According to the research questions, this chapter will first reveal the finalization and the descriptive data of the questionnaires, including the Scientific Epistemological Beliefs survey (SEB), the Metacognitive Awareness regarding Science learning
Inventory (MASI), and the Conceptions of Learning Science (COLS) questionnaire, conducted in the quantitative part of the study. Then, the interactions between these variables (i.e., SEB, MASI, and COLS) in quantitative part of study will be explored.
The qualitative part of study with interview data, presenting selected 60 students’
interview about their scientific epistemological beliefs and their phenomenographic conceptions of learning science and science assessment, will be presented. Also, the interplay between students’ beliefs and conceptions gained from qualitative part will be investigated. Finally, the nested ecology considering students’ qualitative data and the role of metacognitive awareness on science learning by combining both
quantitative and qualitative results will be presented last.
79
V.1. Students’ Scientific Epistemological Beliefs Derived from Scientific Epistemological Beliefs survey (SEB)
V.1.1. The exploratory factor analysis of SEB survey
To validate the SEB survey, an exploratory factor analysis with a varimax rotation was performed to clarify its structure. As a result, the participants’ responses were grouped into the following four orthogonal factors, which were: Source,
Certainty, Development, and Justification. The eigenvalues of the four factors from the principle component analysis were all larger than one. Items with a factor loading of less than 0.40 and with many cross-loadings were omitted from the survey. A total of 18 items were retained in the final version of the SEB survey (shown in Table 5.1), and the total variance explained is 55.87%.
The reliability (Cronbach’s alpha) coefficients for these factors were 0.76, 0.67, 0.79, 0.74, respectively, and the overall alpha was 0.76, suggesting that these factors had high reliability in assessing the students’ scientific epistemological beliefs.
Table 5.1 also shows the 240 students’ average item scores and the standard deviations of the four factors of the SEB survey. According to table 5.3, students attained high scores on the “development” (an average of 4.39 per item) and
“justification” factor (an average of 4.36 per item). Their scores on the “source” factor, an average of 3.07 per item, were relatively lower when compared to those of other factors.
80
Table 5.1. The exploratory factor analysis, reliability, factor means, and standard deviations of the SEB (n=240)
Item Number Factor 1 Factor 2 Factor 3 Factor 4
Total variance explained is 55.87%, Total alpha is 0.76
V.1.2. The confirmatory factor analysis of SEB survey
The confirmatory factor analysis (CFA) further confirmed the construct validity and the structure of the SEB survey. A CFA analysis by LISREL, the CFA factor loadings, and the t-values of the items for each factor of the SEB survey are presented in Table 5.2. According to Table 5.2, all of the factor loadings and the t-values of the 18 items on the four factors of the SEB survey showed significance at the 0.05 level,
81
specifying the relations of the observed measures (i.e., items) to their posited
underlying constructs (i.e., factors). The fitness of the items for each factor of the SEB survey (Chi-square per degree of freedom = 2.01, RMSEA = 0.065, GFI = 0.89, NFI = 0.87, NNFI = 0.92, CFI = 0.93) supported the conclusion that the confirmatory model provided a reasonable fit for the data.
Table 5.2. The confirmatory factor analysis and reliability of the SEB (n=240) Item Number Factor 1 Factor 2 Factor 3 Factor 4
82
V.1.3. The interrelations between factors of the SEB survey
Table 5.3 revealed the correlation results among factors of the SEB survey. As a result, the significant correlation observed between source and certainty factors (r = 0.38, p < 0.001), and between development and justification factors (r = 0.43, p <
0.001). Hofer and Pintrich (1997) proposed that the structure of personal
epistemology consists of two major aspects: the nature of knowledge (including certainty and development of knowledge) and the nature of knowing (including source and justification of knowledge). Thus, based on their framework, there should be some correlations between the certainty and development factors and between source and justification factor. However, the present results indicated that the hierarchical structure proposed by Hofer and Pintrich (1997) was not shown in this sample.
Similar results were also found by Conley et al. (2004), which revealed the high correlation between source and certainty factors. According to the framework
proposed by Hofer and Pintrich (1997), the source and justification beliefs concern the nature of knowing, while the certainty and development factors deal with the nature of knowledge. The above structure of beliefs about the nature of knowing and nature of knowledge provide a powerful framework for thinking about epistemological beliefs, but that framework does not reveal the pattern of results found both in the present and Conley et al. (2004) study.
