Factors influencing interest in STEM careers: An exploratory factor analysis

Exploratory Factor Analysis for Environmental Factors

For environmental factors, a total of 34 items were identified. These 34 items were subjected to principal component analysis (PCA) using SPSS version 23. Prior to performing PCA, the suitability of data for factor analysis was assessed. Inspection of the correlation matrix revealed the presence of many coefficients of .3 and above. The Kaiser-Meyer-Olkin value was .915, exceeding the recommended value of .6 (Pallant, 2011) and the Bartlett's test of sphericity reached a statistical significance, supporting the factorability of the correlation matrix as shown in Table 3.

Table 3. The Findings from Kaiser-Meyer-Olkin and Bartlett's Test for Environmental Factors

Kaiser-Meyer-Olkin's Measure of Sampling Adequacy		.915
Bartlett's Test of Sphericity	Approx. Chi-Square	5729.671
	Df	561
	Sig.	.000

Principal component analysis revealed the presence of seven components with eigenvalues exceeding 1. An inspection of the scree plot revealed a clear break after the fourth component as shown in Figure 1. Thus, the decision was made to retain these four components for further investigation.

Figure 1. Scree Plot for Environmental Factors

To aid the interpretation of the four components, varimax rotation was used to generate orthogonal factors. The four components' solution explained a total of 49.70% of the variance, with component 1 contributing 19.57%, component 2 contributing 12.08%, component 3 contributing 10.31 %, and component 4 contributing 7.72%, as shown in Table 4.

Table 4. Total Variance Explained for Environmental Factors

	Initial Eigenvalues			Extraction Sums of Squared Loadings			Rotation Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	10.604	31.188	31.188	10.604	31.188	31.188	6.656	19.576	19.576
2	2.819	8.291	39.479	2.819	8.291	39.479	4.109	12.085	31.661
3	1.917	5.639	45.117	1.917	5.639	45.117	3.508	10.318	41.979
4	1.558	4.583	49.700	1.558	4.583	49.700	2.625	7.721	49.700

Table 5. Rotated Component Matrix for Environmental Factors

	Component
	Media influences	Activities outside the classroom	Social influences	Activities in the classroom
d7	.825
d4	.819
d1	.801
d3	.795
d6	.782
d5	.780
d8	.732
d2	.713
d10	.622
d9	.566
b12		.728
b13		.710
b8		.681
b10		.656
b9		.626
b11		.595
b7		.489
b1		.459
c3			.592
c5			.552
c6			.546
c11			.512
c9			.510
c8			.507
c1			.506
c2			.495
c7			.491
c10			.472
c4			.430
b3				.775
b2				.720
b4				.645
b5				.601
b6

Based on factor loading values as shown in Table 5, item b6 was removed because of low factor loading whereas item b1 was categorized into another group. We decided to retain item b1 in the outside of classroom activity component. We believe that students are highly likely to conduct design activities outside the classroom and not in the classroom as shown in earlier studies. Overall, based on exploratory factor analysis, environmental factors consist of four sub factors i.e., activities in the classroom (component 4), activities outside the classroom (component 2), social influences (component 3), and media influences (component 1).

Exploratory Factor Analysis for STEM Self-Efficacy

20 items were developed for self-efficacy. These 20 items were subjected to principal component analysis (PCA) using SPSS version 23. Inspection of the correlation matrix revealed the presence of many coefficients of .3 and above and the Kaiser-Meyer-Olkin value was .881, exceeding the recommended value of .6 (Pallant, 2011) and the Bartlett's test of sphericity reached a statistical significance, supporting the factorability of the correlation matrix as shown in Table 6.

Table 6. Findings from Kaiser-Meyer-Olkin and Bartlett's Test for Self-Efficacy

Kaiser-Meyer-Olkin Measure of Sampling Adequacy		.881
Bartlett's Test of Sphericity	Approx. Chi-Square	4192.599
	Df	190
	Sig.	.000

Principal component analysis revealed the presence of four components with eigenvalues exceeding 1. An inspection of the scree plot revealed a clear break after the fourth component as shown in Figure 2.

Figure 2. Scree Plot for Self-Efficacy

Varimax rotation was selected because the factors in self-efficacy were determined according to STEM fields. The four components' solution explained a total of 67.86% of the variance, with component 1 contributing 18.05%, component 2 contributing 17.14%, component 3 contributing 16.64%, and component 4 contributing 16.04 %, as shown in Table 7.

Table 7. Total Variance Explained for Self-Efficacy

	Initial Eigenvalues			Extraction Sums of Squared Loadings			Rotation Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	7.390	36.952	36.952	7.390	36.952	36.952	3.609	18.046	18.046
2	2.671	13.353	50.305	2.671	13.353	50.305	3.429	17.144	35.190
3	1.910	9.552	59.858	1.910	9.552	59.858	3.329	16.643	51.833
4	1.602	8.010	67.868	1.602	8.010	67.868	3.207	16.035	67.868

Based on the factor loading values in the rotated component matrix as shown in Table 8, all the 20 items belonged to the four elements, namely i) Science (e1 to e5), ii) Technology (e6 to e10), iii) Engineering (e11 to e15), and iv) Mathematics (e16 to e20).

Table 8. Rotated Component Matrix for Self-Efficacy

	Component
	Science	Mathematics	Engineering	Technology
e4	.851
e2	.811
e3	.807
e1	.721
e5	.693
e20		.869
e16		.842
e17		.766
e19		.743
e18		.646
e13			.816
e14			.805
e15			.750
e12			.746
e11			.668
e8				.818
e9				.812
e6				.792
e10				.748
e7				.722

Exploratory Factor Analysis for Perception of STEM Careers

14 items were developed for perception of STEM careers. These 14 items were subjected to principal component analysis (PCA). Prior to performing PCA, the suitability of data for factor analysis was assessed. Inspection of the correlation matrix revealed the presence of many coefficients of .3 and above. The Kaiser-Meyer-Olkin value was .880, exceeding the recommended value of .6 (Pallant, 2011) and the Bartlett's test of sphericity reached a statistical significance, supporting the factorability of the correlation matrix as shown in Table 9.

