Chapter III Methodology
3.3 Research Variables
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
45
Purposive sampling was applied with the inclusion criterion of female blog readers of beauty blogs. In order to reach the blog readers of beauty blogs, the researcher observed the comment function in popular beauty blogs, those readers were left messages on their own blogs to invite them to fill-out the questionnaire. Snowball sampling was also applied to allow questionnaire participants to make referrals. Moreover, questionnaire was posted on related online forum in PTT in order to target beauty blog readers.
3.2.2.2 Questionnaire Design
The questionnaire was composed of ten sections. In the first section, demographic information was required to help researcher draw a general blog reader profile. The second section asked about participants’ behavior on searching beauty0related information. In the third section, participants’ general level of satisfaction toward beauty blog content was observed. The fourth, fifth and six section asked participants’ attitude and their perceptions toward three different types of blog posts (Pure Sharing Blog Post, Sample Trial Blog Post, Product Placement Blog Post). In section 7 and 8, participants were asked to evaluate the two CGM platforms –PTT Cosmetics Board and FashionGuide Sample Trial Report. The ninth section was composed of information-sharing as well as purchasing behavior questions.
3.3 Research Variables
Demographic Variables
Age was measured as a continuous variable with “below 18” (coded ‘1’), “18 to 24”
(coded ‘2’ ), “25 to 29” (coded ‘3’), “30 to 34" (coded ‘4’), “35 to 39" (coded 5), “40 to 44”
(coded ‘6’), “45 to 49” (coded ‘7’ ), and “50 and above” (coded ‘8’). Level of education was an ordinal variable measured using seven categories, ranging from “elementary school”
(coded ‘1’) to “Ph. D” (coded ‘6’) and “Others” (coded ‘7’). Monthly income was measured
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
46
with the use of six categories, ranging from “under 10,000 NTD” (coded ‘1’) to “more than 90,000 NTD” (coded ‘6’).
Independent Variables
Level of Content Satisfaction refers to readers’ general level of satisfaction toward the content of beauty blogs and was measured with 10 items using 5-point Likert scale with “Strongly disagree” coded as 1, “Disagree” coded as 2, “Neutral” as 3, “Agree” as 4, and “Strongly agree” as 5.
Attitude toward Sample Trial Blog Post was measured with a single item that asked
participants what their attitudes toward beauty bloggers explicitly notifying readers about the blog posts with material connection are. 5-point Likert scale with “Strongly disagree” coded as 1, “Disagree” coded as 2, “Neutral” as 3, “Agree” as 4, and “Strongly agree” as 5 was applied.
Attitude toward Product Placement Blog Post was measured with a single item that asked what their attitudes toward beauty bloggers writing product placement articles without openly notifying the readers are. 5-point Likert scale with “Strongly disagree” coded as 1,
“Disagree” coded as '2, “Neutral” as 3, “Agree” as 4, and “Strongly agree” as 5 was applied.
WOM Credibility of Pure Sharing Blog Post was measured with Eight items that asked participants how much they agree with eight statements on whether positive as well as negative WOM in Pure Sharing Blog Post is (1) credible, (2) true, (3) can be valuable
reference, and (4) objective. An additive index from these eight items are created and treated as the index of WOM credibility in PSBP (M=3.56, SD=0.50).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
47
Frequency of Beauty Blog Reading was measured with a single item that asked participants the average time spent on reading beauty blogs each day.
Beauty-related Information Collecting involves blog readers pre-purchasing information gathering behavior. The Information collecting is the new index from the mean of following seven items using 5-point Likert scale, with Strongly disagree coded as “1”, Disagree as “2”, Neutral as “3”, Agree as “4”, and strongly agree as “5”: (1) When I need information about beauty-related products/services, I search online; (2) Compared to other information-collecting channels, reading beauty blogs is the most convenient; (3) I can quickly find the information I need in beauty blogs; (4) Reading beauty blogs has become my everyday habit;
(5) If I have any questions while reading the blog posts, I can find answers in the blog posts or relevant responses/comments; (6) I think the content of beauty blogs is detailed; (7) I think beauty bloggers’ personal experience and recommendation can help me make right
purchasing decision. (Cronbach’s alpha=0.82)
Dependent Variables
Interest-arousing in Products/ Services/ brands was measured with a single item using 5-point Likert scale, from “Strongly agree”coded as 5, “Agree” as 4, “Neutral” as 3, “Disagree”
coded as 2, and “Strongly disagree” as 1. (M=4.04, SD=.67) The statement is “I have been influenced by beauty blogs and further grown interests in products, services, or brands.” is treated as dependent variable indicating consumer purchasing behavior.
