An evaluation of English-medium instruction in higher education: influencing factors and effects

Figures & data

Table 1. Latent variables and measurement indicators in the model.

Model types	Factors	Latent variables	Measurement indicators	Notations
Formative models	Personal factors	Prior knowledge	Prior knowledge of English	A1
			Prior knowledge of Subject	A2
		Interest	Interest in learning English	B1
			Interest in learning Subject	B2
		Effort	Effort of learning English	C1
			Effort of learning Subject	C2
	Environmental factors	Teacher	Teachers’ level of English	D1
			Teachers’ level of Subject knowledge	D2
			Teachers’ lecturing skills	D3
		Course	Subject major courses design	I1
			Classroom English use	I2
		Resource	Resources for learning Subject	F1
			Resources for learning English	F2
Reflective models	Student performance	English	English competency	G1
			Confidence about English	G2
			Ability to use English	G3
		Subject	Major courses results	H1
			Academic competitions	H2
			Academic skills	H3
			Application of subject knowledge in practice	H4

Figure 1. The exploratory conceptual model of the EMI.

Figure 2. Questionnaire design and data collection procedure Source: compiled by the author, reference: Bryman (2012).

Table 2. Structure of the survey.

Parts of the survey	Data collected from the survey
Part 0	Informed consent
Part 1	To identify whether the participants are EMI learners
Part 2	Data to measure the indicators for the variables involved in the learning process of EMI programme
Part 3	Background information of participants

Figure 3. The PLS model in the study.

Table 3. Measurement fit for reflective models in this study.

Indices	Assessment Criteria	Results in the study
Indicator reliability	> 0.7	> 0.7
Composite reliability	> = 0.8	>0.8
Cronbach's alpha	> = 0.8	>0.8
Cross-loadings	Lower than outer loadings	Lower than outer loadings
Average variance extracted (AVE)	> 0.5 and the cross-loadings	>0.5 and the cross-loadings
The Fornell–Larcker discriminant validity criterion	The square root of AVE greater than correlations with other factors of the latent variable	The square root of AVE greater than correlations with other factors of the latent variable
Heterotrait-Monotrait Ratio (HTMT)	<0.9	<0.9

Table 4. Measurement fit for formative models in this study.

Indices	Assessment Criteria	Results in the study
Face validity	Indicators formatively present the overall meaning of the construct of the latent variable	Indicators formatively present the overall meaning of the construct of the latent variable
Path loading significance	> 0.7	> 0.7
Measurement weights	> 0.2	> 0.2
Cross-loadings	Lower than outer loadings	Lower than outer loadings
Multicollinearity of the indicators in formative models	Variance inflation factor (VIF) < 4.0	Variance inflation factor (VIF) < 4.0

Table 5. Goodness of fit for the inner structural model in this study.

Indices	Assessment Criteria	Results in the study
Structural path coefficients	Significant	Significant
R-square	0.67, 0.33 and 0.19 to be ‘substantial’, ‘moderate’ and ‘weak’ respectively	Moderate strength of the effect
Multicollinearity	Variance inflation factor (VIF) < 4.0	Variance inflation factor (VIF) < 4.0
Indirect path coefficients	Significant	Significant

Figure 4. The estimated structural model of the diverse effects.

Table 6. Direct and indirect effects.

Hypotheses	Paths	Effects	Coefficients	P Values
H1a	Prior knowledge -> English	Direct effect	0.299	0.000
H1b	Prior knowledge -> Subject	Direct effect	0.254	0.000
H2a	Effort -> English	Direct effect	0.351	0.000
H2b	Effort -> Subject	Direct effect	0.436	0.000
H3	Interest -> Effort	Direct effect	0.544	0.000
	Interest -> English	Indirect effect	0.191	0.000
	Interest -> Subject	Indirect effect	0.238	0.000
H4a	Course -> English	Direct effect	0.156	0.018
H4b	Course -> Subject	Direct effect	0.174	0.020
H5	Teacher -> Course	Direct effect	0.409	0.000
	Teacher -> English	Indirect effect	0.064	0.017
	Teacher -> Subject	Indirect effect	0.071	0.024
H6	Resource -> Course	Direct effect	0.393	0.000
	Resource -> English	Indirect effect	0.061	0.043
	Resource -> Subject	Indirect effect	0.068	0.041

Table 7. Estimation results of the hypotheses in the study.

Hypothesis	Content	Evaluation result
H1a	A student's prior knowledge would directly predict his or her performance in English.	Established/Confirmed
H1b	A student's prior knowledge would directly predict his or her performance in subject.	Established/Confirmed
H2a	A student's effort would directly predict his or her performance in English.	Established/Confirmed
H2b	A student's effort would directly predict his or her performance in subject.	Established/Confirmed
H3	A student's interest would directly predict his or her effort, and thus indirectly predict his or her performance in English and subject.	Established/Confirmed
H4a	Course quality would directly predict student performance in English.	Established/Confirmed
H4b	Course quality would directly predict student performance in subject.	Established/Confirmed
H5	Teacher quality would directly predict course quality, and thus indirectly predict student performance in English and subject.	Established/Confirmed
H6	Resource would directly predict course quality, and thus indirectly predict student performance in English and subject.	Established/Confirmed

Table 8. The strength of the effects of diverse factors on student performance.

Factors		English performance	Subject performance	Effect
Personal factors	Effort	0.351	0.436	Direct
	Prior knowledge	0.299	0.254	Direct
	Interest	0.191	0.238	Indirect (through Effort)
Environmental factors	Course	0.156	0.174	Direct
	Teacher	0.064	0.071	Indirect (through Course)
	Resource	0.061	0.068	Indirect (through Course)

Bryman, A. 2012. Social Research Methods. Oxford: Oxford University Press.

 Google Scholar