towards specificity is an enhanced precision in the prediction of academic choices and academic achievements. Still, the explanation of variance in students’ achievement is not

VALUE ORIENTATION, VALENCES, AND ACHIEVEMENT

203

the sole objective of research into motivation to learn; another one is the development of appropriate theoretical models of the causal chain that leads to differences in preferences for different domains. Within such models, individual values may be of core importance. Taken together, individual values and subjective task value differ in their level of specificity. Hence, subjective task values should not be considered as individual values in the meaning outlined above, since subjective task value refers to a specific task or a class of tasks (e.g., Mathematical tasks), whereas individual values have no explicit object of reference. In order to make this distinction explicit, we prefer to denote the subjective task value in the following as valence (cf. Fries, Schmid, Dietz, & Hofer, 2005). But of course, these valences play a crucial role whenever a value-behavior relation is to be explained (Feather, 1990). Feather (1990, 1999) provides such an explanatory model. He argues that values have an impact on behavior via their influence on the valences underlying the behavioral decisions. In keeping with Lewin’s (1951) analysis, values are believed to have an effect on the valences of concrete actions, objects, and events. Thus, values are said to follow the same functional principles as needs. Valences then function as force fields, to which an individual is attracted or from which it is deflected. The following statement clarifies this view: Values influence behavior but have not the character of a goal (that is, of a force field). For example, the individual does not try to “reach” the value of fairness, but fairness is “guiding” behavior. It is probably correct to say that values determine which types of activity have a positive and which have a negative valence for an individual in a given situation. In other words, values are not force fields but they “induce” force fields. (Lewin, 1951, p. 41) Feather (1990) gave Lewin’s original approach a more cognitive interpretation. He assumed that values induce valences by influencing a person’s subjective definition of a situation. For example, a student valuing achievement highly may interpret the situation of meeting a classmate in a café as a good opportunity to ask about the last math assignment, whereas a student valuing well-being highly may interpret the same situation as a good opportunity for planning some fun activities together. Empirically, Feather studied the capacity of values to induce valences. Indeed, he was able to show that values were related to academic choices at the university (Feather, 1988). Although there was no direct influence of values on behavior, a mediated influence could be shown using path-analytical techniques. Students endorsing values of restrictive control attached higher valence to Mathematics; students endorsing values of prosocial concern favored English. Valences for the different subjects were also related to students’ choices of courses. In another study, Feather (1995) analyzed the relation between values, valences, and choices in hypothetical situations. Values were measured using the Schwartz Value Survey (Schwartz, 1992). Several hypothetical situations were constructed in which alternatives were representing behavior related to a value type from the circumplex model proposed by Schwartz (1992). For every hypothetical situation participants rated the attractiveness (i.e., a measure of valence) of the different alternatives and decided which alternative they would choose. Results indicated that students’ values were related to valences of the activities. There also was a relation of values to choice. In multiple regression analyses only valences had an effect on choice, indicating a mediation of the influence of values by the valences of the activities. Which values are relevant for academic achievement? Values differ with regard to their relevance for academic achievement. In our research, we concentrated on value orientations which are linked to typical demands within school. The theory of value change in postindustrialized societies proposed by the political scientist Ronald Inglehart (1997) provides a theoretical framework for identifying relevant value dimensions. Inglehart assumes that so-called modern (achievement) values like hard work, security, and prosperity are more and more competing with so-called postmodern (well-being) values like tolerance, fulfilling

204

S. FRIES, S. SCHMID, M. HOFER

social relationships, or self-actualization. Modern values are important for the process of modernization as the individual’s achievement heavily relies on them. However, the utility of further achievement is no longer seen once a high level of economic wealth is reached and is taken for granted by the members of society. In such a situation well-being values gain importance. In the three waves of the World Value Surveys, Inglehart and Baker (2000) found that during the last 25 years most societies developed in the predicted way. This analysis can be transferred to the situation of students. Oviada (2003) provides evidence for a high and still growing importance of both types of values for students. The dimensions of achievement and well-being values probably cover the core values of contemporary students, as they are related to the two major fields of students’ life: school and leisure (Lens, Lacante, Vansteenkiste, & Herrera, 2005). Of course there is no one-to-one mapping of value orientations to such broad and complex action domains like school and leisure. However, from students’ perspective school offers primarily opportunities for acting in correspondence with achievement values, whereas in leisure time opportunities for wellbeing values are dominant (Schmid, Hofer, Dietz, Reinders, & Fries, 2005). Individuals differ with respect to the importance they attach to achievement and well-being values. As these values are clustered into two fundamental sets, we speak in the following of an achievement value orientation and a well-being value orientation, meaning the overall importance an individual attributes to achievement and to well-being values. Achievement and well-being value orientations should induce more or less positive valences for the different school subjects (cf. Feather, 1990). Those students granting achievement values a higher importance should experience a greater convergence of the opportunities in school and their value orientation. Therefore, they should attach higher valences to the academic domain. Students with a high well-being value orientation should experience less convergence of the opportunities in school and their value orientation. Hence, they should attach lower valences to the academic domain. Furthermore, since valences are related to achievement (cf. Schiefele, 1999), values should also have an impact on academic achievement. In everyday life these two value orientations can easily come into conflict (Hofer, Schmid, Fries, Dietz, Clausen, & Reinders, 2007), which renders their configuration an important issue. Fries et al. (2005) report results on the impact of value configuration on achievement. However, in the present paper the focus is on the question whether valences mediate the relationship of value orientations and academic achievement. Therefore, we solely analyzed single value orientations but not configurations of values. The aim of this study In the present study, we analyze the relation between value orientation, valences, and academic achievement. First, we assume that value orientations and achievements are related: An achievement value orientation should be positively related to academic achievement, while a well-being value orientation should be negatively related to academic achievement. Second, in line with the model of Feather (1990, 1999), we assume the influence of values on behaviors to be mediated by the valences of the school subject. We tested these hypotheses with a measure of pupils’ value orientation as initial variable, a measure of the valences pupils attach to different school subjects as mediating variable, and their respective grades in these subjects as outcome variable.

