Chapter III Methodology
3.5 Data Collection and Analysis Methods
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
48
3.4.2 In-depth Interview
Due to the constraints of questionnaire design, an in-depth interview will be used to solicit more information from the customer of Brand’s InnerShine beauty drinks.
Five or six respondents of the previous questionnaire survey will be selected as interviewees. In order to add up the diversity of data, respondents who are different from each other on demographic characteristics will be prioritized for interview. The format of interview will be semi-structured for it is considered as an effective way of probing for more information and clarification of answers. Also, samples with the varied professional, educational and personal histories are able to provide more insights, whose answers may not be well captured in the standardized questionnaire (Barriball and While, 1994).
3.5 Data Collection and Analysis Methods
The data collected with both quantitative and qualitative methods will be adjusted and analyzed with selected tools. Details are explained in the session.
3.5.1 Questionnaire Survey
The results of survey will be coded and inputted in SPSS version 18.0 statistical analysis computer package. Selected strategies for data analysis will be described in the session.
1. Descriptive statistics analysis
To observe the essential features and the tendency of the sampling, descriptive statistics will be adopted. Frequency scores, measure of central tendency and dispersion will be summarized and exhibited by calculating the mean scores and standard deviation, for the tendency of conformity and disconformity from the data can be revealed by the two measures.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
49
2. Factor analysis
Factor analysis helps researchers to extract the meaningful and unobserved latent variables from the original ones. To convert a set of possibly correlated variables into a set of linearly uncorrelated variables, principal component analysis (PCA) will be used. Some uncorrelated variables can be therefore excluded via the varimax of orthogonal rotation; the rest of components which are not correlated will be excluded.
3. Reliability analysis
To ensure the consistency of the measured results, reliability test is indispensable.
Cronbach's alpha, often used in testing the summated scales, is able to determine the internal consistency or average correlation of items in a survey instrument and further gauge its reliability (Santos, 1999). For the Likert scale is used in the study, Cronback’s alpha is selected. Reliability coefficient recognized as acceptable should be 0.7 at least, but lower thresholds are used in some studies (Nunnaly, 1978;
Santos, 1999).
4. T-test
T-test is a tool for evaluating the differences in means between two groups. It is usually used when independent variables are categorical variables (e.g. gender, residential area), while dependent variables are continuous ones. Independent samples t-test and paired-samples t-test are most employed methods; the former tests two samples in different populations, while the latter one tests paired sample within the same population (e.g. before and after). In the study, independent samples t-test will be adopted.
5. Pearson product-moment correlation analysis
To examine the strength of relationship between variables, correlation analysis is employed. Since Pearson product-moment correlation coefficients is the most
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
50
frequently applied tool for measuring the association between two interval variables (Sarankatos, 2005), it will also be adopted in the study. The range of coefficient falls into +1.00 and -1.00 which indicates a perfect positive correlation and a perfect negative correlation respectively. Besides, there is no linear relationship if an r is closed to 0 (Baggaley, 1964). It will be applied to analyze the correlations among perceived value, brand image, customer satisfaction, and customer loyalty in the research.
6. Regression analysis
The goal of regression analysis is to determine the values of parameters for a function that results in the function which best fits a set of data observations. To investigate the relationship between two independent variables on a dependent variable, multivariate regression will be adopted in the study. Specifically, the impact of brand image and perceived value on customer satisfaction, the impact of brand image and perceived value on customer loyalty, and the impact of satisfaction on loyalty will be examined by multivariate regression.
7. Analysis of variance, ANOVA
ANOVA tests whether or not the means of several groups are all equal, and it is appropriate for comparing two or more samples. If significant difference is found among the observed variables, the source of variance contains the great impact and can be recognized as the cause of the variance. One-way ANOVA is used when the data are divided into groups according to only one factor; hence, it can be adopted for testing whether there is significant variance of each demographic characteristic on dependent variables (customer satisfaction and customer loyalty).
