«A CROSS-CULTURAL STUDY OF THE RELATIONSHIPS BETWEEN EPISTEMOLOGICAL BELIEFS AND MORAL JUDGMENT AS A PSYCHOLOGICAL FOUNDATION FOR MORAL EDUCATION by ...»
The Defining Issues Test has two validity checks imbedded within the test. The first is an M score that is derived from the number of statements selected that are pretentious and mostly meaningless. There is an overall correlation of zero between M scores and P scores (Rest, 1990). Selection of these answers as the first or second most important consideration in making a moral decision suggests that the subject is making choices based on lofty sounding verbiage rather than personal values. These items do not represent any stage of moral reasoning, and in fact, selection of one of these items in the first four positions is contraindicated by test instructions. When the M score is greater than eight, the authors recommend invalidating the entire test (Rest, 1990).
A further internal validity test involves the consistency of a person’s choices. The instrument requires the subject to rank 12 questions or statements in their relative order of importance. The subject is not being consistent if an item ranked as “little importance” is rated as a subject’s first or second choice and selected ahead of items rated “very important.” If there are inconsistencies on more than two stories, or if the number of inconsistencies on any one story exceeds eight, Rest (1990) recommends invalidation and exclusion of that person’s protocol.
An additional inconsistency check regards a subject’s lack of discrimination.
When a test protocol shows most items ranked the same, there is a suspicion that the subject is not taking the test seriously. If a story has more than 9 items rated the same,
current study, all of these validity measures were utilized according to Rest’s (1986,
The short form of Rest’s (1979) DIT was used in this study. The P score from the short form correlates (r =.93) with the full DIT and separate stage score correlations using the two forms range from.57 (Stage 5b) to.88 (both Stage 5a and 6) (Rest, 1990).
Also, the Korean version of the DIT (Moon, 1994) was used for assessing levels of principled moral reasoning in the Korean sample. A variety of studies have been completed on moral judgment development using the Korean version of the DIT (Moon, 1994). However, only two studies reported test-retest stability and internal consistency of the Korean DIT (Park 1989, test-retest r =.47, α =.64; Moon 1994, α =.52). These tend to be somewhat lower than the original DIT. The short form of the Korean version of the DIT consists of three separate dilemmas (i.e., the Husband, Prisoner, and Doctor stories).
For the collection of the U.S. data, the equivalent stories used for Korean data were selected (i.e., the Heinz, Prisoner, and Doctor stories).
Epistemic Beliefs Inventory (EBI) To measure students’ epistemological beliefs, a 32-item Epistemic Beliefs Inventory (see Appendix B) was used (Bendixen et al., 1998). As previously discussed, Schraw et al. (2002) reported that the EBI had several advantages over an exploratory analysis of the Epistemological Questionnaire (EQ) developed by Schommer (1990).
First, Schommer’s EQ yielded a large number of potentially interpretable factors, each accounting for a relatively small share of total sample variation. In comparison, the five factors identified by the EBI provided a close fit with the five epistemological
explained 35.5% of total variation, while the EBI explained 60% of total sample variation. A one-month replication led to values of 39% and 64% respectively. Third, the EBI had better predictive validity than the EQ when correlated with a test of reading comprehension.
Each of the 32 items was written as a grammatically simple statement to which individuals responded using a five-point Likert-type scale ranging from “strongly disagree” (1) to “strongly agree” (5). Individuals made their ratings by circling the number that most closely reflected their agreement with the statement.
To measure Korean students’ beliefs about knowledge and knowing, a new Korean translation of Bendixen, Schraw, and Dunkle’s (1998) 32-item Epistemic Beliefs Inventory was developed. In order to minimize possible linguistic and cultural discrepancies between the original scale and the new scale, two psychological testing procedures necessary for a reliable and valid test were performed: (a) back-translation, and (b) principle factor analysis as the extraction method. Principle factor analysis was used to examine whether the theoretical factors of the Epistemic Beliefs Inventory (EBI) could be recovered in the Korean translation of the instrument.
The back-translation method is required when transposing psychometric instruments from one language to another in order to assure equivalence between an original scale and its translation (Brislin, 1970). For this study, the EBI was translated using a three-stage translation/back-translation/translation procedure: (a) the translation from English to Korean was performed by the investigator, (b) the back-translation from Korean to English was performed by a fluent English speaking Korean researcher in the
detect any discrepancies by an American researcher in the field of social science education. As a result, the original EBI and the Korean EBI were sufficiently equivalent and translation error was not detected. Additionally, principal factor analysis via SPSS graduate pack 10.0 (SPSS Inc., 2002) was performed to examine whether the theoretical factors of the EBI could be recovered in this Korean translation of the instrument.
Sixty undergraduates enrolled in a cyber-ethics class at Chuncheon National University of Education were involved in the pilot study. Prior to any factor analysis, it is necessary to determine the appropriateness of this type of analysis. The KMO (KaiserMeyer-Olkin Measure of Sampling Adequacy) is a measure of how amenable the matrix is to factoring (Gorsuch, 1983). Specifically, it compares the correlations among pairs of variables to their correlations when the effects of the other variables are removed or partialed out. Reasonably large values are needed for a good factor analysis, so the KMO measure should be greater than.60 for the factor analysis to proceed (Gorsuch, 1983;
Hair, Anderson, & Tatham, 1987). Results for all factors used were.70 or higher, indicating adequate correlation among items.