Method Participants and procedure Participants were 704 students (48.4% male and 51.4% female) from nine schools in Ludwigshafen (German city with 170,000 inhabitants). The sample included 29.5% of

VALUE ORIENTATION, VALENCES, AND ACHIEVEMENT

205

students with immigrant background. Students attended the sixth (43.8%) and eighth (56.3%) grade (mean age: 13.5 years; SD=1.3). They came from different tracks within the German school system: 38.2% attended the highest school track (Gymnasium), 21.7% a middle school track (Realschule), and 40.1% the lowest one (Hauptschule). Students were nested within 36 classrooms. The average classroom sample size was 19.6 (SD=5.4). The students answered a questionnaire containing the relevant instruments as well as some instruments unrelated to the questions under study. The questionnaire was administered in classroom testing sessions. Students worked individually with a trained experimenter being present. The students were informed that the goal of the study was to learn about their attitudes and feelings toward school, leisure, and life in general. Material Value orientation. We followed the common practice in value research to assess value orientations via “value prototypes” (cf. Schwartz, Melech, Lehmann, Burgess, Harris, & Owens, 2001). The item format consisted of two comprehensive descriptions of students with different value orientations. The item for the achievement value prototype depicted a student who has clear goals, struggles through uncomfortable tasks and wants to achieve something in life. The item for the well-being value prototype described a student who spends a lot of time with friends, loves diversion and spontaneous activities, and wants to have fun in life (see Appendix). The descriptions depicted “Idealtypen” (in the sense of Weber, 1949) and were originally developed to organize verbal materials from an interview study (Schmid et al., 2005). Prototype measures seem better suited to verbal competencies of young adolescents, since the conventional item formats with abstract value descriptors like for example in the Schwartz Value Survey (Schwartz, 1992) are usually too difficult for them to understand. The applicability of prototype measures for young adolescents has been shown in various studies (e.g., Schwartz et al., 2001; Bubeck & Bilsky, 2004). The prototype items were presented in gender-congruent versions. The students were asked to first read both descriptions carefully. Then they judged their similarity to each prototype on a 6-point rating scale ranging from “very similar” (5) to “not similar at all” (0). The retest-reliabilities of the value prototype items were determined in an independent study (N=54) with an interval of two weeks between both measurements. Retest-reliabilities were (N rtt=.58 for the achievement value orientation prototype and rtt=.71 for the well-being value orientation prototype. Subject valences. The valences of the school subjects Mathematics and German were assessed by two short scales with four items each. Intrinsic incentives like fun, and extrinsic incentives, like beneficial consequences, were equally present. The wordings of the items were identical for both scales with only the school subjects being exchanged. The items were answered on 4-point rating scales ranging from “totally true” e (3) to “not true at all” (0). The Mathematics valence scale had an internal consistency (Cronbach’s α) of .67 and a mean itemtotal correlation of rit=.46. The German valence scale had a lower internal consistency1 (Cronbach’s α of .52 and a mean item-total correlation of rit=.34). For both scales, the item with the highest item-total correlation was: “Learning for Mathematics/German is fun to me” (Mathematics: rit=.55; German: rit=.47). Academic achievement. A major goal of students motivated by achievement values should be the attainment of good grades. Therefore, it appeared natural to us to use grades in Mathematics and German as a proxy of academic achievement. German grades range from 1 to 6 with 1 being the best grade and 6 being the worst grade. In order to facilitate the interpretation of the results, the polarity of the grades was reversed by subtraction from seven.