3.5.2 In-depth Interview
The raw data from semi-structured interview will be collected by digital recording and further transcribed word by word as print copy. Content analysis will be employed
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
51
for interpreting and systemizing the data. The procedure for data analysis will primarily follow the guidelines established by Thomas (2000): (1) Preparation of raw data files (2) Close reading of text (3) Creation of categories (4) Overlapping coding and uncoded text (5) Continuing revision and refinement of category system. The result of analysis from interview will be compared with the results from quantitative analysis. An overall suggestion for marketing strategy of Brand’s InnerShine beauty drinks will be made accordingly.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
52
Chapter 4 Data Analysis
4.1 Sample Structure
The online survey of the study was administered on the my3q.com website from April 26 to May 13, 2012. 492 samples were collected. After ruling out the missing data, only 465 samples are valid. The response rate for the valid questionnaire is 94.51%. The valid data were analyzed by SPSS 17.0.
A general demographic profile of the respondents was provided in the questionnaire on gender, age, level of education, occupation, marital status, personal monthly income, and residential area (see Table 4-1-1). For a specific understanding of respondents' habits of consumption, question items for the inquiry of their purchase channel and the frequency of their purchase are listed (see Table 4-1-2).
Gender
The samples in the study shows that females are the majority, which accounted for 89.9%, while males are only accounted for 10.1%.
Age
Among 465 samples, group under 25 is the majority with the most accumulated percentage of 40.6%. Group 26 to 30 is the second largest plurality by accounting for 38.8 %. The amount of percentage of other groups including 31 to 35 (4.5%), 36 to 40 (4.3%), 46 to 50 (4.1%), and above 51 (4.5%) are close to one and another. The group with the least amount falls into 41 to 45 year-old respondents, of which is 3.2%.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
53
Level of Education
Shown in the table, group "university/college" and group "master" are nearly as many as each other, with 39.5% and 38.7% of total amount respectively. Group senior high accounted for 21.7%. Other than the above ones only accounted for 0.1%.
Occupation
With the most accumulated percentage of 35.3%, students were the majority of the respondents. Respondents who answered "others" (which are other than the listed occupations) were the second largest group, which accounted for 26.4%. Group service industry had a considerable proportion with percentage of 19.5%.
Marital Status
Most respondents in the study are single, accounting for 93.9% of all. While only 6.1%
of respondents are married.
Personal Monthly Income
The great majority falls into both group 20,001 to 30,000 NTD and group under 20,000 NTD, which accounted for 36.1% and 35.7% respectively. With the accumulated percentage of 13.1, group 30,001 to 40,000 NTD is the third place among all groups.
Group 40,001 to 50,000 NTD and group 50,001 to 60,000 NTD accounted for 9.3%
and 3.2% respectively. The least amount of the group falls in group above 60,001 NTD with only 2.6%.
Residential Area
Being inquired for the residential area, most respondents categorized themselves as people who live in Northern Taiwan, with 81.4%. The second largest group is Southern Taiwan, accounting for 15.6%. Respondents from other area are relatively few; there is
‧
2.8% of them came from Central Taiwan. Only 0.1% came from Eastern Taiwan and Outlying islands.
Table 4-1-1: Descriptive Analysis of Demographics
Frequency Percentage （%）
University/College 184 39.5
Master 177 38.7
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
55
30,001 to 40,000 NTD 61 13.1
40,001 to 50,000 NTD 43 9.3
50,001 to 60,000 NTD 15 3.2
Above 60,001 NTD 12 2.6
Residential Area
Northern Taiwan 377 81.4
Central Taiwan 12 2.8
Southern Taiwan 72 15.6
Eastern Taiwan 3 0.1
Outlying islands 1 0.1
Others 0 0
Aside from the demographic data, consumers’ using experiences are investigated in the questionnaire as well. Numbers and details are shown as follows:
Purchase Channel
The results of the answers for the question item “which channel do you purchase Brand’s InnerShine beauty drinks” are summarized as follows: According to the data, most respondents went to physical stores to buy the products, for the 69.2% of all had such experiences. Those who have purchased Brand’s InnerShine beauty drinks in online stores only accounted for 18.7%. There is also 12.1% of all have shopped for the products in both physical and online stores.
Purchase Frequency
The results of the answers for the question item “how frequent do you purchase Brand’s InnerShine beauty drinks” are precisely shown in the below table. With the most accumulated percentage of 44.6, respondents who haven’t been purchased the products within a year are the majority. Respondents who have been purchased the products “once a month,” “once per three months” and “once per six months”
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
56
accounted for 15.9%, 16.1%, and 13.3% respectively. Group “once a year” is 6.9%.