WOM Referral was measured using two 5-point Likert scale with “Strongly agree” coded as 5,
“Agree” as 4, “Neutral” as 3, “Disagree” coded as 2, and “Strongly disagree” coded as 1.
Correlation analysis was conducted on “I recommend beauty blogs to friends” and “Generally speaking, I will recommend my friends to use the products or services that I have good
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
48
evaluation.” Significance was found between this two items; thus this two items are put together to generate new index for WOM Referral (M=.80, SD.68).
Consumer Purchasing Behavior was measured with a single item that asked participants how much they agree on the statement “I have been influenced by beauty blogs and further
purchase products/services. 5-point Likert scale, from “Strongly agree” coded as 5, “Agree”
as 4, “Neutral” as 3, “Disagree” coded as 2, and “Strongly disagree” coded as 1. (M=3.92, SD=.80) is treated as dependent variable indicating consumer purchasing behavior.
Moderator Variable
WOM Credibility of Sample Trial Blog Post was measured with Eight items that asked participants how much they agree with eight statements on whether positive as well as negative WOM in Sample Trial Blog Post is (1) credible, (2) true, (3) can be valuable
reference, and (4) objective. An additive index from these eight items are created and treated as the index of WOM credibility in STBP (M=2.92, SD=0.87).
WOM Credibility of Product Placement Blog Post was measured with Eight items that asked participants how much they agree with eight statements on whether positive as well as negative WOM in Product Placement Blog Post is (1) credible, (2) true, (3) can be valuable reference, (4) objective. An additive index from these eight items are created and treated as the index of WOM credibility in PPBP (M=2.92, SD=0.87).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
49
CHPATER IV DATA ANALYSIS
The purposes of this study is to examine whether readers/consumers’ satisfaction toward beauty blog content would influence their interest, purchase behavior, and WOM referral. Moreover, the WOM credibility of three different types of blog post was examined to provide with more understanding on readers/consumers’ attitudes as well as tendencies toward buzz marketing among beauty blogs.
The data collected was run and analyzed using SPSS 16.0. The online survey was conducted from May 27, 2010 to June 9, 2010, and posted on My3Q website. 639 samples were collected, and 604 of which are valid, with 28 male consumers and seven system missing being excluded from the study.
4.1 Sample Structure
For the first section of the questionnaire, preliminary descriptive analysis was conducted to provide with a general demographic profile of participants on age, education, marital status, occupation, monthly income, and residency (Table 4-1). To further understand the internet-using frequency and blog-using habit, descriptive analysis was performed on average hours spent online per week, blog owning, average time spent on editing blogs, average time spent on reading beauty blogs, and beauty blog reading motivations (Table 4-2).
Age
Among 603 valid samples, the accumulation of two age groups account for the majority (83.5%); respectively age group 18 to 24 years old (46.7%) and 25 to 29 years old (36.8). Following up by age group 30 to 34 years old that accounts for 10.3% of the total samples.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
50
Level of Education
With the accumulated percentage of 94.4% that receive education up to
college/university or above out of 602 samples, it is shown that the majority of participants are fairly educated. 398 participants hold college/university degree (65.9%), 166 master degree (27.5%) and 6 doctor degree (1%).
Marital Status
552 (91.4%) participants are single, accounting for the great majority of all 600 valid samples. While 46 are married (7.6%), two participants fall under the group of
separated/divorced/widowed (0.3%).
Occupation
Among 598 participants that answered, 289 are students, accounting for 47.8%.
Second largest group falls in education, with 57 participants that account for 9.4%.
Participants who are unemployed (32 persons), in service industry (32 persons), in electronics or information industry (31 persons), in medial or environment industry (26 persons), and in communications industry (24 persons) share close percentages that rang from 4% to 5.3%.
The rest 107 participants accounting for 17.7% distributed in other 21 industries.