206

S. FRIES, S. SCHMID, M. HOFER

Handling of missing data From the items used in the analyses to be presented, 1.31% of the values were missing. However, despite of this small proportion, 10.37% of the students had at least one missing value. Therefore, we decided to adopt a multiple imputation (MI) approach to handle this problem (e.g., Graham, Cumsille, & Ekek-Fisk, 2003). This method produces more accurate parameter estimates and standard errors than traditional methods like listwise or pairwise deletion. In a nutshell, MI produces a number of plausible datasets with each missing value replaced by a likely estimate. These datasets are then analyzed separately by conventional complete-data methods. Finally, the results are combined according to the rules of Rubin (1987) to obtain the overall results. We generated ten imputed datasets using the data augmentation algorithm of NORM (Shafer, 1999). The results based on the analyses of the imputed datasets were combined using HLM 5 (Raudenbush, Bryk, Cheong, & Cogdon, 2000). Mediator analyses Various mediational analyses were run to test whether the relationship between value orientation and academic achievement is due to the subject’s valences, the supposed mediating variable. To establish the case for mediation, we followed the “causal steps”-approach proposed by Baron and Kenny (1986; see Kenny, Kashy, & Bolger, 1998, for an updated version of this approach). This procedure was combined with multilevel modeling techniques (e.g., Raudenbush & Bryk, 2002) in order to reflect the hierarchical structure of the dataset with students nested within classes (see Krull & MacKinnon, 2001). According to Baron and Kenny (1986), four requirements should be met to demonstrate that the data are in line with a (complete) mediation hypothesis: 1) The initial variable X is related to the outcome variable Y (test of path c). 2) The initial variable X is related to the mediator M (test of path a). 3) The mediator M is related to the outcome variable Y while controlling for the initial variable X (test of path b). To establish complete mediation, it should be shown that the XXY Y-relationship disappears when controlling for M (test of path c’). Otherwise, one would speak of partial mediation. A mediation analysis cannot prove causality; it can, however, show that the data are in accordance with the assumption of mediation. Usually the requirements for mediation are tested by a series of ordinary least squares (OLS) regression analyses. However, in clustered data sets the assumption of independent observations frequently does not hold (Raudenbush & Bryk, 2002). For example, as students within the same class are usually subject to the same influences, there might be positively correlated error terms among the individuals within a particular class. Therefore, OLS estimation cannot be applied. But multilevel analysis allows for a consideration of the nested data structure. The equations in Table 1 show how the four paths c, a, b, and c’ can be estimated within a multilevel modeling framework. In multilevel models, the complex error structure associated with clustered data can be explicitly taken into account. Taken together, the equations in Table 1 comprise a lower level mediation model (Krull & MacKinnon, 2001). In this model, all variables of interest – X, X M, M and Y – are located on level-1. No level-2 predictors are included. As can be seen from Table 1, equivalent sets of level-1 and level-2 equations can be used to estimate each of the four paths c, a, b, and c’. The level-1 equations are similar to ordinary regression equations. For example, path c is estimated by a level-1 equation with Yij as criterion variable which is predicted from an intercept term b0j, a predictor variable Xij weighted by bc, and an error term rij. However, in contrast to ordinary regression analysis, Yij, Xij, and rij are

VALUE ORIENTATION, VALENCES, AND ACHIEVEMENT

207

subscribed by two indexes, i for the individual and j for the class. Furthermore, the intercept b0j 0 is indexed by the subscript j which indicates that it may vary across classes. In the level-2 equation, this intercept is predicted from an overall intercept term, γ00, and a class-level error term2, u0j. This error term is crucial because it allows the modeling of the correlated error structure associated within clustered data (Krull & MacKinnnon, 2001).

Table 1 Multilevel equations used for mediational analysis. Path

Level 1

Level 2

c a b and c’

Υij=β0j+βcΧij+rij Μij=β0j+βaΧij+rij Υij=β0j+βc’Χij+βbΜij+rij

β0j=γ00+u0j β0j=γ00+u0j β0j=γ00+u0j

We performed separate mediation analyses with achievement and well-being value orientations as initial variables and grade in Mathematics and grade in German as outcome. The respective valences of Mathematics and German were used as mediator variables. In addition to the Baron-Kenny procedure, mediation was tested using the Sobel-Test for the product of the paths a and b (Sobel, 1982).

Results Table 2 shows descriptive statistics and intercorrelations for all of the instruments relevant for this contribution.

Table 2 Descriptive statistics and intercorrelations of the instruments used in the mediator analyses Correlations

1 2 3 4 5 6

Achievement values Well-being values Val. Math Val. German Grade Math Grade German

Mean

SD

Range

ICC

1

2

3

4

5

3.44 2.81 7.26 7.19 3.92 4.01

1.23 1.44 2.66 2.32 1.02 0.88

0-50 0-50 0-12 0-12 1-60 1-60

.00 .02 .07 .04 .09 .13

-.40** -.27** -.31** -.10** -.09**

-.22** -.29** -.09** -.15**

-.21** -.38** -.09**

-.04** -.24**

.37**

Note. Means, standard deviations, and correlations were calculated based on a single (EM-) imputed dataset. “Val. Math”: Valence of the subject Mathematics; “Val. German”: Valence of the subject German. ICC: Intraclass correlation coefficient. Original grades were reversed such that 6 is indicating the best grade and 1 the worst grade; *p * <.05; **p * <.01.

Table 2 also includes the variables’ intraclass correlation coefficients (ICCs) as an estimate of the non-independence of the data due to the nested data structure. For example, for Mathematics grades the ICC was .09, indicating that about 9% of the variance in the Mathematics grades was between classes. For grades in German, the ICC was .13, indicating that about 13% of the variance in the German grades was between classes. The amount of variance between classes was significant for all variables (all χ2≥68.21, all p<.01) except for

208

S. FRIES, S. SCHMID, M. HOFER

the value orientation prototypes (achievement value orientation prototype: χ2=34.22, p=.51; well-being value orientation prototype: χ2=48.61, p=.06). These results indicate that parameter estimation can be improved by taking the nested data structure into consideration when testing the mediation hypotheses. Figures 1 and 2 give the results for the mediator analyses on valences mediating the influence of achievement value orientation on school grades. 0.08** (.093); t=2.71 t Achievement values

Mathematics grade

-0.01 (-.008); t=-0.21; t p=.84 Achievement values

Mathematics grade 0.14***

0.60***

(.259); t=5.26 t

(.391); t=11.13 t

Valence Mathematics indirect effect: axb=0.08*** (.10); z=4.76

Figure 1. Mediation of achievement value orientation on Mathematics achievement Note. In this figure as in the following figures unstandardised coefficients are reported with standardised coefficients in brackets; **p<.05; **p * <.01; ***p * <.001.

