Paper presented at the 9th International Conference on Motivation, Lisbon,
Portugal.
Can access this presentation: http://faculty-staff.ou.edu/G/Barbara.A.Greene-1/
Although the study of epistemological beliefs has been applied to important questions since the 1980s (e.g., Ryan, 1984; Hofer & Pintrich, 1997; Schommer, Crouse, & Rhodes, 1992; Schommer-Aikens, Mau, Brookhart, & Hutter, 2000), there is still little research on how beliefs about knowledge and learning help us understand student motivation (Hofer & Pintrich, 1997; Qian and Alvermann, 1995). Prior research has shown that at least some aspects of epistemic beliefs are related to learnerís achievement goals (Ravindran, Greene & DeBacker, in press), cognitive engagement and strategy use (Kardash & Howell, 2000; Ravindran, Greene & DeBacker, in press; Ryan, 1984; Schommer, Crouse, & Rhodes, 1992; Tsai, 1998), comprehension (Schommer, 1990; Schommer, et al., 1992), and achievement (Schommer, 1993; Schommer, Calvert, Gariglietti, & Bajaj, 1997; Schommer et al., 1992). The present study extends these findings by exploring previously unexamined aspects of epistemological beliefs that we believe are particularly relevant to instructional settings and their relationships with self-efficacy, motivation type (i.e., intrinsic and extrinsic from Deci & Ryan, 2000) and cognitive engagement.
Work in the area of personal epistemology recently seems to be converging around two categories of epistemological beliefs: beliefs about knowledge (that is, the structure certainty of knowledge) and beliefs about knowing (that is, the source and justification of knowledge) (Hofer, 2000; Pintrich, 2002). We believe that in order to understand the amotivation of some adolescents, we also need to examine beliefs about the purposes of schooling and the knowledge gained in school. To this end we developed the Beliefs about School Learning Questionnaire.
The decision to develop a new instrument was partially based on concerns
noted about current epistemic belief instruments (e.g., Clarebout, Elen,
Luyten, & Bamps, 2001; Duell & Schommer-Aikins, 2001; Hofer &
Pintrich, 1997) and on the perceived need to include dimensions that might
inform motivational differences better than the existing scales (Qian &
Alvermann, 1995). However, our main concern was to develop a measure of
those epistemological beliefs that would help us understand the role that
beliefs about school and school knowledge might play in the achievement
motivation of students, especially students who come from the minority
groups that are dropping out of high school in the US at rates higher than
Caucasian students (e.g., Hispanic and African-American students, NCES,
2000). Ogbu and colleagues (Ogbu, 1992; Fordham and Ogbu, 1986) have argued
that some minority groups develop an oppositional stance toward majority
culture institutions (i.e., schools) that is related to distrust in the
impact those institutions have on their group. Therefore we have developed
and validated an instrument to measure such beliefs.
The Beliefs about School Learning Questionnaire (BASLQ) has four subscales (see Table 1 for sample items). The constructed and commodity subscales identify the degree to which academic knowledge is viewed as either something which a student constructs while interacting with the material or a commodity that students obtain from sources such as teachers and text books without subjecting it to their personal interpretation. The remaining two subscales examine the degree to which academic knowledge is viewed as either a tool for empowerment or an attempt by socializing forces to subjugate the student. The empowerment scale contains items describing the view that academic knowledge is a tool for increasing oneís personal and interpersonal control. The subjugation scale contains items describing the view that academic knowledge represents an attempt by mainstream authorities to control studentsí knowledge and opinions and alienate them from their culture of origin. Importantly, empowerment and subjugation are not conceived as opposing ends of the same construct.
In regard to relationships between our beliefs scales and self-efficacy, motivation type and cognitive engagement, we expected that the subjugation subscale would predict amotivation and be predicted by extrinsic-introjected motivation (Deci & Ryan, 2000) and the commodity view of academic knowledge. Additionally, we examined the related question of whether subjugation predicts shallow "going through the motions" engagement. We also expected that the constructed view of knowledge would predict intrinsic motivation, self-efficacy and deep cognitive engagement.