Monthly Income
More than 90% of the participants make less than 50,000 NTD per month, among which the largest group falls in making less than 10,000 NTD per month (278 participants, 46%), following by those who make 30,001 to 50,000 NTD (145 participants, 24%) and those who make between 10,001 to 30,000 NTD (132 participants, 21.9%). People who make more than 50,001 NTD account for a total of 43 persons, 7.1%. This distribution is in
‧
accordance with the previous question on age and occupation, in which younger group as well as students take up for majority of all participants.
Residency
Among 601 valid samples, 403 indicate their residency in northern Taiwan (66.7%), 71 in central Taiwan (11.8%), 86 in southern Taiwan (14.2%), 8 in eastern Taiwan (1.3%), and 10 in foreign countries (1.7%), including Japan, Hong Kong, Canada, United States, United Kingdom, France, and Guatemala.
Table 4-1. Descriptive Analysis: Demographics
Frequency Percentage (%) Age
Senior / Vocational High School 29 4.8
University/ College 398 65.9
Master 166 27.5
‧
Electronics / Information 31 5.1
Manufacturing 17 2.8
Communications 24 4.0
Logistics / Warehouse 5 0.8
Agriculture / Forestry / Fishery / Farming 1 0.2
Tourism / Leisure 4 0.7
‧
Average Hours Spent Online per Week
The largest group is composed of 278 participants out of 602, accounting for 46%, indicate their average time spent online is more than 20 hours per week, implying nearly half of the participants are heavy users of the internet. The second largest group is those who spend 5 to 8 hours per week, with 120 participants accounting for 19.9%. 78 (12.9%) have indicated their average time spent online per week is 9 to 12 hours, 49 (8.1%) indicated 13 to 16 hours, 42 (7%) indicated less than 5 hours, and 35 (5.8%) indicated 17 to 20 hours.
Blog Owning
There are 597 valid samples, among which 431 participants (71.4%) indicate blog owning, while 166 participants (27.5%) do not own any blogs.
Daily Average Time Spent on Editing Blog(s)
Among the 431 participants that suggest blog-owning, 423 has answered this question.
The majority, 383 participants that account for 88.9%, averagely spend less than 2 hours on
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
54
editing blog(s). 34 participants (7.9%) spend 2 to 4 hours, two spend 8 to 10 hours (0.7%), and only one spends more than 10 hours (0.4%) per day.
Average Time Spent on Reading Beauty Blogs
Up to 93.4% of the participants spend averagely less than 2 hours reading beauty blog(s) each day; among which those who spend less than 30 minutes per day account for the majority (290 participants, 48%). With the percentage of 29.8%, 180 participants spend 30 minutes to less than an hour to read beauty blogs. 93 participants accounting for 15.4% spend 1 to 2 hours. Participants that spend more than 2 hours each day account for 6.2%.
Beauty Blog Reading Motivations
Blog readers may have more than one motivation to read beauty blogs, therefore multiple choices are offered for better describing reading motivations. Among 603
participants, “to exchange information” was chosen by 363 participants (60.1%), “to learn knowledge” was chosen by 358 participants (59.3%), “to kill time” was chosen by 299 participants (49.5%), “to relax” was chosen by 243 participants(40.2%), “to seek suggestion”
was chosen by 197 participants (32.6%), “to exchange opinion” was chosen by 52 participants (8.6%).
From the descriptive result, it is clear that beauty blogs are considered more functional for readers as information-exchanging and knowledge-learning are the top two purposes of reading. Nonetheless, beauty blogs are also viewed as good tools for time-killing and relaxation. In other words, beauty blogs serve multiple functions; readers can not only look for helpful information or practical knowledge, but also can relax as well as get rid of boredom. Moreover, readers can even seek for suggestions on decision-making.
‧
Table 4-2. Descriptive Analysis: Internet-using Frequency and Blog-using Habit Frequency Percentage (%) Average Online Hours Per Week
Less than 5 hours 42 7.0
Average Time Spent on Editing Blog(s)
Less than 2 hours 383 88.9
Average Time Spent on Reading Beauty Blog(s)
Less than 30 minutes 290 48.0
‧
To exchange information 363 60.1
To exchange opinion 52 8.6
4.2 Factor Analysis and Reliability Analysis
In the second section, questions 13 to 23 are asked about consumer’s behavior on searching beauty-related information. From the descriptive analysis, it is observed that the mean of question 20 is 3.13 (SD=1.07), indicating the overall attitude toward “The
impression/number of clicks of beauty blogs will influence my willingness to read” is quite neutral. Therefore, question 20 was excluded. The rest of data was subjected to Factor analysis that applied Principle Axis Factoring extraction method and Promax Rotation method. A two -factor solution for 10 items was deemed appropriate.
