CHAPTER 3 RESEARCH METHOD
3.3 Q UESTIONNAIRE DESIGN AND DATA COLLECTION
3.3.1 Questionnaire design
The first draft of this research questionnaire is made according to the information obtained from the above-mentioned literature including “ information literacy”, “Personality
Formulate a research project
Examine related literature
Consult the views of scholars and experts
Develop the questionnaire and conduct an early test of the survey
Conduct the survey
Collect data, collate, and analyze
Write research papers
trait anxiety”, “Mobile computer anxiety”, “Use intention of mobile computer in teaching”,
“Mobile computer self-efficacy”. (see Table3-1)
3.3.1A Information literacy scale
The part “Information Literacy” is made based on the teacher basic information literacy indicator for junior high school and elementary school teachers announced by the Ministry of Education as well as on the teacher information literacy scale made by Pai (2004) and Lin (2007), in addition to the views from scholars both at home and abroad. This scale consists of five aspects: information awareness, ability to operate hardware and software, ability to make use of the Internet network, the combination of information technology and teaching, and information communication. The items are based on Likert seven-point scale. The format of the response consists of seven levels, ranging from “strongly disagree” to “strongly agree”. As each item on the response scale is circled, the score will be calculated from1-7. The higher the score is, the higher a teacher’s information literacy is, and vice versa. There are a total of eight items.
3.3.1B Personality trait anxiety inventory
This study is complied by referring to Chung and Long’s(1984) State-Trait Anxiety Inventory, a revised form of Spielberger’s (1983).The items are based on Likert seven-point scale. The format of the response consists of seven levels, ranging from “strongly disagree” to
“strongly agree”. As each item on the response scale is circled, the score will be calculated
from1-7. The higher the score is, the higher a teacher’s trait anxiety is, and vice versa. There are a total of six items.
3.3.1C Mobile computer anxiety scale
This scale consists of three aspects: computer learning anxiety, anxiety about facing the Information Age, and anxiety about applying computers to instruction. It is made according to the Computer Anxiety Sub-scale of Computer Attitude Scale made by Loyd and Gressard (1984) and to Shen’s (2002) the Computer Anxiety Scale for Teachers. The items are based on
Likert seven-point scale. The format of the response consists of seven levels, ranging from
“strongly disagree” to “strongly agree”. As each item on the response scale is circled, the
score will be calculated from1-7. The higher the score is, the higher a teacher’s computer anxiety is, and vice versa. There are a total of six items.
3.3.1D Use intension of mobile computer in teaching scale
This scale is made based on the perspectives of Ajzen (1991), Venkatesh and Davis (1996), scholars of theory of rational behavior. The items are based on Likert seven-point
scale. The format of the response consists of seven levels, ranging from “strongly disagree” to
“strongly agree”. As each item on the response scale is circled, the score will be calculated
from1-7. The higher the score is, the higher a teacher’s intention of using mobile computer in
teaching is, and vice versa. There are a total of five items.
3.3.1E Mobile computer self-efficacy scale
This scale is compiled by referring to Computer Self-efficacy Scale Wong’s (2000), which is made according to the viewpoints of Compeau and Higgins (1995), as well as to Hsieh’s ( 2001) Computer Self-efficacy Scale of Teachers. The items are based on Likert seven-point scale. The format of the response consists of seven levels, ranging from “very little confident” to “ quite a lot of confident”. The higher the score is, the higher a teacher’s mobile computer self-efficacy is, and vice versa. There are a total of six items.
3.3.1F Basic personal data
(1)Sexuality: Male and Female
(2)Age: Under 30 years, 31-40 years, 41-50 years, over 51 years
(3)Educational background: Normal university, Ordinary university Graduate school, others
(4)Teaching experience: Less than 5 years, 6–10 years,11–15 years, 16–20 years, more than 21 years
(5)Working position: Teacher with concurrent administrative duty, class teacher, subject teacher
(6)Experience of using mobile computer: Yes or No (see appendix)
Table 3-1 Questionnaire design
3.3.2 Sampling methods and data collection
This study aims to understand the intention of elementary school teachers to use mobile computers for teaching, and the target population is the public elementary school teachers in Taichung. Stratified random sampling is used to select sample. The sample survey is conducted, with a sample of 114 people being taken as a pretest sample, and about 392 people as a formal sample.
