Verification and analysis of factors influencing mobile learning behavior



On the basis of constructing the model of factors influencing the mobile learning of liberal arts students in colleges and universities, the samples are measured, the reliability and validity are analyzed, and the constructed mobile learning model is verified.

Key words:TAM3; Mobile learning; Reliability and Validity Analysis

Introduction

To ensure the validity of the questionnaire, we conducted a pre-survey before the questionnaire was issued. Initially issued seven questionnaires, the questionnaire recovery and statistics, found the problem described in the questionnaire. After revising the questions of the questionnaire, finalize the questionnaire and sample plan, and issue and recall the questionnaire formally.

1.Sample description

We take the second grade undergraduates majoring in Chinese Language and Literature in the College of Liberal Arts as the research object, and issue and recycle paper questionnaires. A total of 80 questionnaires were distributed and 73 questionnaires were returned, of which 68 were valid, and the effective rate was 85 %. The data collected by a single sample t test, the statistics are Table 1, we can see that the variable confidence interval up and down the smaller floating, the sample has further description of the feasibility and reliability of the analysis.

Table 1

Independent samples t-test

Independent samples t-test
	T	df	Significance(two-tailed)	The average difference	95 % difference between the number of confidence intervals
					Lower limit	Upper limit
ID1	19.632	67	0.000	2.10294	1.8891	2.3168
ID2	7.436	67	0.000	0.82353	0.6025	1.0446
ID3	10.114	67	0.000	1.35294	1.0859	1.6200
ID4	14.360	67	0.000	1.79412	1.5447	2.0435
ID5	12.906	67	0.000	1.60294	1.3550	1.8508
MDF1	12.883	67	0.000	1.51471	1.2800	1.7494
MDF2	14.866	67	0.000	1.95588	1.6933	2.2185
MDF3	12.310	67	0.000	1.51471	1.2691	1.7603
EE1	13.047	67	0.000	1.32353	1.1211	1.5260
EE2	19.674	67	0.000	2.19118	1.9689	2.4135
EE3	7.548	67	0.000	0.92647	0.6815	1.1715
EE4	14.557	67	0.000	2.00000	1.7258	2.2742
EE5	13.615	67	0.000	1.79412	1.5311	2.0571
AF1	13.999	67	0.000	1.13235	0.9709	1.2938
AF2	13.460	67	0.000	1.29412	1.1022	1.4860
AF3	11.284	67	0.000	1.26471	1.0410	1.4884
AF4	11.515	67	0.000	1.60294	1.3251	1.8808
AF5	11.090	67	0.000	1.51471	1.2421	1.7873

2. Reliability Analysis

The gauge portion of this paper consists of four subscales. In order to effectively ensure the reliability of the scale, we also test the reliability of the four subscales in the analysis of the reliability of the total scale, as shown in Table 2 below.

Table 2

Total scale reliability analysis

	Average of scales	Scale variance	Relevant nature of the total number of items corrected	Cronbach's Alpha
AF1	43.5735	129.144	0.677	0.921
AF2	43.4118	126.664	0.704	0.920
AF3	43.4412	124.489	0.703	0.919
AF4	43.1029	120.303	0.724	0.918
AF5	43.1912	119.470	0.776	0.917
ID1	42.6029	130.959	0.401	0.926
ID2	43.8824	128.643	0.501	0.924
ID3	43.3529	134.023	0.181	0.932
ID4	42.9118	125.455	0.578	0.922
ID5	43.1029	119.974	0.839	0.916
MDF1	43.1912	123.709	0.705	0.919
MDF2	42.7500	122.847	0.658	0.920
MDF3	43.1912	125.261	0.597	0.922
EE1	43.3824	128.031	0.588	0.922
EE2	42.5147	126.761	0.592	0.922
EE3	43.7794	125.936	0.568	0.922
EE4	42.7059	122.002	0.662	0.920
EE5	42.9118	121.604	0.712	0.919

