# «A CROSS-CULTURAL STUDY OF THE RELATIONSHIPS BETWEEN EPISTEMOLOGICAL BELIEFS AND MORAL JUDGMENT AS A PSYCHOLOGICAL FOUNDATION FOR MORAL EDUCATION by ...»

The purpose of this study was to explore whether cultural patterns exist in the relationships between epistemological beliefs and moral reasoning. Therefore, two sets of analyses were conducted. In the first, descriptive statistics and correlations among all variables were used to assess the relationships between moral reasoning and each of the eleven independent variables between Korean and U.S. college students. The second set of analyses consisted of all possible regressions that examined whether epistemological beliefs explain a substantial proportion of variation above and beyond other variables and which of five epistemological beliefs contributes substantially to higher levels of principled moral reasoning.

** Table 4.1 summarizes the means and standard deviations for all variables and results of the t-tests.**

Simple independent t-tests were used to assess the differences between Korean and U.S. college students in all variables. The results indicated that Korean college students tended to believe more strongly than U.S. college students that authorities have access to otherwise inaccessible knowledge, t(432) = -11.05, p.001, ability to learn is innate, t(432) = -4.05, p.001, and knowledge is simple, t(432) = p =.003. Korean students believed more strongly than their counterparts that knowledge is certain, but the difference was not statistically significant. U.S. students believed more strongly than their counterparts that learning is quick, t(432) = 5.9, p

does not happen quickly. The analyses indicated that U.S. college students obtained significantly higher P scores (M = 49.20, SD = 14.86) in comparison to the Korean students (M = 43.01, SD = 14.37).

As recommended by Cohen (1988), effect sizes with respect to each of the independent variables were computed. Cohen’s criteria for evaluating the effect sizes suggest that the effect size pertaining to omniscient authority was quite large and quick learning indicated a medium effect size. Also, the effect sizes pertaining to P scores and syllogisms approached moderate levels, whereas those pertaining to simple knowledge and innate ability were small.

The Korean Sample Of the five epistemological variables, three were significantly related to P scores (see Table 4.2). Those variables were omniscient authority, certain knowledge, and quick learning. Scores high on those dimensions were correlated negatively with P scores, indicating that higher levels of principled moral reasoning were associated with a more sophisticated and presumably less conventional, epistemological belief system.

shown a negative relationship between the acceptance of authority and P scores (Bendixen et al., 1998; Curtis et al., 1988; Haan, Smith, & Block, 1968; Presley, 1985);

the findings were also consistent with those of Walker et al. (1991), who reported that the DIT scores increased as epistemological beliefs measured on a unidimensional scale became more sophisticated.

A statistically significant association was noted between GPA and the P score (r =.33, p.01). This result supports the view that educational achievement has a significant correlation with the P score (Ji, 1997; Kohlberg, 1969; Rest, 1986). Age was significantly correlated with P score, but this correlation (r =.14) was very small. On the other hand, Gilligan’s (1982) charge of gender-bias in Kohlberg’s model was not warranted by the present evidence. Male and female students were not significantly different in terms of their P scores (r = -.09, p.05). However, age, education, and GPA had statistically significant correlations with gender. A significant correlation between GPA and gender indicated that GPA was statistically higher for female students than for male students.

Significant correlations of age and education with gender indicated that age and education were statistically higher for male students than for female students. Education, major, and syllogistic reasoning had no significant correlations with P scores.

All possible regressions were used to compare the proportion of variance in the principled moral reasoning explained by each variable. Scale scores for each of the five epistemic dimensions consisted of the average among all items with a loading of.30 or higher on that dimension. The results were shown in ascending order, beginning with one-predictor equations and concluding with an eleven-predictor equation. At each stage,

descending order (see Table 4.3). For this study, the criterion mainly used was the maximum proportion of variance explained (R2), because it not only provides an important measure of effect size (Cohen, 1988), but also the largest R2 and the Cp statistic are related to each other, in fact (Draper & Smith, 1998).

As single predictors were used at this stage, the R2’s are, of course, the squared zero-order correlations of each predictor with the criterion. Omniscient authority was listed first because it has the highest R2 with P scores, whereas simple knowledge was listed last because its correlation with P scores is the lowest. Omniscient authority explained about 12.1% of the variance in P scores. GPA, the next predictor, explained 10.6% of the variance in P scores, followed by certain knowledge (10.3%).

Table 4.3 Values of R2 and Cp for All Possible Regressions (Korea) Number in Model R-Square C(p) Variables in Model

Moving on to the results with two predictors, the combination of OA and GPA appeared to be the best. OA and GPA accounted for about 20.4% of the variance in P scores. The next best (i.e., CK and GPA) accounted for about 1% less of the variance as compared with that accounted by OA and GPA. Two epistemological beliefs (CK and OA) accounted for about 17% of the variance.

together accounting for about 24.7% of the variance. The increment from the best subset of two predictors to the best subset of three was about 4%. Of the four-variable equations, the best combination was CK, OA, gender, and GPA, together accounting for about 27.3% of the variance. The increment from the best subset of three predictors to the best subset of four was about 2.6%. The combination among all eleven predictors accounted for about 30.3% of the variance in the P score. The increment from the best subset of four predictors to the best subset of all eleven was only 3%.

The analysis revealed that variables (OA and CK) from the five epistemological predictors explained a substantial proportion of the variance in P scores over and above the effects of gender, age, education, GPA, academic major, and syllogistic reasoning.