0.06* (.09); t=2.06 t Achievement values

German grade

0.01 (.01); t=0.25; t p=.81 Achievement values

German grade 0.15***

0.36***

(.31); t=7.50 t

(.24); t=5.06 t Valence German

indirect effect: axb=0.05*** (.07); z=4.20

Figure 2. Mediation of achievement value orientation on German achievement

As presented in the upper parts of the figures, achievement value orientation was related to grades. The figures show unstandardised regression coefficients (with standardised regression coefficients in brackets): A unit of increase in achievement value orientation was related to an increase in grades of .08 for Mathematics and of .06 in German, respectively. Thus, the mean difference between a student who reports that he or she is “very similar” (5) to the achievement value orientation prototype and a student who reports that he or she is “not similar at all” (0) was .40 grade units for Mathematics and .30 grade units for German. Achievement value orientation was also related to the valence of Mathematics as well as to the valence of German. An increase by one unit in the achievement value orientation resulted in an increase of .14 in the valence of Mathematics and .15 in the valence of German, respectively. Furthermore, valences were strongly related to grades when achievement value

VALUE ORIENTATION, VALENCES, AND ACHIEVEMENT

209

orientation was controlled for. An increase of one unit in valence was related to an increase of .60 in grade in Mathematics and to an increase of .36 in grade in German, respectively. In order to establish complete mediation, the relationship between value orientation and grade should disappear when controlling for the valences of the two school subjects, the supposed mediator. As can be seen from the figures, this was actually the case. Taken together, the analyses indicated a complete mediation of the influence of achievement value orientation on grade by the valences of the respective domains. As the Sobel-Test (Sobel, 1982) showed, the indirect effect, calculated as the product of the indirect paths a and b, was significant in both cases (Mathematics: z=4.76, p<.001; German: z=4.20, p<.001)3. Taken together achievement value orientation and valences explained 15.9% of variance in Math grades and 6.4% of variance in German grades on the individual level. Figures 3 and 4 give the results for the mediator analyses on valences mediating the influence of well-being value orientation on school grades. -0.05** (-.07); t=-1,89; t p=.06 Well-being values

Mathematics grade

0.01 (.019); t=0.63; t p=.53 Well-being values

Mathematics grade -0.10***

.60***

(-.21); t=-6.92 t

(.39); t=12.72 t

Valence mathematics indirect effect: axb=-0.06*** (-.08); z=-6.03

Figure 3. Mediation of well-being value orientation on Mathematics achievement

-0.07** (-.12); t=3.38 t Well-being values

German grade

-0.03 (-.06); t=-1.50; t p=.13 Well-being values

German grade -0.12***

.34***

(-.29); t=-7.38 t

(-.23); t=5.08 t

Valence mathematics indirect effect: axb=-0.04*** (-.07); z=-4.18

Figure 4. Mediation of well-being value orientation on German achievement

Well-being value orientation was significantly related to grades in German, too. An increase by one unit in well-being value orientation resulted in a decrease of the grade by .07 units. For the domain of Mathematics, the relation between well-being value orientation and grade fell short to reach significance (unstandardized coefficient: -0.05, t=-1.89, p=.06). Also, well-being value orientation was related to the valences of Mathematics and German. An increase of one unit in well-being value orientation resulted in a decrease of -.10 in valence of

210

S. FRIES, S. SCHMID, M. HOFER

Mathematics and a decrease of -.12 in valence of German, respectively. Again, both valences were connected to the respective grades. An increase of one unit in the valence of Mathematics resulted in an increase of .60 in Mathematics achievement and an increase of one unit in the valence of German resulted in an increase of .36 in German achievement, respectively. When controlling for the valences, there were no longer significant relations between value orientation and grades. Taken together, the analyses again indicated a complete mediation of the influence of well-being value orientation on grades by the respective valences of the domains as well. The indirect effects for both school subjects were significant (Mathematics: z=-6.03, p<.001; German: z=-4.18, p<.001.). Taken together, well-being value orientation and valences explained 16.0% of variance in Math grades and 6.6% of variance in German grades on the individual level.