Table 3-2 Sampling methods
SPSS 12 software is used as statistical method to perform descriptive statistical analysis, item analysis, reliability and validity analysis, factor analysis. Stepwise regression analysis is also performed.
3.4.1 The descriptive statistical analysis
The descriptive statistical analysis is getting initial understanding of research object's basic data. The Likert seven-point scale is used for agreement level and for description of the mean value, frequency distribution, standard deviation, and skewness of different variables.
This leads to an initial understanding of sample collection.
3.4.2 Factor Analysis
The main purpose of factor analysis is data simplification. Many original variables are simplified to few factors, yet most information of the original data is still kept. The Bartlett's Test of Sphericity and KMO sample adequacy measurement are used in this research for testing. When the KMO measurement is higher, the outcome of factor analysis is better and there are more common factors between variables. If the chi-square value of the Bartlett's Test of Sphericity is significant, there are common factors between the relative matrices of the parent group and is suitable for factor analysis. For the factor extraction, only the factors with eigenvalue greater than 1 are selected. Furthermore, the factor loading of items of the selected factors are checked if it is higher than 0.5 according to the relative facet of the factor.
3.4.2 Reliability
The Cronbach's α coefficient, the most used reliability coefficient in social science, is used to estimate internal consistency. The consistency is good and the scale is stable when the reliability coefficient is higher than 0.7. If it gets lower than 0.35, the outcome is poor (Yang, 2009).
Chapter 4 Data analysis and results
In this chapter we discuss the questionnaire data and the result of statistical analysis.
Section 4-1 presents the descriptive statistical analysis. Section 4-2 focuses on item analysis, reliability and validity analysis and factor analysis of questionnaire data, and Section 4-3 discusses the statistical analysis result and regression analysis.
4.1 Descriptive statistical analysis
The subjects of this study are elementary school teachers in Taichung City. A total of 430 questionnaires are distributed, and 390 are returned. After eliminating 4 invalid samples, there are 386 effective samples and the valid return rate is 99%. The data are analyzed by descriptive statistics, and obtain the sample number and percentage of demographic variables of the elementary school teachers, so as to find out the proportion of sample data to each variable. The characteristics of sample data are also discussed.
The demographic variables of elementary school teachers in the questionnaire data are shown in Table 4-1. In terms of gender, male teachers account for 32.64%, and female teachers account for 36.01%. As for age, those under 30 years old account for 16.58%, those aged 31-40 account for 36.01%, those aged 41-50 account for 44.04%, and those over 51
years old account for 3.37%. Regarding the educational background, 162 teachers are from normal universities (including teachers colleges) accounting for 41.96%, 112 are from general universities (including teacher training classes, educational programs) accounting for 29.02%, and 112 are from graduate institutes (including 40-credit classes) accounting for 29.02%. In terms of seniority, 12.95% have taught for less than 5 years, 28.50% have taught for 6-10 years, 14.51% have taught for 11-15 years, and 44.04% have taught for more than 15 years.
As for the administrative duties, class teachers account for 60.10%, teachers with concurrent administrative duties account for 28.76, and subject teachers account for 11.14%. Regarding the time of using mobile computer, 10.36% have used it for less than 1 year, 21.24% have used it for 1-2 years, 15.03% have used it for 2-3 years, 51.81% have used it for 3-4 years, and 11.92% have used it for more than 4 years. As seen, the demographic variables of teacher samples are consistent with the norms of elementary school teachers in Taichung City, where the teachers are mostly 31-50 years, the educational background and teaching seniority of teachers are evenly distributed, most are class teachers, and have used mobile computers for more than 3 years. It indicates that the sampling teachers of this study are the main active teacher groups in elementary schools.