As can be seen from the above table, the correlation coefficient between ID3(I was hesitant to use the APP. I am afraid that once the mistake will have an impact on normal learning.) and the remaining items was 0.181, the correlation coefficient between ID1(Among my peers around me, I am usually the first to try new information technologies.) and the remaining items was 0.401, The correlation coefficient is very low. After the item is removed from the scale of internal consistency Alpha coefficient value changes in terms of point of view, After ID3 is deleted, the Alpha coefficient of the mobile learning scale changed from 0.925 to 0.932. After the deletion of ID1, the Alpha coefficient of the mobile learning scale changed from 0.925 to 0.926. The remaining 17 items, the items deleted after the scale coefficients are smaller than 0.925. As can be seen from the table to correct problems related items and the total score, ID3, the homogeneity of entry ID1 and the rest of the title is not high, consider deleting it.

Alpha coefficient of 0.6 on behalf of the reliability of the scale is acceptable, 0.7 reliability rating scale above is better, More than 0.8 reliability scale on behalf of very good. Alpha coefficients of the four subscales in the reliability analysis table are 0.691, 0.796, 0.797 and 0.874, respectively. In addition, the alpha coefficient of the total scale is 0.925, indicating that the reliability of the influencing factors is good, as shown in Table 3.

Table 3

Reliability analysis

Variable	Item	Cronbach's Alpha	Cronbach's alpha in the subscale
Individual Differences（ID）	ID2	0.704	0.691
	ID4	0.486
	ID5	0.574
Mobile Device Features（MDF）	MDF1	0.662	0.796
	MDF2	0.706
	MDF3	0.792
External Environment（EE）	EE1	0.757	0.779
	EE2	0.742
	EE3	0.781
	EE4	0.707
	EE5	0.688
Application Function（AF）	AF1	0.865	0.874
	AF2	0.855
	AF3	0.845
	AF4	0.842
	AF5	0.825

3.Validity analysis

To test the construct validity of the scale, we conducted a factor analysis as shown in Table 4.

Table 4

KMO and Bartlett test

Project	Value
Kaiser-Meyer-Olkin measures sampling suitability	0.877
Bartlett's ball check	698.040
df	120
Significance	0.000

KMO is the Kaiser-Meyer-Olkin sampling suitability measure (values range from 0 to 1), when the KMO value is bigger (closer to 1), it means that there are more common factors among the variables, and the lower the net correlation coefficient between the variables, the more suitable for the factor analysis. The KMO value here is 0.877, the indicator statistic is greater than 0.80, and the property presented is «good», indicating that there are common factors between the variables and the variables are suitable for the factor analysis. As shown in Table 5.

Table 5

Rotation component matrix

Items	Ingredient
	1	2	3	4
ID2	0.704
ID4	0.845
ID5	0.803
MDF1		0.879
MDF2		0.856
MDF3		0.795
EE1			0.668
EE2			0.717
EE3			0.596
EE4			0.806
EE5			0.843
AF1				0.780
AF2				0.808
AF3				0.831
AF4				0.833
AF5				0.871

The loading of component 1 on ID2, ID4, ID5 is greater than 0.7 and the first factor can be determined by ID2, ID4, ID5. Similarly, the second factor is determined by MDF1, MDF2 and MDF3. The third factor is determined by EE1, EE2, EE3, EE4 and EE5. The fourth factor is determined by AF1, AF2, AF3, AF4 and AF5, Table dimensions are basically the same, indicating that the scale has good structural validity.

4.Model verification

（1）Personal factor analysis

Through the path analysis of the gender of the individual factors, we found that the significance of the individual factors was 0.000, located in the range of significant differences, as shown in Table 6 below. From this we can judge that the personal factors of college liberal arts students have an impact on their mobile learning behavior. Gender coefficient value is positive, we can see that gender has a significant positive impact on mobile learning in liberal arts students in colleges and universities, and then validates the hypothesis of the influence of personal factors on mobile learning behavior.