Discussion An analysis of the relation between value orientations, valences, and academic achievement was presented. Value orientations were related to academic achievement. Thus, our study showed – unlike other studies in the field (e.g., Feather, 1988) – a direct relation of value orientation to a behavioral outcome measure in an educational setting. Value orientations were also related to the valences attached to the different school subjects. Achievement value orientation was positively related to the valences, whereas well-being value orientation was negatively associated. Consistent with findings from other studies (e.g., Schiefele, 1999), valences were related to academic achievement. Furthermore, there seems to be an indirect effect of value orientation on achievement mediated by valences, but no direct effect. If one controlled for valences, a direct relation between value orientation and achievement was no longer found, and a Sobel-Test for mediation turned out to be significant. Taken together, the data were consistent with the assumption that the relation between value orientation and academic achievement is completely mediated by the valences attached to the different school subjects. Value orientations were only weakly related to academic achievement. This might cast doubt of the relevance of our results. However, we found value orientations to be a factor lying behind valences which themselves relate much stronger to grades. As such, our study contributes to an understanding of the origin of the valences attached to school subjects. In addition, positive evaluations of academic domains – as they are expressed in high valences – represent a goal of schooling on its own. Therefore, the reported positive relation between value orientation and valences is an essential finding, since valences can be considered not only as a mediator but also as an important outcome variable within scholastic settings. Two limitations of the reported study need to be considered. Value orientations were measured using single prototype items. Although these prototype items consisted of comprehensive descriptions, future studies should operationalise value orientations by means of more prototypes combined with different items. Furthermore, achievement was solely assessed by grades. However, grades do not only result from academic achievement but also from other factors such as conformity to social norms (cf. Pansu, Bressoux, & Louche, 2003). Since the tendency to conform to social norms might also be associated with achievement values, future research should also consider standardised achievement tests. The results of our study can be interpreted in line with a model proposed by Feather (1990, 1999). Feather states that values are motivationally relevant by inducing valences. According to his model, values serve as guidelines for the evaluation of behavior, objects, and events. It is, therefore, likely that values are also connected with valences attached to various school subjects. Usually, school offers mainly opportunities for achievement and hence for the expression of achievement value orientations. Thus, whenever such opportunities are encountered a student’s achievement or well-being value orientation may guide his or her evaluations and decisions. Of course it is unlikely that value orientations are the only

VALUE ORIENTATION, VALENCES, AND ACHIEVEMENT

211

determinant of valences. The finding that students differ in the valences they associate with different school subjects indicates that more specific factors must be at work as well. Such factors could be teacher behaviors, student competencies, self-efficacy, or epistemological beliefs about differences in school subjects (cf. Eccles & Wigfield, 2002). So far, little is known about the process of valence induction on the basis of values. Feather (1990) argues in favor of mechanisms similar to those suggested for attitudes (Fazio, 2001). According to Fazio, strong attitudes can be activated automatically when a person gets confronted with the corresponding attitude object. Once an attitude is activated, it will influence the individuals’ subjective definition of the object or event. The cognitive-affective evaluation of the object or event contains its representation in terms of its desirability or – in other words – its valence (Feather, 1990). If situational norms allow for a behavior congruent with the individual’s attitudes, such behavior will be shown. Similar to attitudes, values also have to be activated in order to influence the subjective definition of any objects or events encountered by the individual. Such activation is likely to occur whenever value-related objects, situations, or events are in the focus of attention (Verplanken & Holland, 2002). For example, a discussion among students about some behavior of a highly restrictive teacher might activate the value of self-direction. This activation will then in turn influence their subjective definition of the behavior shown by the teacher. Based on the principles of judging valence (e.g., value supportiveness, cf. Brendl & Higgins, 1996), the activated values influence the valence attached to the object or event. Values might also become activated in the case of a motivational conflict. Whenever a situation involves a choice between alternatives related to different values, these values might become activated. This is the case if a student has to decide between studying or going out with friends, a situation in which values constituting achievement value orientation and values constituting well-being value orientation will possibly be activated and influence the individual’s definition of the conflicting alternatives. In accordance with this assumption, Fries et al. (2005) found a relation between value orientation and decisions in hypothetical situations in which students had to choose between an academic and a leisure opportunity. Finally, values can be activated whenever the self-concept is activated. The reason for this is that values represent a central structure in the self (Verplanken & Holland, 2002). Especially school offers many opportunities for activating the self-concept (Kessels, 2005) and, hence, for activating the values being held by the individual. In the long run, such activations of values influencing the momentary judgments of objects and events should result in consistent and rather stable valence judgments attached to domains of academic interest. A potential criticism of our study might address our assumption that values influence achievement and not the other way round. One might argue that good grades result in approval of achievement values and disapproval of well-being values. However, such a direction seems unlikely considering the assumed mediation. The valences for the school domains are much more specific than the value orientations. Such specific valences should not be capable of inducing the respective value orientations, but the value orientations should be capable of inducing the respective valences. Whereas a change in an attitude does not affect the underlying value structure, a change in the value system should have an impact on the attitudes expressed by an individual (cf. Rokeach, 1973). However, a longitudinal approach to test the direction of influences is still found wanting. Taking a broader perspective, the construct of value orientation might play a crucial role in future motivational analysis of learning. Boekaerts (2003; see also Hofer & Peetsma, 2005) argues that the learning individual must be seen in its cultural and societal context, since these contexts have been shown to influence the student’s construal of learning and teaching as well as the learning process itself (cf. Greenfield, Keller, Fuligni, & Maynard, 2003). Motivational analysis restricted to domain specific measures cannot achieve such demands: Context is already considered, but only in the sense of directly encountered characteristics such as classmates or teachers, but not in the sense of societal influences (cf. Urdan, 1999). Value orientations seem to be a highly useful construct for the motivational analysis of learning processes and outcomes, since it also takes into account the cultural and societal context

212

S. FRIES, S. SCHMID, M. HOFER

(Boehnke, 2005). Usually, value orientations are measured on an individual level; however, they are often seen as a reflection of a nation’s culture as well as economic and political processes (Hofer & Peetsma, 2005). In future studies with students from different nations, the effects of such background variables on motivation to learn should be explored. This will be a step beyond today’s purely individual perspective on motivation to learn to a more multifaceted perspective in which societal aspects are also considered.