The combination of omniscient authority and certain knowledge accounted for about 17% of the variance. In other words, collectively, the two beliefs explained more variance in P scores than either gender, age, education, GPA, academic major, and syllogistic reasoning considered separately. However, the increment from the subset of omniscient authority and certain knowledge to the subset of the five epistemological variables was only about 1.6%. Therefore, these results indicated that simple knowledge, innate ability, and quick learning may be of little or no use for prediction.

The assessment of the relative importance of the five epistemological predictors addresses the issue about which variables contribute substantially to Korean college students’ moral reasoning. Results obtained from the largest R2 and the Cp statistic in an all possible regressions analysis with the Korean sample indicated that omniscient authority was the best discriminator among the five epistemological predictors, followed

omniscient authority and certain knowledge) parallels that of a previous study (Lee, 1995), the result of which had shown the association of omniscient authority and certain knowledge with academic writing among Korean graduate students.

The U.S. Sample

and show that the correlations between subjects’ P scores and evaluations of beliefs in omniscient authority (r = -.35, p.01), simple knowledge (r = -.31, p.01), and quick learning (r = -.18, p.05) were both significant and indicative of an inverse relationship.

A significant association was noted between GPA and the P score (r =.26, p.01). This result supports the view that educational achievement has a significant correlation with the P score (Ji, 1997; Kohlberg, 1969; Rest, 1986).

Gilligan’s (1982) charge of gender-bias in Kohlberg’s model was not warranted by the present evidence. Male and female students were not significantly different in terms of their P scores (r =.03, p.05). Also, age, education, major, and syllogistic reasoning had no significant relationships with P scores.

The results of an all possible regressions analysis on U.S. college students were reported in Table 4.5. Omniscient authority was the best predictor, explaining 12.3% of the variance in P scores. Simple knowledge, the next predictor, explained 9.4% of the variance in P scores, followed by GPA (about 6.6%) and quick learning (about 3.2%).

** Table 4.5 Values of R2 and Cp for All Possible Regressions (U.**

S) Number in Model R-Square C(p) Variables in Model

appeared to be the best. OA and GPA accounted for about 17.8% of the variance in P scores. The next best (i.e., SK and OA) accounted for about 1% less of the variance as compared with that accounted for by OA and GPA. Also, the combination of another two epistemological beliefs (QL and OA) accounted for about 14.7% of the variance.

Of the three-variable equations, the best combination was SK, OA, and GPA, together accounting for about 20.8% of the variance. The increment from the best subset of two predictors to the best subset of three was about 3%. Of the four-variable equations, the best combination was SK, QL, OA, and GPA, together accounting for about 21.7% of the variance. The increment from the best subset of three predictors to the best subset of four was about 1%. The combination among all eleven predictors accounted for about 23.4% of the variance in the P score. The increment from the best subset of four predictors to the best subset of all eleven was only 1.7%.

The analysis revealed that variables (OA, SK, and QL) from the five epistemological predictors explained a substantial proportion of the variance in P scores over and above the effects of gender, age, education, GPA, academic major, and syllogistic reasoning. The combination of omniscient authority, simple knowledge, and quick learning accounted for about 17.7% of the variance. In other words, collectively, the three beliefs explained more variance in P scores than either gender, age, education, GPA, academic major, and syllogistic reasoning considered separately. However, the increment from the subset of omniscient authority, simple knowledge, and quick learning to the subset of the five epistemological variables was only about 0.5%. Therefore, these

prediction.

The relative importance of omniscient authority, simple knowledge, and quick learning of the five epistemological predictors in moral reasoning provides support for findings in a series of studies (Bendixen et al., 1998; Curtis et al; 1988; Presley, 1985;

Walker et al., 1991). Bendixen et al.s’ findings (1998) indicated that simple knowledge, omniscient authority, and quick learning each explained a significant proportion of the variation in P scores over and above the effects of gender, age, education, and syllogistic reasoning. Walker et al. (1991) reported that P scores increased as epistemological beliefs measured on a unidimensional scale became more sophisticated and that epistemological beliefs are related to P scores even when other variables are removed from the equation.

The present study revealed similar results in the relationship among epistemological beliefs and moral reasoning between Korean and U.S. college students.

First, the results showed that the epistemological belief omniscient authority and GPA were the strongest predictors in Korean and U.S. college students’ P scores. Also, the analysis revealed that variables from the five epistemological predictors explained a substantial proportion of the variance in P scores over and above the effects of gender, age, education, GPA, academic major, and syllogistic reasoning. With the U.S. sample, the combination of omniscient authority, simple knowledge, and quick learning accounted for about 17.7% of the variance in P scores. With the Korean sample, the combination of omniscient authority and certain knowledge accounted for about 17% of the variance in P scores.

gender-bias in Kohlberg’s model was not warranted by the present evidence. Male and female students were not significantly different in terms of their P scores. Also, education, major, and syllogistic reasoning had no significant correlations with P scores.

With the Korean college students, age was significantly correlated with P score, but this correlation (r =.14) was very low.

The present study also revealed differences between the two groups. The results revealed that Korean college students who viewed the nature of knowledge as certain scored lower on the DIT, whereas U.S. students’ beliefs about certain knowledge had no significant relationship with P scores and accounted for little variance in P scores. On the other hand, U.S. college students who endorsed simple knowledge produced lower principled moral reasoning scores, whereas Korean students’ beliefs about simple knowledge had no significant relationship with P score and accounted for little variance in P scores.

This chapter begins with an introduction prior to a discussion of the results of this research. The results will be discussed and related to the research questions and the general purpose of the study. Following the discussion, there will be some conclusions presented regarding the findings and suggestions for their practical application. The chapter will conclude with limitations of this research.