Factor 1 (Eigenvalue = 4.22) was labeled “Beauty-related Information Collecting”
and accounted for 42.20% of the total variance. Factor 2 (Eigenvalue = 1.77) was labeled
“Willingness to Read” and accounted for 17.71% of the total variance. Table 4-3 shown below contains the 10 items and their corresponding factor loadings. The reliability (Cronbach’s alpha) of two factors are respectively .82, and .88.
Table 4-3. Factor Analysis: Consumer Behavior on Searching Beauty-related Information
# of items Reliability
(Cronbach’s α) Items
‧
Extraction Method: Principle Axis Factoring.
Rotation Method: Promax with Kaiser Normalization.
Loadings are greater than .50, Eigenvalues > 1
Level of Content Satisfaction
In Section 3, 10 questions were asked to observe readers’ general level of satisfaction toward the content of beauty blogs. 575 samples are valid among the total of 604 samples.
The content satisfaction refers to readers’ general level of satisfaction toward the content of beauty blogs by measuring 10 items using 5-point Likert scale. With Strongly disagree coded as “1”, Disagree as “2”, Neutral as “3”, Agree as “4”, and strongly Agree as “5”, 10 items are as follows: (1) Generally speaking, the content of beauty blogs is interesting; (2) Generally speaking, the content of beauty blogs is frequently updated; (3) Generally speaking, the content of beauty blogs is abundant; (4) Generally speaking, the content of beauty blogs is correct; (5) Generally speaking, the content of beauty blogs is credible; (6) Generally speaking, the content of beauty blogs is interactive; (7) Generally speaking, the content of beauty blogs is true; (8) Generally speaking, the content of beauty blogs is immediate; (9) Generally speaking, the content of beauty blogs is close to my life; (10) Generally speaking,
‧
the content of beauty blogs is influential. An additive index “Level of Content Satisfaction”
from these 10 items was created with Cronbach’s alpha=0.85 (M=3.65, SD=.49).
Section 4, 5 and 6 paid closer attentions to readers’ attitudes toward three different types of blog post—(1) Pure Sharing Blog Post (PSBP), which is the non-commercial content that is not operation of buzz marketing; (2)Sample Trial Blog Post (STBP), which involves commercial content and is operationalized as explicit buzz marketing; (3)Product Placement Blog Post (PPBP), which involves commercial content and is operationalized as implicit buzz marketing. First, descriptive analysis was conducted on all items from these three sections. The mean scores of corresponding items were compared to provide with a general profile on readers’ attitudes on specific aspects, shown as Table 4-4.
Table 4-4. Mean Scores of Readers’ Attitudes toward PSBP, STBP, and PPBP.
Pure
Positive WOM increase purchase intention 3.80 3.34 2.88 Interested in negative WOM 3.84 3.75 3.31
Negative WOM is credible 3.73 3.63 3.21
Negative WOM is true 3.66 3.61 3.20
Negative WOM can be reference 3.93 3.80 3.33
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
59
Negative WOM is objective 3.38 3.45 3.05
Negative WOM is commercial 2.91 3.16 3.18
Negative WOM decrease purchase intention 4.00 3.86 3.55
For a clearer comparison, mean scores are put into a horizontal bar graph (Figure 4-1).
Figure 4-1. Horizontal Bar Graph on Readers’ Attitudes toward PSBP, STBP, and PPBP.
Negative WOM decrease purchase intention Negative WOM is commercial
Negative WOM is objective Negative WOM can be reference Negative WOM is true Negative WOM is credible Interested in negative WOM Positive WOM increase purchase intention Positive WOM is commercial Positive WOM is objective Positive WOM can be reference
Positive WOM is true Positive WOM is credible Interested in positive WOM
0 1 2 3 4 5
Mean Score
Pure Sharing Blog Post Sample-trial Blog Post Product Placement Blog Post
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
60
With regard to positive WOM, readers show much agreement on the statement that WOM in Pure Sharing Blog Post arouse their interest, rather than the other two types of blog posts. In general, readers agree the most on Pure Sharing Blog Post being more credible, true, worthwhile to be reference, objective, as well as increasing purchase intention while readers also appear tendency to agree on the above statements in Sample Trial Blog Post. As to Product Placement Blog Post, readers show more disagreement on its credibility and
recognize those posts to be the most commercial. It was also indicated that positive WOM in Product Placement Blog Post has the lowest influence on consumers’ purchase intention.