Table 4-1 Statistical description of elementary school teachers
4.2 Item, reliability, and factor analysis
4.2.1 Item analysis
The results of item analysis by SPSS are shown in Table 4-2. The average mean is 4.09~6.22,and the standard deviation is 0.88~1.60. The skewness and kurtosis coefficients are considered to observe the distribution characteristic of data in questionnaire variables. The skewness coefficient is -1.89~0.06, conforming to the skewness extreme outliers test, and there is no absolute value greater than 3. The kurtosis coefficient is -0.65~6.73, conforming to the kurtosis extreme outliers test, and there is no absolute value greater than 10. This study calculated the relevancy between the questionnaire items and the total scale to inspect the isomorphism type of specific items. The items with correlation coefficient under .5 were deleted, and the correlation coefficient above .5 was set as the screening criteria. It indicates the items have significant discrimination.
Table 4-2 Summary of item analysis
Item No Mean S.D. skewness Kurtosis Item-Total
Correlation decision
1 5.62 1.26 -0.91 0.52 0.61
2 4.99 1.46 -0.58 -0.24 0.62
3 6.02 1.08 -1.51 2.87 0.63
4 4.72 1.60 -0.32 -0.74 0.65
5 5.77 1.23 -1.27 1.73 0.53
Table 4-2 Summary of item analysis (cont.)
Three scholars are invited to review the content validity of the questionnaire, and assess the relevancy between the research purpose and items, as well as the wording properness.
Some items with poor discrimination are deleted, and ambiguous expressions are revised. The questionnaire contain 31 items, and is measured based on a Likert seven-point scale, where 1 is “Strongly disagree” and 7 is “Strongly agree”. A total of 114 questionnaires are distributed for pre-test, and the Cronbach's α coefficient was 0.928. The reliability refers to the consistency or stability of the score of the questionnaire. If a questionnaire is answered by different respondents at different times or similar items are used, the scores obtained should be the same or similar, in order to be regarded as having good reliability. As suggested by Nunnally (1978) that Cronbach's α coefficient should be greater than 0.7, this study distributed 390 questionnaires in the formal survey, and the Cronbach's α coefficient obtained is 0.947, indicating a high internal consistency and good stability.
4.2.3 Factor analysis
Prior to the factor analysis, KMO (Kaiser-Meyer-Olkin) goodness of fit and Bartlett’s Test of Sphericity are conducted for verifying the correlation among variables, in order to determine whether the variables are applicable to factor analysis. The results reveal that, KMO test value is 0.939, indicating a good relevancy among the items. In addition, Bartlett's chi-square value is 10205.654, which is significant. It indicates that the survey data of this study have common factor and are suitable for subsequent factor analysis. This study adopts
common interpretation of all variables to screen the factors with eigenvalue above 1.
Table 4-3 Summary of KMO and Barlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy. .939 Bartlett's Test of Sphericity
Approx. Chi-Square
10205.654
df
378
Sig.
.000
Five factors are extracted after the factor analysis. The question content, factor loading, and Cronbach's α coefficient of each factor are shown in Table 4-4. As suggested by Richard, Youjai and Kent (1999), the factor loading should be above 0.5, and the ideal value of factor loading should be 0.5~0.95. Table 4-5 shows the percentage of eigenvalue and variance of the extracted factors. The eigenvalue of the first factor is 5.06, that of the second factor is 4.84, that of the third factor is 4.51, that of the fourth factor is 3.86, and that of the fifth factor is 2.66. The explained variance is 18.06%, 17.28%, 16.10%, 13.80% and 9.48%, respectively, 74.73% of variance are explained. Factor 1 has 6 items, which factor loading is 0.80~0.85;
Factor 2 has 7 items, which factor loading is 0.62~0.78; Factor 3 has 5 items, which factor loading is 0.76~0.84; Factor 4 has 6 items, which factor loading is 0.58~0.76; and Factor 5 has 4 items, which factor loading is 0.50~0.91. According to the characteristics of the items, these five factors are named Mobile computer anxiety (MCA), Information literacy (IL), Use
intention of mobile computing in teaching (UI), Mobile computer self-efficacy (MCE) and
Personality trait anxiety (PA), respectively.