Table 9. Findings from Kaiser-Meyer-Olkin and Bartlett's Test for Perception of STEM Careers

Kaiser-Meyer-Olkin Measure of Sampling Adequacy		.880
Bartlett's Test of Sphericity	Approx. Chi-Square	2533.864
	Df	91
	Sig.	.000

Initially, principal component analysis revealed the presence of three components with eigenvalues exceeding 1. An inspection of the scree plot revealed a clear break after the thirdcomponent as shown in Figure 2.

Figure 3. Scree Plot for Perception of STEM Careers

However, we decided to keep to two components as determined earlier using varimax rotation. The two solutions explained a total of 56.86% of the variance, with component 1 contributing 30.03% and component 2 contributing 26.83 %, as shown in Table 10.

Table 10. Total Variance Explained for Perception of STEM Careers

	Initial Eigenvalues			Extraction Sums of Squared Loadings			Rotation Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	5.811	41.506	41.506	5.811	41.506	41.506	4.204	30.029	30.029
2	2.149	15.350	56.856	2.149	15.350	56.856	3.756	26.827	56.856

Based on the factor loading values in the rotated component matrix as shown in Table 8, all the 14 items belonged to the two components as was decided earlier in this study. Table 11 shows the rotated component matrix for perception of STEM careers belong into two sub factors i.e., job prospect in STEM careers and skills needed in STEM careers.

Table 11. Rotated Component Matrix for Perception of STEM Careers

	Component
	Job Prospect	Skills Needed
f2	.774
f6	.766
f3	.749
f4	.740
f1	.736
f5	.620
f8	.612
f7	.453
f10		.858
f11		.856
f12		.844
f13		.763
f14		.638
f9		.515

Exploratory Factor Analysis for Interest in STEM Careers

12 items were developed for interest in STEM careers. These 12 items were subjected to principal component analysis (PCA). Prior to performing PCA, the suitability of data for factor analysis was assessed. Inspection of the correlation matrix revealed the presence of many coefficients of .3 and above. The Kaiser-Meyer-Olkin value was .875, exceeding the recommended value of .6 (Pallant, 2011) and the Bartlett's test of sphericity reached a statistical significance, supporting the factorability of the correlation matrix as shown in Table 12.

Table 12. Findings from Kaiser-Meyer-Olkin and Bartlett's Test for Interest in STEM Careers

Kaiser-Meyer-Olkin Measure of Sampling Adequacy		.875
Bartlett's Test of Sphericity	Approx. Chi-Square	1439.306
	Df	66
	Sig.	.000

Initially, principal component analysis revealed the presence of two components with eigenvalues exceeding 1. An inspection of the scree plot revealed a clear break after the two components as shown in Figure 2.

Figure 4. Scree Plot for Interest in STEM Careers

To aid interpretation of the two components, varimax rotation was performed. The two solutions explained a total of 51.450% of the variance, with component 1 contributing 25.749% and component 2 contributing 25.701%, as shown in Table 13.

Table 13. Total Variance Explained for Interest in STEM Careers

	Initial Eigenvalues			Extraction Sums of Squared Loadings			Rotation Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	4.842	40.348	40.348	4.842	40.348	40.348	3.090	25.749	25.749
2	1.332	11.102	51.450	1.332	11.102	51.450	3.084	25.701	51.450

Based on the factor loading values in the rotated component matrix as shown in Table 14, all the 12 items belonged to the two components i.e., physical sciences and life sciences. Physical sciences are items/work related to physics, mathematics, computer sciences, energy, engineering and entrepreneur or business scientist careers. In contrast, careers in life sciences consist of jobs related to environmental works, biology and zoology, earth science, medical science, chemistry and science teachers or educators.

Table 14. Rotated Component Matrix for Interest in STEM Careers

	Component
	Life Sciences	Physical Sciences
g7	.821
g3	.777
g8	.656
g5	.609
g2	.536
g12	.494
g10		.832
g9		.805
g6		.634
g1		.571
g11		.547
g4		.424

Reliability Analysis

Overall, the Cronbach's alpha value for each factor was between .817 and .933. Thus, each value indicated that all items showed high reliability as shown is Table 15.

Table 15. Cronbach's Alpha Value

No.	Construct	Element	Cronbach's Alpha Value
1	Learning experiences	In the classroom	.704
1	Learning experiences	Outside the classroom	.833
2	Social influences	-	.817
3	Media influences	-	.933
4	Self-efficacy	Science	.892
		Technology	.849
		Engineering	.856
		Mathematics	.897
5	Perception of STEM careers	Job prospect	.858
5	Perception of STEM careers	Skills needed in STEM career	.873
6	Interest in STEM careers		.863

Interpretation of Total Score Mean

The level for each factor was interpreted through the total score mean value and categorized as shown in Table 16. The total score mean interpretation was adapted from Nunnally (1997) where scores are indicated as low, medium low, medium high and high based on the total score mean obtained.

Table 16. Mean Value and Interpretation of Total Score Mean

Total Score Mean

Interpretation of Total Score Mean

1.00 - 2.50

2.51 - 5.00

5.01 - 7.50

7.51 - 10.00

Low

Medium Low

Medium High

High

(Adapted from Nunnally, 1997)