Understanding consumer online shopping behaviour from the perspective of transaction costs

4.4 DATA COLLECTION 
In order to make the sample represent the whole online shoppers in China, one economically 
developed city and one economically less-developed city were chosen. The developed city 
has a relatively higher level of education, income, computer literacy, higher frequency of 
information exchange, wider Internet usage, and better development of information industry, 
leading to a higher percentage of online shoppers, while the less-developed city has a lower 
level of education and income, and a lower scale of development of information industry, 
resulting in a lower percentage of online shoppers. In China, most developed cities are 
located in the coastal areas of China. Due to the Reforming and Opening Policy, coastal areas 
have developed at the very fast speed and become the most developed economic region since 
the convenient sea transportations have raised a large number of exports. On the contrary, 
inland areas of China are lack of transport facilities and thus become the less-developed 
regions. 

---

169
In this study, the developed city was randomly selected from the pool of coastal cities in 
China, and the less-developed city was randomly selected from the pool of inland cities of 
China. The sample can be used as a good representative of general online shoppers in China. 
Specifically, the names of five large coastal cities from China were written on pieces of paper 
and folded in the same manner. One had been drawn at random. Similarity of the paper and 
the way the pieces of paper were folded ensure that a random selection was made. Ultimately, 
data for the study was gathered from the city of Shanghai as a representative of the developed 
city. Similarly the names of five small inland cities from China were written on similar pieces 
of paper and folded in the same manner. One was selected at random. Similarity of the paper 
and the way the pieces of paper were folded ensure that a random selection was made. Finally, 
the data from the less-developed cities was gathered from the city of Nanchong. 
It can be noted that many previous studies have used a sample drawn only from universities 
and colleges. The research took the view that the survey of just universities would exclude 
people who work in offices and shop online from locations other than universities (Rinnawi 
2002), therefore such samples are not representative of the total online shoppers’ population 
(Van Slyke
 et al.
2004). In addition, student characteristics may differ from the general 
online shoppers’ characteristics. For example, cost-conscious student consumers may weight 
price heavily in their purchase decisions, which may be different from the population of 
online shoppers as a whole. Students may also be more risk-taking, more innovative and 
more trusting in online vendors than the elderly consumers. Accordingly, this study suggests 
that the student samples may limit the generalizability of the whole online shopper population.
In addition, a large number of studies have used the Internet as the data collection cool. 
Studies using email to announce their surveys sent an invitation email to a mailing list, with a 
URL link to the survey web site. Although e-mail-based web surveys have demonstrated 

---

170
superiority over postal surveys in terms of response speed and cost efficiency (Flaherty
 et al.
1998, Weible and Wallace 1998, Sheehan and Hoy 1999, Sheehan and McMillan 1999), its 
disadvantages should not be overlooked.
The largest defect is its response rate. The response rates for e-mail-based web surveys may 
not match those of other survey methods (Cook
 et al.
2000, Couper 2000), since individuals' 
overall attitudes toward the unsolicited e-mail survey may be unfavourable. The increase in 
unsolicited e-mail to Internet users and the ill will generate among potential respondents can 
be viewed as an important reason for the lower response rates (Frost and Strauss 2013). This 
would also increase the difficulty for the researcher in planning to use e-mail surveys as it is 
likely that some types of unsolicited respondent contacts will be necessary when using 
random sampling techniques. Studies show that some Internet users receive more than 39 
unsolicited e-mails per day at the workplace alone (NUA Internet Surveys 2000a). The 
information overload causes individuals to develop ways for dealing with e-mail, which 
includes using filtering software or developing heuristics such as deleting all unsolicited e-
mail without opening it. Besides, potential survey participants may be concerned about the 
Internet security, such as the threat of viruses delivered from unsolicited e-mail (Sills and 
Song 2002), which discourages potential participants from reading unsolicited e-mail survey. 
In addition to the low response rates of the e-mail survey, obtaining thousands of email 
addresses of online shoppers is a big challenge. Also, issues such as changing Internet 
Service Provider and e-mail address, and holding of multiple e-mail addresses by a single 
individual have consequences for under-representation (Bradley 1999). Given the 
aforementioned reasons, e-mail-based web surveying would not be an appropriate data 
collection method in the study.