However, it is evident that generally, despite different types of blog post, readers consider negative WOM relatively more credible, truer, more worthy as a reference, more objective, less commercial than its positive counterpart. Readers also agree more on the statement that negative WOM may decrease their purchase intention than positive WOM may increase their purchase intention. This is consistent with previous literature on negative WOM may have more influence than positive WOM on consumers’ decision making and purchasing behavior.
In addition, for the purpose of observing readers’ attitudes toward two kinds of buzz marketing, explicit and implicit, participants were asked on their level of agreement on Sample Trial Blog Post (explicit buzz marketing) and Product Placement Blog Post (implicit buzz marketing). It is found that on being notified, readers will have relatively higher
acceptance on buzz marketing (Sample Trial Blog Post, M=3.48) while the implicit buzz marketing will cause lower agreement (Product Placement Blog Post, M=2.37).
WOM Credibility of Pure Sharing Blog Post was measured with eight items on 5 point Likert scale with Strongly Disagree coded as “1”, Disagree as “2”, Neutral as “3”, Agree as “4”, and strongly Agree as “5”: (1) I think the positive WOM in Pure Sharing Blog
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
61
Post is credible, (2) I think the positive WOM in Pure Sharing Blog Post is true, (3) I think the positive WOM in Pure Sharing Blog Post can be valuable reference, (4) I think the positive WOM in Pure Sharing Blog Post is objective. (5) I think the negative WOM in Pure Sharing Blog Post is credible, (6) I think the negative WOM in Pure Sharing Blog Post is true, (7) I think the negative WOM in Pure Sharing Blog Post can be valuable reference, (8) I think the negative WOM in Pure Sharing Blog Post is objective. 586 out of a total of 604 samples are valid; with Cronbach’s alpha= .85, an additive index from these eight items are created and treated as the index of WOM Credibility in PSBP (M=3.56, SD=.50).
WOM Credibility of Sample Trial Blog Post was measured with eight items on 5 point Likert scale with Strongly Disagree coded as “1”, Disagree as “2”, Neutral as “3”, Agree as
“4”, and strongly Agree as “5”: (1) I think the positive WOM in Sample Trial Blog Post is credible, (2) I think the positive WOM in Sample Trial Blog Post is true, (3) I think the positive WOM in Sample Trial Blog Post can be valuable reference, (4) I think the positive WOM in Sample Trial Blog Post is objective. (5) I think the negative WOM in Sample Trial Blog Post is credible, (6) I think the negative WOM in Sample Trial Blog Post is true, (7) I think the negative WOM in Sample Trial Blog Post can be valuable reference, (8) I think the negative WOM in Sample Trial Blog Post is objective. 580 out of a total of 604 samples are valid; with Cronbach’s alpha= .82, an additive index from these eight items are created and treated as the index of WOM Credibility of STBP (M=3.42, SD=.51).
WOM Credibility of Product Placement Blog Post was measured with eight items on 5 point Likert scale with Strongly Disagree coded as “1”, Disagree as “2”, Neutral as “3”, Agree as “4”, and strongly Agree as “5”: (1) I think the positive WOM in Product Placement Blog Post is credible, (2) I think the positive WOM in Product Placement Blog Post is true, (3) I think the positive WOM in Product Placement Blog Post can be valuable reference, (4) I
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
62
think the positive WOM in Product Placement Blog Post is objective. (5) I think the negative WOM in Product Placement Blog Post is credible, (6) I think the negative WOM in Product Placement Blog Post is true, (7) I think the negative WOM in Product Placement Blog Post can be valuable reference, (8) I think the negative WOM in Product Placement Blog Post is
think the positive WOM in Product Placement Blog Post is objective. (5) I think the negative WOM in Product Placement Blog Post is credible, (6) I think the negative WOM in Product Placement Blog Post is true, (7) I think the negative WOM in Product Placement Blog Post can be valuable reference, (8) I think the negative WOM in Product Placement Blog Post is