Table 5.3. The interrelations between factors of the SEB survey (n=240)
Source Certainty Development Justification
Source 1
Certainty 0.38*** 1
Development 0.11 0.10 1
Justification 0.03 0.00 0.43*** 1
*** p < 0.001.
83
The design of the SEB survey may interpret the correlation results of this study.
First, the design of the reversed item may cause the pattern of correlations. In the SEB survey, the item descriptions of source and certainty factors reveal the unsophisticated beliefs, and those of development and justification factors represent the sophisticated beliefs. Second, the source and certainty factors highlight the thinking about
knowledge claims (i.e., knowledge comes from external authority, and knowledge has right answer). And, the development and justification factors emphasize on the ideas about scientists’ work (i.e., the evolving and justifying scientific ideas). Thus, the different foci of item descriptions may result in students’ responses. Although the framework proposed by Hofer and Pintrich (1997) is embraced in this study, more works including revising the SEB survey and further examination of the framework proposed by Hofer and Pintrich (1997) are necessary.
84
V.2. Students’ Metacognitive Awareness Obtained from Metacognitive Awareness regarding Science learning Inventory (MASI)
V.2.1. The exploratory factor analysis of MASI
To validate the MASI, an exploratory factor analysis with a varimax rotation was performed to clarify its structure. As a result, the participants’ responses were grouped into the following four orthogonal factors, which were: Self-regulation, Critical judgment, Metastrategy, and Reflection. The eigenvalues of the four factors from the principle component analysis were all larger than one. Items with a factor loading of less than 0.40 and with many cross-loadings were omitted from the survey. A total of 15 items were retained in the final version of the MASI (shown in Table 5.4), and the total variance explained is 57.77%. It should be noted that, there were only three items in the metastrategy factor in MASI. Although Bollen (1989) recommended that each factor be assessed with a minimum of three or four items each, further study may need to add more items to this factor.
The reliability (Cronbach’s alpha) coefficients for these factors were 0.76, 0.71, 0.54, 0.76, respectively, and the overall alpha was 0.84, suggesting that these factors had sufficient reliability in assessing the students’ metacognitive awareness regarding science learning. Overall, the reliability coefficients were acceptable in this study.
However, it is necessary to note that the reliability coefficient for the metastrategy factor was somewhat low (0.54), but acceptable for educational studies (e.g.,
Kizilgunes, Tekkaya, & Sunger, 2009). Still, when interpreting the results of current findings, researchers should consider the low reliability.
Table 5.4 also shows the 240 students’ average item scores and the standard deviations of the four factors of the MASI. According to table 5.7, students attained moderately high awareness regarding each factor of the MASI. The MASI was newly
85
developed in this study. This inventory not only provides a tool for detecting students’
metacognitive awareness regarding science learning, but also can be used to investigate with other variables which may influence students’ learning science.
Table 5.4. The exploratory factor analysis, reliability, factor means, and standard deviations of the MASI (n=240)
Item Number Factor 1 Factor 2 Factor 3 Factor 4
Total variance explained is 57.77%
Total alpha is 0.84
V.2.2. The confirmatory factor analysis of MASI
The confirmatory factor analysis (CFA) further confirmed the construct validity and the structure of the MASI. A CFA analysis by LISREL, the CFA factor loadings,
86
and the t-values of the items for each factor of the MASI are presented in Table 5.5.
According to Table 5.5, all of the factor loadings and the t-values of the 15 items on the four factors of the MASI showed significance at the 0.05 level, specifying the relations of the observed measures (i.e., items) to their posited underlying constructs (i.e., factors). The fitness of the items for each factor of the MASI (Chi-square per degree of freedom = 1.68, RMSEA = 0.053, GFI = 0.93, NFI = 0.93, NNFI = 0.96, CFI = 0.97) indicated a sufficient fit and also confirmed the questionnaire’s structure.