---

171
To overcome the potential sampling problems, this study was designed to gather data on the 
streets through using a face-to-face survey in Shanghai and Nanchong, at the regular 
thoroughfares or near the shopping mall entrances, where the researchers could meet people 
from all walks of life. The method of surveying people on the streets was appropriate for this 
study in terms of the quality and complexity of the data collected. The reasons can be 
specifically explained as follows. On-street data collection method could include different 
segments of online shoppers, for example those who adopt online shopping at home, in 
offices or in Internet cafés (Legris
 et al.
2003), resulting in generalizability and 
representativeness of the whole online shopper population in China.
Compared with e-mail surveys, face-to-face surveys offer significant advantages in terms of 
the amount and complexity of the data that can be collected. For example, face-to-face 
surveys can be significantly longer (Holbrook
 et al.
2003). Most people will allow a 
researcher to conduct the research for up to an hour, whereas respondents will typically not 
tolerate e-mail surveys that require more than 15 or 20 minutes of effort (Doyle 2009). The 
additional length allows researchers to design more questions, longer questions, more 
detailed questions, and more complicated questions (Doyle 2009). Since sixteen variables 
were included in the research model of this study, the survey questionnaire was very long. As 
such, the face-to-face survey was considered as the most appropriate method for this study.
The face-to-face survey could improve the response rate. Since it is much more difficult for 
people to refuse the invitation to participate in the face of a live human being than toss a 
written survey into the recycling bin with the junk mail or hang up on a disembodied voice, 
face-to-face surveys typically offer the highest response rates obtainable (over 90% in some 
cases) (Doyle 2009). From the respondent’s point of view, face-to-face survey could 

---

172
effectively address any questions or problems that arise (Doyle 2009). If the respondent finds 
a question to be confusing or ambiguous, the researcher can immediately clarify it. Based on 
the above arguments, the sample drawing from different streets through a face-to-face survey 
in Shanghai and Nanchong allows the capture of most of the population segments that shop 
online. The sample can be generalizable to Chinese populations of online shoppers. 
To collect data on the streets, the researcher contacted people and inquired whether they 
would like to participate by undertaking the survey and if yes, whether they had online 
purchase experience. Only respondents with positive answers were asked to continue on and 
answer the questionnaire. This screening question was employed to ensure that the sample 
consisted only of online shoppers. In order to ensure the respondents recalled an experience 
they were familiar with, the survey instructed them to recall the most recent purchase they 
made from an online vendor. They were also asked to write down the online store’s name and 
the product(s) and/or service(s) they bought from the online store, and attempted all questions 
based on that particular purchase experience. 
The survey was conducted on a one-to-one basis considering it might be hard to gather 
people to fill out the questionnaire at the same time. In order to randomise the selection 
process, the researcher selected every 3th, 5th or 10th person (depending on the flow of 
people) to overcome any selection bias. Also, the researcher ignored children in this selection 
process if they happened to be the 3
th
, 5
th
or 10
th
person, because children did not represent 
the desired sample features. However, age is not of major concern of this study because 
previous research (Dawson 1997) has suggested that people (especially women) were 
constantly lying about their age which had brought significant errors to the research. Data 
were collected at different times of the day (morning and afternoon) and during different days 

---

173
of the week (Tuesday through Saturday) to ensure a representative sample. The data-
gathering process lasted for two months and 984 responses were obtained. For the purpose of 
this study one assumption was made regarding the sample and population. It assumed that 
each respondent answered only one survey. The chances of the same respondent filling in the 
questionnaire more than once were further reduced by the fact that the data collection 
conducted in China for a short span of time, which reduced the chance of someone forgetting 
and filling out the survey instrument again.