Table 4-4 Reliability and factor analysis
No. Items Factor
loading
Cronbach’s α Factor 1: Mobile computer anxiety (MCA)
MCA1 I am afraid of attending mobile computer study
courses. 0.824
MCA2 It is very difficult for me to learn to use mobile
computers. 0.835
MCA3 I am worried about causing mobile computers down
when using it. 0.793
MCA4 I feel afraid when encountering mobile computer
technology related products. 0.849
MCA5 The mobile computers related teaching methods used
in class make me nervous. 0.831
MCA6 I am afraid of using mobile computers to implement
information technology integration teaching. 0.807 Factor 2: Information literacy (IL)
0.940
IL1 I can describe the names of major computer
peripheral devices. 0.712
IL2 I know how to prevent and handle computer virus
infection and safety protection. 0.779
IL3 I can operate office applications (ex. Excel,
PowerPoint) 0.721
IL4 I can do simple maintenance when computer is down
or malfunctioned. 0.732
0.893
Table 4-4 Reliability and factor analysis (cont.)
No. Items Factor
loading
Cronbach’s α IL5 I can set up and use browser to browse the internet. 0.702
IL6 I can guide students to exhibit works through class
web pages. 0.683
IL7 I know how to use projector to display teaching
materials. 0.623
Factor 3: Use intention of mobile computers in teaching(UI)
UI1 I am willing to use mobile computers to teach. 0.759 UI2 I will like to use mobile computers in the future to
assist teaching. 0.841
UI3 I will try my best to use mobile computers to assist
teaching. 0.835
UI4 I think it is worthwhile to use mobile computers to
upgrade teaching result. 0.828
UI5 I will recommend other associates to use mobile
computers in teaching. 0.791
0.945
Factor 4: Mobile computer self-efficacy (MCE)
MCE1 I can learn how to use mobile computers even
without instructions from others. 0.667
MCE2 As long as there is a manual, I can learn to use a
mobile computer. 0.709
MCE3 As long as there are instructions from others, I can
learn to use a mobile computer. 0.578
MCE4 If there is enough time, I can learn to use a mobile
computer myself. 0.759
0.958
Table 4-4 Reliability and factor analysis (cont.)
No. Items Factor
loading
Cronbach’s α MCE5 I myself can easily apply mobile computers in
teaching 0.698
MCE6 If I have experience of using similar mobile computers devices, I will be able to use mobile computers.
0.719
Factor 5: Personality trait anxiety(PA)
PA1 I am an emotionally stable person. 0.500
PA2 I tend to overestimate things. 0.860
PA3 I tend to worry too much about little things. 0.910 PA4 I tend to worry about bad things in the future. 0.840
0.818
Table 4-5 Total variance explained and percentage of variance
Component Eigenvalue Percentage of variance Cumulative percentage
1 5.06 18.07 18.07
In order to find out the effect of each factor on the use intention of mobile computer in
teaching, this study used Pearson Correlation to analyze the correlativity among the factors, find out the significant intercorrelation among the factors, and evaluate the path prediction of these factors. The analytical results are shown in Table 4-6. Factors PA, IL, MCE, MCA and UI all indicate significant positive correlations; with p values lower than 0.01. The results of
research model are also shown in Figure 4-1.
Table 4-6 Summary of Pearson correlation anlaysis
Factor PA IL MCE MCA UI
PA 1.00 0.136∗∗ 0.202∗∗ 0.353∗∗ 0.146∗∗
IL 1.00 0.711∗∗ 0.535∗∗ 0.553∗∗
MCE 1.00 0.634∗∗ 0.705∗∗
MCA 1.00 0.526∗∗
UI 1.00
∗ p < 0.05,∗∗p < 0.01,∗∗∗p < 0.001
Since there are more than one type of predictive factor, in order to discuss the explanatory ability and prediction relations among the factors, this study used stepwise regression model for multiple regression statistical analysis, and observed each path coefficient, so as to find out the strength of impact of the factors. In the first regression analysis, MCE factor is set as the dependent variable, PA and IL as independent variables for stepwise regression analysis, so as to discuss whether PA and IL factors have significant predictive ability in MCE factor.