Method
The participants were 147 undergraduate students at a large mid-western university who received course credit for volunteering. Although a large portion of these volunteers were from upper division educational psychology courses, about 35% were freshman and sophomores from a career exploration class. Approximately two thirds of the respondents were female. Approximately 22% of the participants were from minority groups, which is consistent with the racial/ethnic make-up of the university.
Instrumentation
Each respondent completed a packet consisting of five surveys.
All surveys were comprised of six-point Likert scales anchored with "Strongly
Disagree" and "Strongly Agree."
Beliefs about School Learning Questionnaire (BASLQ). The original number of items developed for the sub-scale of this new instrument were as follows: subjugation had 10 items; commodity had 12 items; constructed knowledge had 11; and empowerment had 10.
Self-efficacy. This four-item instrument measures student confidence in their ability to learn, understand, and master material encountered in their college classes. It has been reliable and valid in prior work (e.g., Greene & Miller, 1996)
The Academic Motivation Scale (AMS). The sub-scales used from the AMS (Vallerand et al., 1993) were amotivation, extrinsic-introjected, and intrinsic motivation.
The Epistemological Beliefs Inventory (EBI). This instrument was used to provide convergent validity evidence for the BASLQ. The EBI possesses five sub-scales: Omniscent Authority, Certain Knowledge, Quick Learning, Simple Knowledge, and Innate Ability (Bendixen et al., 1998).
Cognitive Engagement. This subscale assesses both deep cognitive engagement and shallow cognitive engagement . The items have been shown to be valid and reliable in prior work (e.g., Greene & Miller, 1996; Kardash, & Amlund, 1991).
Procedure
Volunteers completed the surveys in groups of 10 to 15. The average
completion time was approximately 20 minutes.
Results
Confirmatory factor analyses were performed on the items for each BASLQ sub-scale using LISREL 8.52 (Joreskog & Sörbom, 2002). We used the following fit indices: Root Mean Residual (RMR), with optimal values falling below .10; Goodness of Fit Index (GFI), Comparative Fit Index (CFI), and Norm Fit Index ó with values above .90 indicating good fit (Kline, 1998) Additionally, we used the ratio of Chi Sq/df, with ratios less that 3 being considered acceptable (Kline, 1998).
Inter-correlations Between Sub-scales of the BASLQ and Theoretically Related Constructs.
As expected theoretically, statistically significant positive
correlations were found between the commodity subscale and the EBI's subscales
for simple knowledge (r =.34), certain knowledge (r =.26), quick learning
(r =.20), and fixed ability to learn (r =.31). The constructed subscale
was expected to be negatively correlated with certain knowledge (r = -.27),
and quick learning (r = -.12), and simple knowledge (r = -.01, ns). See
Table 3 for all correlations.
The empowerment subscale, as expected, was positively correlated
with the constructed subscale (r =.58), self-efficacy (r = .19), deep cognitive
engagement (r =.27), and intrinsic motivation (r =.35). It was also correlated
with extrinsic motivation (r =.25) and shallow cognitive engagement (r
= -.15). The subjugation subscale was expected to be positively correlated
with the commodity subscale (r =.59), shallow cognitive engagement (r =.29),
and amotivation (r =.34). See Table 4 for all correlations.
Path Models
We tested four path models to explore the relationships among
the BASLQ scales and the motivation-related variables. Each of the four
path models tested utilized LISREL 8.52 (Joreskog & Sörbom, 2002).
Results are included in Table 2 as the interpretation of fit indices for
these models parallels that used in CFA.
Model 1
The intent of the first model was to predict deep cognitive engagement
using direct paths from self-efficacy and constructed knowledge, as well
as indirect paths from these variables mediated through their impact on
intrinsic motivation. Empowerment was also included in this model as being
predicted by constructed knowledge and predicting intrinsic motivation.