Appendix Descriptions of students representing value prototypes (Achievement) For John it is mainly important to achieve something in life. He has clear goals he consequently tries to reach. He struggles even through uncomfortable tasks if his goal is important to him. Then he puts other activities back. John wants to find a good job in the future in which he earns much money and can afford everything he would like to have. (Well-being) For Simon it is above all important to have fun in life and to experience a lot. His favourite way of spending his time is with his friends. They are very important to him. He loves diversion and spontaneous actions. Therefore, he avoids committing himself to something or to plan for a longer period of time. If it were according to him life would only consist of free time.

Notes 1

The rather low consistency of the German scale is not surprising if one considers that our valence scales refer to different incentives in a student’s psychological environment which do not necessarily co-occur. As such, our valence measure has features in common with so-called composite variables (Bollen & Lennox, 1991). The indicators of a composite variable are not caused by an internal attribute of a person; a composite variable rather results from different and possibly independent sources measured by its indicators. For example, students describing German as fun need not to agree on describing German as being important for their later lives. Following this reasoning, the low homogeneity of the valence scales is not necessarily an indicator of insufficient measurement (Bollen & Lennox, 1991; Streiner, 2003).

2

In principle, the level-1 slope could be modeled as random coefficient, too. In such a case it would be predicted from another level-2 equation. However, this would render the calculation of the indirect effect difficult. At present, there is only an ad hoc method for handling this problem (see Kenny, Korchmaros, & Bolger, 2003).

3

Due to the estimation with HLM, axb does not always equal c-c’. However, the difference is usually small (Krull & MacKinnon, 2001).

References Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 11731182. Boehnke, K. (2005). Value orientations in relation to mathematical self-esteem: An exploratory study of their role in mathematical achievement among German, Israeli, and Canadian 14-year-olds. European Journal of Psychology of Education, 20, 227-241. Boekaerts, M. (2003). Adolescence in Dutch culture: A self-regulation perspective. In F. Pajares & T. Urdan (Eds.), Adolescence and education: International perspectives on adolescence and education (vol. 3, pp. 101-124). Greenwich, CT: Information Age Publishing. Bollen, K., & Lennox, R. (1991). Conventional wisdom on measurement: A structural equation perspective. Psychological Bulletin, 110, 305-314. Brendl, C.M., & Higgins, E.T. (1996). Principles of judging valence: What makes events positive or negative? In M.P. Zanna (Ed.), Advances in experimental social psychology (vol. 28, pp. 95-160). San Diego, CA: Academic Press.

VALUE ORIENTATION, VALENCES, AND ACHIEVEMENT

213

Bubeck, M., & Bilsky, W. (2004). Value structure at an early age. Swiss Journal of Psychology, 63, 31-41. Cantor, N. (1994). Life task problem solving: Situational affordances and personal needs. Personality and Social Psychology Bulletin, 20, 235-243. Deci, E.L., & Ryan, R.M. (Eds.). (2002). Handbook of self-determination research. Rochester, New York: University of Rochester Press. Eccles, J.S., & Wigfield, A. (2002). Motivational beliefs, values, and goals. Annual Review of Psychology, 53, 109-132. Fazio, R.H. (2001). On the automatic activation of associated evaluations: An overview. Cognition and Emotion, 15, 115-141. Feather, N.T. (1988). Values, valences, and course enrollment: Testing the role of personal values within an expectancyvalence framework. Journal of Educational Psychology, 80, 381-391. Feather, N.T. (1990). Bridging the gap between values and actions: Recent applications of the expectancy-value model. In E.T. Higgins & R.M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (vol. 2, pp. 151-192), New York: Guilford Press. Feather, N.T. (1995). Values, valences, and choice: The influence of values on the perceived attractiveness and choice of alternatives. Journal of Personality and Social Psychology, 68, 1135-1151. Feather, N.T. (1999). Values, achievement, and justice: Studies in the psychology of deservingness. New York: Kluwer Academic/Plenum. Fries, S., Schmid, S., Dietz, F., & Hofer, M. (2005). Conflicting values and their impact on learning. European Journal of Psychology of Education, 20, 259-273. Graham, J.W., Cumsille, P.E., & Ekek-Fisk, E. (2003). Methods for handling missing data. In J.A. Schinka & W.F. Velicer (Eds.), Methods in Psychology (pp. 87-114). New York: John Wiley & Sons. Greenfield, P.M., Keller, H., Fuligni, A., & Maynard, A. (2003). Cultural pathways through universal development. Annual Review of Psychology, 54, 461-490. Himmelfarb, S. (1993). The measurement of attitudes. In A.H. Eagly & S. Chaiken (Eds.), The psychology of attitudes (pp. 23-87). Forth Worth, TX: Harcourt Brace Jovanovich. Hitlin, S., & Piliavin, J.A. (2004). Values: Reviving a dormant concept. Annual Review of Sociology, 30, 359-393. Hofer, M., & Peetsma, T. (2005). Societal values and school motivation. Students’ goals in different life domains. Introduction to the special issue. European Journal of Psychology of Education, 20, 203-208. Hofer, M., Schmid, S., Fries, S., Dietz, F., Clausen, M., & Reinders, H. (2007). Individual values, motivational conflicts, and learning for school. Learning and Instruction, 17, 17-28. Inglehart, R. (1997). Modernization and postmodernization. Princeton, NJ: Princeton University Press. Inglehart, R., & Baker, W.E. (2000). Modernization, cultural change, and the persistence of traditional values. American Sociological Review, 65, 19-51. Kenny, D.A., Kashy, D.A., & Bolger, N. (1998). Data analysis in social psychology. In D. Gilbert, S.T. Fiske, & G. Lindzey (Hrsg.), Handbook of social psychology (4th ed., vol. 1, pp. 233-265). New York: McGraw-Hill. Kenny, D.A., Korchmaros, J.D., & Bolger, N. (2003). Lower level mediation in multilevel models. Psychological Methods, 8, 115-128. Kessels, U. (2005). Fitting into the stereotype: How gender-stereotyped perceptions of prototypic peers relate to liking of school subjects. European Journal of Psychology of Edcuation, 20, 309-323. Krull, J.L., & MacKinnon, D.P. (2001). Multilevel modeling of individual and group level mediated effects. Multivariate Behavioral Research, 26, 249-277. Lens, W., Lacante, M., Vansteenkiste, M., & Herrera, D. (2005). Study persistence and academic achievement as a function of the type of competing motivational tendencies. European Journal of Psychology of Education, 20, 275-287. Lewin, K. (1951). Field theory in social science: Selected theoretical papers by Kurt Lewin (D. Cartwright, Ed.). New York: Harper & Row. Maio, G.R., & Olson, J.M. (2000). What is a “value-expressive” attitude? In G.R. Maio & J.M. Olson (Eds.), Why we evaluate: Functions of attitudes (pp. 249-269). Mahwah, NJ: Erlbaum.