Table 5.5. The confirmatory factor analysis and reliability of the MASI (n=240) Item Number Factor 1 Factor 2 Factor 3 Factor 4 Factor 1: Self-regulation
1 0.72 (11.50*)a
87
V.2.3. The interrelations between factors of the MASI
Table 5.6 revealed the correlation results among factors of the MASI survey. As a result, the significant interrelations were observed among self-regulation, critical judgment, metastrategy, and reflection factors (r = 0.33 to 0.47, p < 0.001). The coefficients ranged between 0.33 and 0.47, showing somewhat overlapping with other factors.
Table 5.6. The interrelations between factors of the MASI (n=240)
Self-regulation Critical judgment Metastrategy Reflection Self-regulation 1
Critical judgment 0.42*** 1
Metastrategy 0.44*** 0.35*** 1
Reflection 0.33*** 0.47*** 0.35*** 1
*** p < 0.001.
88
V.3. Students’ Conceptions of Learning Science Derived from Conceptions of Learning Science questionnaire (COLS)
V.3.1. The exploratory factor analysis of COLS
Similarly, an exploratory factor analysis with a varimax rotation was performed to clarify the structure of COLS. As shown in Table 5.7, the participants’ responses were grouped into the following six orthogonal factors, which were: Memorizing, Testing, Calculate and Practice, Increase of Knowledge, Applying, and
Understanding and Seeing in a new way. The eigenvalues of the six factors from the principle component analysis were all larger than one. Items with a factor loading of less than 0.40 and with many cross-loadings were omitted from the survey. A total of 33 items were retained in the final version of the COLS (shown in Table 5.7), and the total variance explained is 66.35%.
Table 5.7. The exploratory factor analysis, reliability, factor means, and standard deviations of the COLS (n=240)
Item Number Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 1: Memorizing alpha = 0.87, Mean = 2.81, S.D. = 0.88
1 0.75
2 0.79
3 0.78
4 0.79
5 0.68
6 0.67
Factor 2: Testing alpha = 0.88, Mean = 2.95, S.D. = 0.98
9 0.63
12 0.72
13 0.74
14 0.84
15 0.77
89 Total variance explained is 66.34%
Total alpha is 0.86
The original questionnaire items were grouped into six factors through an
exploratory factor analysis. The items of Memorizing, Testing, Calculate and Practice, Increase of Knowledge, Applying factor were respectively loaded on the expected factor. However, the items of Understanding factor and those of the Seeing in a new
90
way factor were loaded on a single factor. After omitting the items with the factor loading less than 0.40 and those with many cross-loadings, six factors were retained in the survey (shown in Table 5.7). The retained items of the sixth factor consisted of original items from the Understanding factor and the Seeing in a new way factor.
Accordingly, we named the sixth factor Understanding and Seeing in a new way factor. This result is paralleled to the findings of Lee, Johanson, and Tsai (2008) which also found six factors of the COLS.
As shown in Table 5.7, the reliability (Cronbach’s alpha) coefficients for these factors were 0.87, 0.88, 0.86, 0.87, 0.85, 0.90, respectively, and the overall alpha was 0.86, suggesting that these factors had high reliability in assessing the students’
conceptions of learning science.
Table 5.7 also shows the 240 students’ average item scores and the standard deviations of the four factors of the COLS. According to table 5.7, students attained high scores on the “increase of knowledge” factor (an average of 4.04 per item) and
“understanding and seeing in a new way” factor (an average of 4.04 per item).
V.3.2. The confirmatory factor analysis of COLS
The confirmatory factor analysis (CFA) further confirmed the construct validity and the structure of the COLS. A CFA analysis by LISREL, the CFA factor loadings, and the t-values of the items for each factor of the COLS are presented in Table 5.8.
According to Table 5.8, all of the factor loadings and the t-values of the 33 items on the four factors of the COLS showed significance at the 0.05 level, specifying the relations of the observed measures (i.e., items) to their posited underlying constructs (i.e., factors). The fitness of the items for each factor of the COLS (Chi-square per degree of freedom = 2.01, RMSEA = 0.065, NFI = 0.92, NNFI = 0.96, CFI = 0.97) indicated a sufficient fit and also confirmed the questionnaire’s structure.
91
Table 5.8. The confirmatory factor analysis and reliability of the COLS (n=240)
Item Number Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6
92
(Continued) Item Number Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6
42 0.81 (14.87*)
43 0.65 (10.88*)
45 0.76 (13.62*)
a t-value
* Significant t-value, p < 0.05.