The overall model was supported as indicated by the fit indices shown in
Table 2. Except for the path from empowerment to intrinsic, all specified
paths were significant. Deep cognitive engagement was predicted significantly
by self-efficacy (ß = .31, t = 3.55), constructed knowledge (ß
= .20, t = 2.16) and intrinsic motivation (ß = .21, t = 2.20). Intrinsic
motivation was, in turn, predicted by constructed (ß = .41,
t = 4.66), but less so by self-efficacy (ß = .18, t = 2.06). Additionally
empowerment was significantly predicted by constructed (ß = .59,
t = 7.51)
Models 2 and 3
The intent of the next set of analyses was to compare two models
examining the inter-relationships between the epistemic constructs of commodity
and subjugation, the motivational constructs of extrinsic-introjected,
and shallow cognitive engagement.
Model 2 suggested that shallow cognitive engagement would be predicted
by direct paths from commodity, extrinsic-introjected, and subjugation.
Although the overall model had some support, as the fit indices in Table
2 show, the c 2 approached significance (p = .06) and the c 2/df was greater
than the optimal 3. This is likely to be a result of the non-significant
path from subjugation to shallow cognitive engagement (ß = .15, t
= 1.39).
Model 3 specified that shallow cognitive engagement would be predicted by direct paths from commodity ( ß = .33, t= 3.79), and extrinsic-introjected (ß = .29, t= 3.31), and that subjugation would be predicted by direct paths from these scales as well (commodity = ß.58, t= 7.47; extrinsic-introjected (ß = .33, t= 1.93, ns). The data fit this model better than Model 2, which had included the link between subjugation and shallow cognitive engagement. A good fit of this model to the data was indicated by the non-significant c 2 and strong fit indices shown in Table 2.
Model 4
This model tested amotivation as predicted by a direct, positive path
from extrinsic- introjected motivation, a direct, negative path from intrinsic
motivation, and a positive indirect path from commodity to subjugation
and from subjugation to amotivation. Although all paths were in the expected
direction in terms of negative or positive relationships, but only the
paths from commodity to subjugation and from subjugation to amotivation
were significant (ß = .59, t = 7.37, and ß = .35, t = 3.82,
respectively). Despite the lack of significance in the other paths, this
indirect path predicting amotivation from commodity as mediated through
subjugation was strong enough to support the model with a c 2 value of
5.67 (non-significant) and acceptable fit indices as shown in Table 2.
Discussion
Table 3
Correlations between the EBI subscales and the Beliefs about School
Learning Questionnaire
1 2 3 4 5 6 7 8 9
1. Fixed ability 1
2. Omn.authority -.013 1
3. Quick .513** -.06 1
4. Simple .27** .33** .34** 1
5. Certain .07 .25** .16* .17* 1
6. Subjucation .435** -.03 .37** .28** .07 1
7. Empowerment .01 .00 -.19* -.12 -.25** -.07 1
8. Commodity .31** .11 .20* .34** .26** .595** -.20* 1
8. Constructed -.04 .05 -.18* -.012 -.27** -.11 .595** -.30** 1
Notes: ** Correlation is significant at the 0.001 level (1-tailed);
* Correlation is significant at the 0.05 level (1-tailed); Listwise N=142
Table 4
Correlations between the Beliefs about School Learning Questionnaire
and the motivation and cognitive engagement variables
1 2 3 4 5 6 7 8 9 10
1 Subjugation 1
2 Empowerment -.05 1
3. Commodity .59** -.20* 1
4. Constructed -.11 .60** -.31** 1
5. Deep -.06 .275** -.06 .35** 1
6. Shallow .30** -.14 .30** -.055 -.30** 1
7. Self-efficacy -.021 .23* -.03 .28** .45** -.22* 1
8. Intrinsic Motivation .05 .36** -.12 .39** .235* -.04 .23* 1
9. Extrinsic Introjected .16 .22* .05 .295** .06 .26** .045 .32** 1
10. Amotivation .335** -.03 .18* .00 -.28** .15* -.18* -.14* -.03 1
Notes: ** Correlation is significant at the 0.001 level (1-tailed);
* Correlation is significant at the 0.05 level (1-tailed); Listwise N=139