Table 6

Regression Analysis of Personal Factors and Mobile Learning Behavior

Personal variable	Non-standardized coefficients		Standardized coefficients	T	Significance
	B	Standard error	Beta
Equation (constant)	34.319	6.923	-	4.957	0.000
Gender	3.226	3.693	0.107	0.874	0.385

（2）Perception factor analysis

By studying the mobile learning behavior of college liberal arts students, it is found that most students think that blue ink cloud class APP is helpful for the course of «Ancient Chinese» and «Office Automation Practice». At present, this mode is also applied in teaching practice. Through the reliability analysis and validity analysis, it is proved that the research model of mobile learning behavior of college liberal arts students has strong fittingness.

Regression analysis of the observed factors, such as Table 7. Based on the model of mobile learning behavior of liberal arts students in colleges and universities, this paper analyzes the individual differences of mobile learning behavior, the characteristics of mobile devices, external environment factors, application functions, personal factors and mobile learning behaviors from the perspective of perceived ease of use and perceived usefulness Relationship.

Table 7

Regression coefficient analysis of influential factors

	Regression coefficient
Individual Differences >>> Learning intention	0.873
Mobile device features>>>Learning intention	0.845
External environment >>> Learning intention	0.929
Application function >>> Learning intention	0.921
Learning intention >>> Learning behavior	0.986

Through the above measurements, we can draw the following conclusions:

①Individual differences positively influence learning intent

②Mobile device features positively influence learning intent

③The external environment positively affects the learning intent

④Application function positively affects learning intent

Appendix：Observation variable dimension measurement

Variable	Serial number	Problem description
Individual Differences (ID)	ID1	Among my peers around me, I am usually the first to try new information technologies. (deleted)
	ID2	If others use mobile phones and other mobile terminals to demonstrate how to use the APP, I can accomplish the same operation.
	ID3	I was hesitant to use the APP. I am afraid that once the mistake will have an impact on normal learning. (deleted)
	ID4	I think the APP help reduce plagiarism.
	ID5	I think the learning effect of using the APP is trustworthy.
Mobile Device Features (MDF)	MDF1	The navigation design of the APP is conducive to more convenient reading.
	MDF2	The APP makes my study more proactive.
	MDF3	The APP is easy to use, you can not be limited by time and place restrictions.
External Environment (EE)	EE1	If my classmates, friends or teachers recommend me to use the APP for secondary study, I will try it..
	EE2	Using the APP can improve my class performance.
	EE3	I am able to use the APP on a variety of end devices.
	EE4	Under the open network and plenty of internet traffic, I will do my own self-study using the APP.
	EE5	When I encounter problems with the APP, I can get help from service providers or others.
Application Function (AF)	AF1	The issue of the APP is clear and understandable.
	AF2	The issue of the APP is related to the syllabus.
	AF3	A lot of information can be uploaded to the APP, which helps me to understand and study the course..
	AF4	The APP's sign-in function has enhanced my learning autonomy.
	AF5	The APP's function of homework assignments can enhance my learning motivation, improve learning efficiency, and be quick and easy.

References：

1. Park, S. Y., Nam, M. W., & Cha, S.B. (2012). University Students' Behavioral Intention to Use Mobile Learning: Evaluating the Technology Acceptance Model [J]. British Journal of Educational Technology, 43(4), 592–605.
2. Chang, S. C., & Tung, F. C. (2007). An Empirical Investigation of Students’ Behavioral Intentions to Use the Online Learning Course Websites [J]. British Journal of Educational Technology, 39(1):71–83.
3. Briz-Ponce, L., Pereira, A., Carvalho, L., Juanes-Méndez, J. A., &García-Peñalvo, F. J. (2017). Learning with Mobile Technologies-Students’ Behavior [J]. Computers in Human Behavior, 72:612–620.
4. Wang, Y. S., Wu, M. C., & Wang, H. Y. (2009). Investigating the Determinants and Age and Gender Differences in the Acceptance of Mobile Learning [J]. British Journal of Educational Technology, 40(1):92–118.
5. Nikou, S. A., & Economides, A. A. (2017). Mobile-Based Assessment: Investigating the Factors that Influence Behavioral Intention to Use [J]. Computers & Education, 109:56–73.