214

S. FRIES, S. SCHMID, M. HOFER

Maio, G.R., Olson, J.M., Bernard, M.M., & Luke, M.A. (2003). Ideologies, values, attitudes, and behavior. In J. DeLamater (Eds.), Handbook of social psychology (pp. 283-308). New York: Plenum. Marini, M.M. (2000). Social norms and values. In E.F. Borgatta & R.J.V. Montgomery (Eds.), Encyclopedia of sociology (pp. 2828-2840). New York: Macmillan. Oviada, S. (2003). Suggestions of the postmodern self: Value changes in American high school students, 1976-1996. Sociological Perspectives, 46, 239-256. Pansu, P., Bressoux, P., & Louche, C. (2003). Theory of the social norm of internality applied to education and organizations. In N. Dubois (Ed.), A social cognitive approach to social norms (pp. 195-230). London: Routledge. Raudenbush, S.W., & Bryk, A.S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage. Raudenbush, S.W., Bryk, A.S., Cheong, Y.F., & Cogdon, R.T. (2000). HLM 5: Hierarchical linear and nonlinear modeling. Lincolnwood, IL: Scientific Software International, Inc. Rokeach, M. (1973). The nature of human values. New York: Free Press. Rubin, D.B. (1987). Multiple imputation for nonresponse in surveys. New York: Wiley. Schiefele, U. (1999). Interest and learning from text. Scientific Studies of Reading, 3, 257-280. Schmid, S., Hofer, M., Dietz, F., Reinders, H., & Fries, S. (2005). Value orientations and everyday action conflicts: An interview study. European Journal of Psychology of Education, 20, 243-258. Schwartz, S.H. (1992). Universals in the content and structure of values: Theory and empirical test in 20 countries. In M. Zanna (Ed.), Advances in experimental social psychology (vol. 25, S. 1-65). New York: Academic Press. Schwartz, S.H., Melech, G., Lehmann, A., Burgess, S., Harris, M., & Owens, V. (2001). Extending the cross-cultural validity of the theory of basic human values with a different method of measurement. Journal of Cross-Cultural Psychology, 32, 519-542. Seligman, C., Olson, J.M., & Zanna, M.P. (Eds.) (1996). The psychology of values: The Ontario Symposium (vol. 8). Mahwah, NJ: Erlbaum. Shafer, J.L. (1999). NORM: Multiple imputation of incomplete multivariate data under a normal model, version 2 [Computer Software for Windows 95/98/NT]. Retrieved January 12, 2005 from http://www.stat.psu.edu/~jls/ misoftwa.html. Sobel, M.E. (1982). Amsymptotic confidence intervals for indirect effects in structural equation models. In S. Leinhardt (Ed.), Sociological methodology (pp. 290-312). San Francisco: Jossey-Bass. Streiner, D.L. (2003). Being inconsistent about consistency: When coefficient alpha does and doesn’t matter. Journal of Personality Assessment, 80, 217-222. Urdan, T. (Ed.). (1999). Advances in motivation and achievement: The role of context (vol. 11). Stamford, CT: JAI Press. Verplanken, B., & Holland, R.W. (2002). Motivated decision making: Effects of activation and self-centrality of values on choices and behavior. Journal of Personality and Social Psychology, 82, 434-447. Weber, M. (1949). Max Weber on the methodology of the social sciences (E.A. Shils & H.A. Finch, Eds. & Trans.). New York: Free Press. Wigfield, A., & Eccles, J. (2000). Expectancy-value theory of achievement motivation. Contemporary Educational Psychology, 25, 68-81.