With the least proportion, group “once a week” accounted for only 3.2%.
Table 4-1-2: Descriptive Analysis of Consumers’ Using Experiences
Frequency Percentage （%）
Purchase Channel Physical stores
（e.g. over-the-counter stores, convenience stores, shopping malls, supermarket, Brand’s chain stores）
322 69.2
Online stores（e.g. auction websites, Brand’s official website）
87 18.7
All of the above 56 12.1
Purchase Frequency
Once a week 15 3.2
Once a month 74 15.9
Once per three months 75 16.1
Once per six months 62 13.3
Once a year 32 6.9
Less than once within a year 207 44.6
‧
After being ruled out the missing data, the valid data of the questionnaire will be firstly subjected to descriptive analysis. Mean and standard deviations of each variable (brand image, perceived value, customer satisfaction, and customer loyalty) is analyzed and displayed in order.
1. Brand Image
There are 18 question items in brand image construct. The mean of each item ranges 3.68 to 5.23, and the standard deviation ranges 0.832 to 1.416. The highest mean occurs in the item no. 14 “Compared with other brands of functional drinks, Brand’s is a trust-worthy brand” (5.23); while the lowest mean occurs in the item no. 10 “Drinking Brand’s InnerShine beauty drinks makes more attractive to the opposite sex due to the prettier appearance” (3.68). Means and standard deviations in each item are described in Table 4-2-1 as follows:
Table 4-2-1: Frequency Distribution of Question Items in Brand Image
Question Items M Order S.D.
14. Compared with other brands of functional drinks, Brand’s is a trust-worthy brand. 5.23
1 0.832
17. Brand’s is a reliable brand due to its long history. 5.23 2 0.975
18. Due to the higher market share of beauty drinks, Brand’s is a trust-worthy brand. 5.13
3 1.009
15. The brand image of Brand’s is associated with the concept of heath. 5.13
3 1.009
16. Brand’s is a reliable brand due to its several national authentications. 5.1
4 0.996
5. Drinking Brand’s InnerShine beauty drinks makes me a good complexion. 4.68
5 1.148
4. Brand’s InnerShine beauty drinks make my skin healthier. 4.65
6 1.003
3. The product of Brand’s InnerShine beauty drinks is reliable due to its natural
ingredients additive. 4.55 7 1.043
12. The fancy packing of Brand’s InnerShine beauty drinks facilitates me to purchase
them. 4.52 8 1.105
1. Brand’s InnerShine beauty drinks help beautify my skin. 4.42
9 1.041
‧
2. Drinking Brand’s InnerShine beauty drinks make me healthier by fortifying my
metabolism. 4.35 10 1.124
11. Drinking Brand’s InnerShine beauty drinks gives me a refreshing feeling. 4.29
11 1.225
9. Buying Brand’s InnerShine beauty drinks makes me more connected with others
by sharing the topic in common. 4 12 1.416
8. Drinking Brand’s InnerShine beauty drinks makes me more confident to myself. 3.97
13 1.284
6. I can show my taste by drinking Brand’s InnerShine beauty drinks. 3.87
14 1.071
13. Drinking Brand’s InnerShine beauty drinks makes me a good complexion and
getting more complements. 3.87 14 1.315
7. After drinking Brand’s InnerShine beauty drinks, I was admired by friends and
colleagues due to the prettier appearance. 3.84 15 1.222
10. Drinking Brand’s InnerShine beauty drinks makes more attractive to the opposite
sex due to the prettier appearance. 3.68 16 1.12
2. Perceived Value
There are 15 question items in perceived value construct. The mean of each item ranges 3.65 to 4.77, and the standard deviation ranges 0.87 to 1.348. The highest mean occurs in the item no. 19 “The quality of Brand’s InnerShine beauty drinks is good” (4.77); whereas the lowest mean occurs in the item no. 24 “Brand’s InnerShine beauty drinks have good value for money” (3.65). Means and standard deviations in each item are described in Table 4-2-2 as follows:
Table 4-2-2: Frequency Distribution of Question Items in Perceived Value
Question Items Mean Order S.D.