V.3.3. The interrelations between factors of the COLS
Table 5.9 revealed the correlation results among factors of the COLS. As a result, tow tendencies of correlation results were revealed. First, the positive interrelations were observed among “memorizing,” “testing,” and “calculate and practice” factors (r
= 0.43 to 0.49, p < 0.001). Second, the positive relationships were also identified among “increase of knowledge,” “applying,” and “understanding and seeing in a new way” factors (r = 0.54 to 0.68, p < 0.001).
Lee, Johanson, and Tsai (2008) revealed that “memorizing,” “testing,” and
“calculate and practice” conceptions could be categorized as reproductive conceptions of learning science, and the conceptions as “increase of knowledge,” “applying,” and
“understanding and seeing in a new way” could be categorized as constructivist conceptions. Accordingly, it seems reasonable to identity these two clusters of correlation results. Especially, as shown in Table 5.9, the “testing” factor was positively related to the “memorizing” and “calculate and practice” factors, but was negatively related to the “increase of knowledge,” “applying,” and “understanding and seeing in a new way” factors. The results implied that the “testing” conception was the most sensitive one which positively related to the reproductive conceptions of learning science and negatively related to the constructivist ones. Lee, Johanson, and Tsai (2008) also indicated that the “testing” conception was the strongest positive
93
predictor to reproductive learning approaches and was negatively related to the constructivist learning approaches to learning science.
Table 5.9. The interrelations between factors of the COLS (n=240)
M T CP IK A US
Memorizing (M) 1
Testing (T) 0.43*** 1
Calculate and Practice (CP) 0.48*** 0.49*** 1
Increase of Knowledge (IK) 0.00 -0.26*** 0.09 1
Applying (A) -0.12 -0.39*** -0.04 0.54*** 1
Understanding and Seeing in a new way (US)
-0.05 -0.34*** -0.01 0.68*** 0.65*** 1
*** p < 0.001.
94
V.4. The Interactions between Students’ Responses toward the SEB, MASI, and COLS
V.4.1. The relations between students’ responses toward SEB and MASI
The Pearson correlation analysis was conducted to examine the relations between students’ responses of SEB and MASI. As shown in Table 5.10, all factors of MASI (i.e., self-regulation, critical judgment, metastrategy, and reflection) were positively related to the “justification” factor of SEB (r = 0.18, 0.31, 0.25, 0.19 respectively).
Such results seemed to indicate that students held more sophisticated beliefs about the role of experiments and scientists’ ideas on scientific knowledge tended to express more metacognitive awareness regarding science learning.
Table 5.10. The correlations between students’ responses toward SEB and MASI (n=240)
SEB
MASI Source Certainty Development Justification
Self-regulation 0.01 -0.04 0.09 0.18**
Critical judgment 0.03 0.00 0.22*** 0.31***
Metastrategy -0.00 0.02 0.10 0.25***
Reflection 0.05 -0.10 0.06 0.19**
** p < 0.01, *** p < 0.001
As Conley et al. (2004) suggested that the justification factor, in the domain of science, is mainly concerned with the role of experiments and the use of evidence to support arguments. Hofer and Pintrich (1997) also argued that the justification for knowing includes how individuals evaluate knowledge claims. Consequently, when students held the sophisticated beliefs about “justification” factor, they may consider about whether there are enough evidence to support the knowledge claims. To this end,
95
students’ metacognitive awareness may be encouraged by such belief. This result, to some extent, parallels to Kitchener’s (1983) suggestion about the hierarchical relation between epistemic cognition and metacognition. Kitchener (1983) suggested that metacognition deals with the knowledge about cognitive tasks, particularly the application of strategies and the monitoring of their usage; Epistemic cognition, operates in conjunction with metacognition, and involves the monitoring of the epistemic nature of problem solving, including an awareness of the limits and certainty of knowing, and the criteria for the process of knowing.
V.4.2. The relations between students’ responses toward SEB and COLS
The Pearson correlation analysis was also conducted to examine the relations between students’ responses of SEB and COLS. As shown in Table 5.11, the “source”
factor of SEB was negatively related to the “memorizing,” “testing,” and “calculate and practice” factors of COLS (r = -0.23, -0.15, -0.16 respectively), and was
positively related to the “understanding and seeing in a new way” factor (r = 0.17).