Les orientations valeurs sont supposées influencer l’apprentissage. Nous croyons que cette influence est médiate par les valences des disciplines scolaires. Dans un sondage (704 étudiants de 36 classes) les orientations des valeurs de réussite et de bien-être ont été mesurées. En outre, les étudiants ont indiqué leur valence pour les disciplines d’allemand et de mathématique et communiqué leurs notes respectives. Pour tenir compte de la structure imbriquée des dates, les hypothèses

VALUE ORIENTATION, VALENCES, AND ACHIEVEMENT

215

médiatiques ont été analysées en utilisant le Modèle Hiérarchique Linéaire avec des models intercept-only. Les notes ont été prédites de façon significative par les orientations valeur. L’analyse de médiation indique une dépendance strictement médiate par les valences des différentes disciplines scolaires.

Key words: Academic achievement, Hierarchical linear model, Mediation analysis, Valence, Value orientation.

Received: December 2005 Revision received: March 2006

Stefan Fries. Lehrstuhl für Erziehungswissenschaft II (Pädagogische Psychologie), University of Mannheim, D-68131 Mannheim, Germany. E-mail: [email protected]; Web site: www.ew2.uni-mannheim.de/fries Current theme of research: Value orientation and motivation to learn. The impact of attractive alternatives on learning. Effects of quality information on media-based learning. New media in higher education. Supporting self-regulated learning. Most relevant publications in the field of Psychology of Education: Fries, S. (2006). Zu Defiziten und möglichen Weiterentwicklungen aktueller Theorien der Lernmotivation [Contemporary Theories on Motivation to Learn: Deficits and Future Prospects]. Zeitschrift für Pädagogische Psychologie, 20, 73-83. Fries, S., & Dietz, F. (2007). Learning with temptations present: The case of motivational interference. Journal of Experimental Education, 76(1). Fries, S., Horz, H., & Haimerl, C. (2006). Pygmalion in media-based learning: Effects of quality expectancies on learning outcome. Learning and Instruction, 16, 339-349. Fries, S., Schmid, S., Dietz, F., & Hofer, F. (2005). Conflicting values and their impact on learning. European Journal of Psychology of Education, 20, 259-274.

Sebastian Schmid. Lehrstuhl für Erziehungswissenschaft II (Pädagogische Psychologie), University of Mannheim, D-68131 Mannheim, Germany. E-mail: [email protected]; Web site: www.ew2.uni-mannheim.de/schmid Current theme of research: Value orientation and motivation to learn. Curiosity. Epistemological beliefs. Most relevant publications in the field of Psychology of Education: Fries, S., Schmid, S., Dietz, F., & Hofer, M. (2005). Conflicting values and their impact on learning. European Journal of Psychology of Education, 20, 259-273. Hofer, M., Schmid, S., Fries, S., Dietz, F., Clausen, M., & Reinders, H. (2007). Individual values, motivational conflicts, and learning for school. Learning and Instruction, 17, 17-28. Schmid, S., Hofer, M. Dietz, F., Reinders, H., & Fries, S. (2005). Value orientations and action conflicts in students’ everyday life: An interview study. European Journal for the Psychology of Education, 20, 243-257.

216

S. FRIES, S. SCHMID, M. HOFER

Manfred Hofer. Lehrstuhl für Erziehungswissenschaft II (Pädagogische Psychologie), University of Mannheim, D-68131 Mannheim, Germany. E-mail: [email protected]; Web site: www.ew2.uni-mannheim.de/hofer Current theme of research: Value orientation and motivation to learn. Adolescent Development. Service Learning. Parent-Child-Interaction. Most relevant publications in the field of Psychology of Education: Hofer, M. (1999). Community Service and Social Cognitive Development in German Adolescents. In M. Yates & J. Youniss (Eds.), Roots of Civic Identity. International Perspectives on Community Service and Activism in Youth (pp. 114-134). Cambridge: University Press. Hofer, M. (2004). The role of discourse in the transformation of Parent-Adolescent Relationships. In A.-N. PerretClermont, C. Pontecorvo, L.B. Resnick, T. Zittoun, & B. Burge (Eds.), Joining society: Social Interaction and Learning in Adolescence and Youth (pp. 241-251). Cambridge: University Press. Hofer, M. (2005). Goal-dependent perception in relations between teachers and students. In P.M. Denicolo & M. Kempf (Eds.), Teacher thinking and professional action (pp. 121-128). London: Routledge. Hofer, M., Sassenberg, K., & Pikowsky, B. (1999). Discourse asymmetries in adolescent daughters’ disputes with mothers. International Journal of Behavioral Development, 22, 1001-1022. Hofer, M., Schmid, S., Fries, S., Dietz, F., Clausen, M., & Reinders, H. (2007). Individual values, motivational conflicts, and learning for school. Learning and Instruction, 17, 17-28.