Gender differences in high school students’ interests in physics

Results of Research

After confirmatory factor analysis, values regarding standardized model was examined. It is represented in Figure 2.

Figure 2. The result of confirmatory factor analysis for five-factor solution

D1: Dimension 1 (Learning Time), D2: Dimension 2 (Tentativeness), D3: Dimension 3 (Structure of Knowledge), D4: Dimension 4 (Source of Knowledge), D5: Dimension 5 (Learning Ability) P: Physics, B: Biology, C: Chemistry

The fit values regarding the data also showed that fit indexes are in acceptable ranges. The values are represented in Table 2.

Table 2. Fit indexes for confirmatory factor analysis

Fit Indexes

CMIN/DF

GFI

CFI

RMSEA

RMR

1.10

.97

.99

.02

.03

As seen in Table 2, all of the fit indexes for the scores of the participants are in acceptable range. Chi-square ratio (CMIN/DF) index on self-efficacy scores is under 3.00 as a highest cut-off acceptable value (Arbuckle, 1997). The Comparative Fit Index (CFI) and Goodness of Fit Index (GFI) for the scores of the participants respectively are higher than .90 cut-off lower limit for CFI and .85 for GFI (Hoyle, 2000, Marsh, Balla, & McDonald, 1988). As the other indexes considered in this study, RMR is smaller than .10 as an acceptable value (Jaccard & Wan, 1996) and RMSEA is lower than cut-off .08 (Raykov & Marcoulides, 2006).

Table 3. Correlation coefficients for factors

	Learning Time	Tentativeness	Structure of Knowledge	Source of Knowledge	Learning Ability	Total
Learning Time	-
Tentativeness	.02	-
Structure of Knowledge	.07	.01	-
Source of Knowledge	.15*	.24*	.14*	-
Learning Ability	.10	.06	.10	.11*	-
Total	.48*	.42*	.50*	.60*	.56*	-

Note: All of the correlation coefficients indicated are significant at the level of .05.

Table 3 summarized correlations between dimensions of scientific epistemological beliefs of prospective science teachers. As seen in the table, they are partially independent from each other.

Table 4. Factor loadings and Factor Score Weights of the scores on the Items

Item Content

Items

Factors

Learning Time

Tentativeness

Structure of Knowledge

Source of Knowledge

Learning Ability

Factor Loading

FSW

Factor Loading

FSW

Factor Loading

FSW

Factor Loading

FSW

Factor Loading

FSW

Learning Time

.54

.18

.01

.53

.17

.01

.54

.18

.01

Tentativeness

.01

.51

.16

.01

.03

.00

.01

.52

.16

.01

.03

.01

.02

.52

.18

.01

Structure of Knowledge

.01

.43

.11

.01

.00

.01

.54

.16

.02

.01

.02

.01

.50

.14

.02

.00

Source of Knowledge

.01

.02

.01

.65

.08

.00

.02

.04

.02

.58

.14

.00

.02

.05

.03

.65

.19

.01

Source of Knowledge

.01

.00

.65

.23

.02

.01

.02

.00

.74

.28

.01

.00

.69

.20

Cronbach Alpha Values (Total = .59)

.55

.48

.56

.73

Note: Factor loading refers to standardized regression weight and FSW refers to Factor Score Weights, (* ) indicates factor loading below .40. P: Physics, B: Biology, C: Chemistry.

In table 4, factor loadings of each dimension and Cronbach alpha reliabilities regarding to the dimensions are represented. All of the dimensions have a factor loading over .40 that is actable cut-off value for factor loadings (Li, McCoach, Swaminathan & Tang, 2008). However Cronbach alpha values are low but this situation is common (Liang et al., 2008; Koksal & Cakiroglu, 2010; Topcu & Yilmaz-Tuzun, 2008; Schommer-Aikins, Duell & Hutter, 2005). Hatcher and Stepanski (1994) stated that Cronbach alpha value as like 0.55 can be accepted for statistical processes in social studies.

Table 5. Minimum group scores, maximum group scores regarding each factor and mean of the scores for each factor.

Descriptive Statistics Regarding Minimum Group Scores, Maximum Group Score, Mean of Scores
	N	Minimum Group Scores	Maximum Gourp Scores	Mean	Std. Deviation
F1	364	1.00	3.00	1.85	.67
F2	364	.67	3.00	1.48	.59
F3	364	.33	3.00	1.81	.69
F4	364	.33	3.00	1.65	.65
F5	364	1.00	3.00	1.77	.76
Total	364	1.00	2.60	1.71	.35
Valid N (listwise)	364

In Table 5 scores of the participants on the inventory are represented, it can be seen that the participants had unsophisticated beliefs regarding F1 while they had mixed beliefs regarding F2, F3, F4 and F5.