The “certainty” factor was negatively related to the “memorizing” factor (r = -0.13).
Moreover, the “development” and “justification” factor were positively related to the
“increase of knowledge,” “applying,” and “understanding and seeing in a new way”
factors of COLS. As aforementioned, Lee, Johanson, and Tsai (2008) revealed that
“memorizing,” “testing,” and “calculate and practice” conceptions could be
categorized as reproductive conceptions of learning science, and the conceptions as
“increase of knowledge,” “applying,” and “understanding and seeing in a new way”
could be categorized as constructivist conceptions. Accordingly, the results shown in Table 5.11 seem to indicate two tendencies that, on the one hand, students with more sophisticated beliefs about the source of scientific knowledge tended to hold
constructivist conceptions of learning science (i.e., “understanding and seeing in a
96
new way” conception) and not to hold reproductive conceptions (i.e., “memorizing,”
“testing,” and “calculate and practice” conceptions); on the other hand, students who held mature beliefs about the development and justification of scientific knowledge were oriented to embrace the constructivist conceptions.
Table 5.11. The correlations between students’ responses toward SEB and COLS (n=240)
SEB
COLS Source Certainty Development Justification
Memorizing -0.23*** -0.13* 0.04 -0.03
Testing -0.15* -0.03 -0.00 -0.09
Calculate and Practice -0.16* -0.11 -0.01 0.02
Increase of Knowledge 0.09 -0.11 0.24*** 0.30***
Applying 0.09 -0.03 0.15* 0.31***
Understanding and Seeing in a new way
0.17** -0.01 0.18** 0.37***
* p < 0.05, ** p < 0.01, *** p < 0.001.
To further understand the relations between students’ responses to SEB and COLS, a series of regression analyses with stepwise method were performed. This study explored the predictive value of the four factors of SEB on students’
conceptions of learning science. As shown in Table 5.12, the regression analyses indicate that beliefs about source of knowledge contribute negatively to the conceptions of learning science as “memorizing,” “testing,” and “calculate and practice,” and positively to the “understanding and seeing in a new way” conceptions.
Moreover, the beliefs about justification of knowledge contribute positively to the conceptions of learning science as “increase of knowledge,” “applying,” and
“understanding and seeing in a new way.”
97
Table 5.12. The regression analysis for SEB factors predicting COLS factors (n=240)
COLS SEB B R t F
The results seem to indicate that students endorsing less sophisticated beliefs about the source of knowledge tend to hold the reproductive conceptions of learning science, such as “memorizing,” “testing,” and “calculate and practice” conceptions.
Less sophisticated beliefs about the source of knowledge view knowledge as external to the self, originating and residing in outside authorities (Conley et al., 2004; Hofer
& Pintrich, 1997). As a result, students with this belief may view learning science as learning the facts external to the self, originating and residing in outside authorities. In other words, students with this belief may hold reproductive conceptions of learning science. The belief about the justification of knowledge concerns with the role of experiments and the use of data to support the arguments. Hofer and Pintrich (1997) suggested that students with this belief may move through a continuum of dualistic beliefs to the multiplistic acceptance of opinions to reasoned justification for beliefs
98
(p. 120). Accordingly, it seems reasonable to find that students with more
sophisticated beliefs about justification of knowledge tend to hold more constructivist conceptions of learning science.
Furthermore, the beliefs about source of knowledge and justification of
knowledge could be referred to the beliefs about the nature of knowing (Conley et al., 2004; Hofer & Pintrich, 1997). Thus, the results of this study reveal that beliefs about the nature of knowing contribute significantly to the conceptions of learning science.
Students’ learning, to some extent, involves the process of knowing and knowledge acquisition. Perry (1970) suggested that students’ epistemology develop through their educational experiences (c.f., Hofer & Pintrich, 1997). And, students’ conceptions of
Students’ learning, to some extent, involves the process of knowing and knowledge acquisition. Perry (1970) suggested that students’ epistemology develop through their educational experiences (c.f., Hofer & Pintrich, 1997). And, students’ conceptions of