19. The quality of Brand’s InnerShine beauty drinks is good. 4.77
1 0.87
20. The quality of Brand’s InnerShine beauty drinks meets my expectations
for it is the leading brand among functional drinks. 4.74 2 1.078
22. The quality of Brand’s InnerShine beauty drinks is trust-worthy. 4.74 3 0.916
30. Brand’s InnerShine beauty drinks are tasty. 4.71 4 1.225
29. Brand’s InnerShine beauty drinks are delicious. 4.68 5 1.306
33. The Brand’s InnerShine beauty drink is a popular functional drink. 4.52 6 1.133 21. Based on my experiences of consumption, I think Brand’s InnerShine
beauty drinks is effective. 4.39 7 0.94
‧
28. Drinking Brand’s InnerShine beauty drinks makes my energetic. 4.35 8 1.234 27. Drinking Brand’s InnerShine beauty drinks makes me refreshed. 4.06 9 1.218 25. The function of its product and price meet my expectation. 4 10 1.048 32. After drinking Brand’s InnerShine beauty drinks, I was admired by others
due to my pretty and healthy appearance. 3.97 11 1.093
31. Consuming Brand’s InnerShine beauty drinks makes me a favorable
impression in front of others. 3.9
12
1.255 26. Drinking Brand’s InnerShine beauty drinks makes me pretty. 3.84 13 1.348 23. The cost of Brand’s InnerShine beauty drinks is reasonable. 3.68 14 1.23 24. Brand’s InnerShine beauty drinks have good value for money. 3.65 15 0.936
3. Customer Satisfaction
There are 11 question items in perceived value construct. The mean of each item ranges 4.1 to 5.35, and the standard deviation ranges 0.735 to 1.187. The highest mean occurs in the item no. 42 “It is convenient to purchase Brand’s InnerShine beauty drinks via physical stores, TV channels, or online stores” (5.35); while the lowest mean occurs in the item no. 41 “When there are some problems with the products, the customer service (e.g. telephone or official website) of Brand’s are available for assisting me to solve the problems” (4.1). Means and standard deviations in each item are described in Table 4-2-3 as follows:
Table 4-2-3: Frequency Distribution of Question Items in Customer Satisfaction
Question Items M Order S.D.
42. It is convenient to purchase Brand’s InnerShine beauty drinks via physical
stores, TV channels, or online stores. 5.35 1
0.97 35. I am satisfied with the design of Brand’s InnerShine beauty drinks. 4.87 2 0.943 43. The instruction of Brand’s InnerShine beauty drinks is precise. 4.84 3 1.168 44. Overall, Brand’s InnerShine beauty drinks are satisfactory. 4.81 4 1.03 34. I am satisfied with the ingredients (e.g. berry essence and plum extract). 4.74 5 0.95 36. I am satisfied with the utility of Brand’s InnerShine beauty drinks. 4.58 6 1.072 37. I am satisfied with the packing of Brand’s InnerShine beauty drinks (e.g. the
volume, the design of bottle) 4.58 6
1.187
‧
38. I am satisfied with the flavor of Brand’s InnerShine beauty drinks. 4.55 7 1.043 39. The product-relevant information which Brand’s InnerShine beauty drinks
offer is enough for me to make decisions during the purchase. 4.55 7
0.946 40. The post-purchasing service (e.g. refund or consultation) is satisfying. 4.19 8 0.859 41. When there are some problems with the products, the customer service (e.g.
telephone or official website) of Brand’s are available for assisting me to solve the problems. ranges 3.19 to 4.55, and the standard deviation ranges 0.735 to 1.187. The highest mean occurs in the item no. 45 “I am willing to purchase Brand’s InnerShine beauty drinks next time” (4.55); while the lowest mean occurs in the item no. 55 “A slight increase of price of Brand’s InnerShine beauty drinks is acceptable due to the fluctuation of price.” (3.19). Means and standard deviations in each item are described in Table 4-2-4 as follows:
Table 4-2-4: Frequency Distribution of Question Items in Customer Loyalty Section
Question Items M Order S.D.