The result shows that the value of R2 Change is 0.505, indicating that the overall explanatory
ability of IL factor to predict MCE factor is 50.5%, which is significant (p=0.000), and F-value is 393.097, as shown in Table 4-7 and Table 4-8. The path coefficient result indicates that IL factor can predict MCE factor effectively, the β value is 0.697, and t value is 19.457, as shown in Table 4-9. The statistical significance shows that higher information literacy of elementary school teachers leads to higher mobile computer self-efficacy.
The result shows that the value of R2 Change is 0.011, indicating that the overall explanatory ability of PA factor to predict MCE factor is 1.1%, which is significant (p=0.003), and F-value is 205.141, as shown in Table 4-7 and Table 4-8. The path coefficient result indicates that IL factor can negatively predict MCE factor effectively, the β value is -0.107, and t value is 3.001, as shown in Table 4-9. The statistical significance shows that higher personality trait anxiety of elementary school teachers leads to lower mobile computer
self-efficacy.
Table 4-7 Summary of multiple regression model
Change Statistics
Table 4-8 ANOVA Summary
In the second regression analysis, MCA factor is set as the dependent variable, PA and IL as independent variables for stepwise regression analysis, so as to discuss whether PA and IL factors have significant predictive ability in MCA factor. The result shows that the value of R2 Change is 0.287, indicating that the overall explanatory ability of IL factor to predict MCA factor is 28.7%, which is significant (p=0.000), and F-value is 155.318, as shown in Table
4-10 and Table 4-11. The path coefficient result indicates that IL factor can predict MCA factor effectively, the β value is -0.497, and t value is 12.136, as shown in Table 4-12. The statistical significance shows that higher information literacy of elementary school teachers leads to lower mobile computer anxiety.
The result shows that the value of R2 Change is 0.079, indicating that the overall explanatory ability of PA factor to predict MCE factor is 7.9%, which is significant (p=0.000), and F-value is 111.148, as shown in Table 4-10 and Table 4-11. The path coefficient result indicates that PA factor can positively predict anxiety factor effectively, the β value is 0.284, and t value is 6.932, as shown in Table 4-12. The statistical significance shows that lower personality trait anxiety of elementary school teachers leads to lower mobile computer
anxiety.
Table 4-10 Summary of multiple regression model
Change Statistics
Table 4-11 ANOVA Summary
In the third regression analysis, MCA factor is set as the dependent variable, MCE as the independent variable for stepwise regression analysis, so as to discuss whether the MCE factor have significant negatively predictive ability in MCA factor. The result shows that the value of R2 Change is 0.402, indicating that the overall explanatory ability of MCE factor to
predict MCA factor is 40.2%, which is significant (p=0.000), and F-value is 260.464, as shown in Table 4-13 and Table 4-14. The path coefficient result indicates that MCE factor can predict MCA factor effectively, the β value is -0.634, and t value is 16.139, as shown in Table 4-15.The statistical significance shows that higher mobile computer self-efficacy of
elementary school teachers leads to lower mobile computer anxiety.
Table 4-13 Summary of multiple regression model
Change Statistics
Table 4-15 Coefficients Summary
Unstandardized Coefficients Standardized Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 1.643 0.234 7.027 .000
Mobile computer self-efficacy
0.698 0.043 -0.634 16.139 .000
In the forth regression analysis, UI factor is set as the dependent variable, MCE and MCA as independent variables for stepwise regression analysis, so as to discuss whether MCE and
In the forth regression analysis, UI factor is set as the dependent variable, MCE and MCA as independent variables for stepwise regression analysis, so as to discuss whether MCE and