It was also necessary to test the hypothesis that the correlation matrix came from a population of variables that are independent. Attempting to correlate independent items will, by definition, yield poor results. If this hypothesis is not rejected, the data are not appropriate for factor analysis. Bartlett’s test of sphericity of the residual covariance matrix tests the hypothesis that the correlation matrix is an identity matrix and therefore unsuitable for further analysis. In order to not accept this hypothesis, Bartlett’s test
Results of this analysis produced an approximate chi-square value of p.001 in all cases, supporting the rejection of the hypothesis and suggesting that the data set was a sample from a multivariate population. The KMO Measure of Sampling Adequacy and Bartlett’s test of Sphericity and significance (p.001) indicated that the necessary prerequisite conditions existed in order to proceed with the factor analysis.
The principle factor analysis as the extraction method yielded 12 factors with eigenvalues greater than one that explained 73% of the total sample variation. The researcher selected the first five observed factors for closer inspection to determine whether they corresponded to the five factors proposed by Bendixen et al. (1998; Schraw et al., 2002). These factors explained 43% of the total sample variation. Items with loadings greater than.30 were used to construct composite scores for each factor. The remaining seven factors included a couple of items but did not suggest clearly interpreted factors in the context of Schommer’s hypothesized five-factor model (1990) and Bendixen, Schraw, and Dunkle’s EBI (1998, Schraw et al., 2002). Factor labels, item-tofactor loadings, eigenvalues, and values of coefficient α for each of the five factors are shown in Table 3.2.
Table 3.2 Factor Structure of the Korean Translation of the Epistemic Beliefs Inventory (N = 60) Factor 1: Innate Ability (Eigenvalue = 3.
59; α =.64) Some people are born with special gifts and talents. (.74) Some people just have a knack for learning and others don’t. (.60) Smart people are born that way. (.43) Some people will never be smart no matter how hard they work. (.39) Factor 2: Quick Learning (Eigenvalue = 3.47; α =.31)
Factor 3: Omniscient Authority (Eigenvalue = 2.44; α =.36) People who question authority are trouble makers. (.39) Children should be allowed to question their parents’ authority. (.35) When someone in authority tells me what to do, I usually do it. (.33) Factor 4: Simple Knowledge (Eigenvalue = 2.21; α =.33) Too many theories just complicate things. (.65) The best ideas are often the most simple. (.33) You can study something for years and still not really understand it. (.42) It bothers me when instructors don’t tell students the answers to complicated problems.
(.31) Instructors should focus on facts instead of theories. (.30) Factor 5: Certain Knowledge (Eigenvalue = 1.92; α =.38) Truth means different things to different people. (.51) The moral rules I live by apply to everyone. (.37) Absolute moral truth does not exist. (.32) The five factors were labeled Innate Ability, Quick Learning, Omniscient Authority, Simple Knowledge, and Certain Knowledge. These factors were identical to the five epistemological dimensions hypothesized by Schommer (1990) and Schraw, Bendixen, and Dunkle’s (2002) findings (see Table 3.3). Each factor included at least three items with loadings in excess of.30.
Table 3.3 Factor Structure of The Epistemic Beliefs Inventory Factor 1: Omniscient Authority (Eigenvalue = 1.
63; α =.68) People should not question authority. (.73) Children should be allowed to question their parents’ authority. (.66) When someone in authority tells me what to do, I usually do it. (.62) Factor 2: Certain Knowledge (Eigenvalue = 1.63; α =.62) The moral rules I live by apply to everyone. (.72) What is true today will be true tomorrow. (.63).
Parents should teach their children all there is to know about life. (.50) Factor 3: Quick Learning (Eigenvalue = 1.47; α =.58)
Factor 4: Simple Knowledge (Eigenvalue = 1.43; α =.62) Instructors should focus on facts instead of theories. (.78) Too many theories just complicate things. (.57) Most things worth knowing are easy to understand. (.44) Factor 5: Innate Ability (Eigenvalue = 1.36; α =.62) How well you do in school depends on how smart you are. (.76) Smart people are born that way. (.56) Really smart students don’t have to work as hard to do well in school. (.30) From Schraw et al., “Development and validation of the epistemic belief inventory (EBI)” (pp. 261-275). In B. K. Hofer & P. R. Pintrich (Eds.), Personal epistemology, 2002, Mahwah, NJ: Erlbaum.
These findings indicated that the Korean translation of the EBI might yield the five epistemic dimensions identified by Bendixen, Schraw, and Dunkle’s EBI (1998;
Schraw et al., 2002) in a similar way. However, the results of this study also suggest three limitations. Firstly, the Korean translation of the EBI yielded 12 factors with eigenvalues greater than one, while the EBI provided a close fit with the five epistemic dimensions hypothesized by Schommer (1990). Secondly, the first five factors on the Korean translation of the EBI explained 43% of total variation, while the EBI explained 60% of total sample variation. Lastly, a comparison of internal consistency coefficients using Cronbach’s α indicated that the Korean translation of the EBI was less reliable than the EBI. These pilot study results may be due to the use of relatively few participants (i.e., sixty undergraduates). Therefore, after completing official data collection at three universities in Korea, the factor structure of the EBI was analyzed again using varimax factor analysis with a larger sample (n = 243).
The varimax solution yielded ten factors with eigenvalues greater than one and explained 58.1% of the total sample variation. The first five observed factors
Bendixen et al. (1998; Schraw et al., 2002). These factors explained 40.5% of the total sample variation. The internal consistency was equal to.65 for omniscient authority,.63 for certain knowledge and quick learning,.49 for innate ability, and.44 for simple knowledge. The internal consistency was low with simple knowledge and innate ability, although it was high with the overall 32-item questionnaire (alpha =.71). The factor structure, loadings, eigenvalues, and values of coefficient alpha for each of the five factors are reported in Table 3.4.
Table 3.4 Factor Structure of the Korean Translation of the Epistemic Beliefs Inventory (N = 241) Factor 1: Quick Learning (Eigenvalue = 3.