45. I am willing to purchase Brand’s InnerShine beauty drinks next time. 4.55 1 1.043 52. I am willing to buy Brand’s InnerShine beauty drinks with others. 4.45 2 1.161 58. Based on my personal needs, I am willing to buy other product series of
Brand’s (e.g. chicken essence, clam essence, bird’s nest) 4.42 3
1.041 51. I am willing to share my positive experiences of usage of Brand’s InnerShine
beauty drinks. 4.26 4
1.047 46. Brand’s is my top priority for choosing the beauty drinks. 4.23 5 1.362 49. I am willing to recommend Brand’s InnerShine beauty drinks to my families
and friends.。 4.23 5
1.157 50. I am willing to recommend Brand’s InnerShine beauty drinks to my
classmates/colleagues. 4.19 6
1.091 57. Concerning with the price and the product quality, I think it is a right decision
to purchase Brand’s InnerShine beauty drinks. 4.19 6
1.061 47. I am willing to purchase Brand’s InnerShine beauty drinks constantly. 4.16 7 1.051
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
61
48. I am willing to become a loyal customer of Brand’s InnerShine beauty drinks. 4.03 8 1.093 54. The price of Brand’s InnerShine beauty drinks is acceptable. 3.74 9 1.192 59. I would like to receive DMs of other product series of Brand’s. 3.68 10 1.355 53. I am willing to share the product-relevant information of Brand’s InnerShine
beauty drinks with others (e.g. To forward the DM ) 3.65 11
1.034 56. Brand’s InnerShine beauty drinks will be my top priority even if there are other
beauty drinks with a favorable price. 3.61 12
1.288 55. A slight increase of price of Brand’s InnerShine beauty drinks is acceptable due
to the fluctuation of price. 3.19 13
1.331
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
62
4.3 Factor Analysis and Reliability Test
In the section, factor analysis and reliability test will be conducted on all the variables in the research including brand image, perceived value, customer satisfaction, and customer loyalty. To better examine the interrelationship among these parameters with the most concise information, factor analysis is required in the preliminary phase of data analysis. Reliability test means to examine the interior consistency and stability of variables. The exact methods for conducting the analysis are described as follows:
1. A test the correlation among variables
To ensure the existence of correlation among variables, Kaiser-Meyer-Olkin (KMO) measure is used to test the sample adequacy before factor analysis. The value of KMO on parameters falls into the range of zero to one; the higher the value, the more suitable for the application of factor analysis. Kaiser (1974) proposed a calibration of the index as a standard to gauge the appropriateness of variables for factor analysis as follows: In the .90s — marvelous; In the .80s — meritorious; In the .70s — middling; In the .60s — mediocre; In the .50s — miserable;
Below .50 — unacceptable. By the standard, once the value of KMO is below .05, the parameter is considered unsuitable for factor analysis.
The other test should be used prior to the implementation of factor analysis is Bartlett’s test of sphericity (Bartlett, 1950), which speculates the significance of the extent of correlation between parameters by displaying the chi-square value (𝑋2).
The higher the chi-square value along with the significant p value is less than .05, the higher correlation between parameters. A high chi-square value also suggests several factors are likely to be extracted from those examined variables (Hsieh，
2008). Table 4-3-1 with KMO value of each parameter higher than .07 and lower than .05 p value, suggesting that data collected from each parameter are suitable for
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
63
factor analysis.
Table 4-3-1: Results of KMO and Bartlett’s Test for Constructs
Variables KMO Bartlett’s Test of Sphericity
𝑥2 Goodness-of-fit Test df p Value
Brand Image .835 4729.012 55 .000***
Perceived Value .757 7923.871 91 .000***
Customer Satisfaction .742 2571.806 28 .000***
Customer Loyalty .789 6298.789 55 .000***
2. Factor extraction and rotation
To extract factors from raw data, principle component solution with orthogonal rotation is applied. Varimax rotation subjected here ensures the variance of loading within factors is the most. The criterion of choosing the factors agreed in most studies is suggested by Kaiser (1966): to choose factors with eigenvalue >1.
Regarding to the criterion for choosing factors according to factor loading, Hairs et
Regarding to the criterion for choosing factors according to factor loading, Hairs et