Introduction to the three parts

The research is dealt with in three separate parts in this chapter, matching the main questions which were introduced in the last chapter

Part A: User Demographics

It was decided to use a multiple choice questionnaire of users at Supernet’s Internet cafés in order to survey their demographic profile.

Design of the survey

The questions were chosen to keep the questionnaire length less than 20 questions, so that respondents would not get too bored, possibly reducing the reliability of the results.  Some of the questions were chosen and phrased so that some comparisons could be made to other surveys,  such as Rainie and

Horrigan (2005) and Kolko, Wei and Spyridakis (2003).  Other questions were designed to test some of the preconceptions held about the user population and some hypotheses about possible correlations between demographics and use of Internet cafés.  For example, is there a correlation between mobile phone ownership and Internet café use? 

It was decided that the questionnaire would be self-administered, and therefore it was particularly important that it could be completed easily by the respondents without reference to anyone else. The questions were carefully phrased to try and ensure that they could be clearly and unambiguously under-stood.  Likewise, the multiple choice answers were chosen to try and ensure that all possible answers were covered.  Both questions and answers were refined to ensure that there was no bias towards any particular answer. 

In order for the results to be statistically significant, the sample size had to be considered. A Sample Size Calculator (Creative Research Systems, 2006) suggested a sample size of 96 would give a 95% confidence level in a 10 point confidence interval over a population of 20,000.  20,000 was a rough estimate of the potential Internet café user population within easy commuting distance of the Supernet locations.  An informal target of 100 completed questionnaires was set as a result.

A sample of the questionnaire is available in the Appendices.

Translation: Language and culture.

It was important that language was not a barrier to accurate collection of information.  On discussion with locally based colleagues, it was determined that all potential survey respondents would be very fluent in either English or Indonesian.  Although the local Acehnese language is predominant, the location is a large city and it was considered that local people there have a good command of Indonesian.

The next concern was to get an accurate translation of the survey completed.  An undergraduate of dual American and Indonesian citizenship was chosen for this task. She is completely fluent in English and Indonesian.  Another Indonesian friend was consulted  to re-translate back to English from the Indonesian version of the survey.  This way, the accuracy of the translation was checked.  Since this research work was being undertaken in the context of an unfamiliar culture, the content and expected reaction to each question were discussed in detail to ensure that the questions would be easily understood as intended and that the multiple choice answers proposed would make sense to most people. The questionnaire was also specifically checked for its cultural sensitivity.  It would not be helpful to offend respondents or have them avoid particular questions or answers for cultural reasons.

It would have been good practice to pilot the questionnaire before running the full survey, but time and logistical considerations did not permit this.

Administration of the survey

It was originally intended that a pilot survey would be conducted with about twenty users.  The early results might have indicated a need for some fine- tuning before the main survey was administered. In the event, logistics prevented this from happening, the questionnaire was administered without a pilot.  

Supernet Internet has two locations.  Piles of surveys in both English and Indonesian were left with the receptionist at both locations, with instructions to invite each customer to complete the survey, offering a choice of preferred language.

It was decided to provide an inducement to customers to complete the surveys.  It was felt that the survey would typically take about 20 minutes to fill out properly; respondents were given 30 minutes free Internet use for completing the surveys.  The receptionist kept a list of the names of the respondents, their survey number and confirmation that they had been paid.  This was designed to prevent users from filling out the form many times.  After the first few forms were completed, it was noted that many respondents were skipping some questions.  The receptionist was then instructed to briefly review the completed surveys to check that they had been completed before paying the respondents. The receptionists work in shifts and the manager phoned or visited each time a new receptionist came on duty to make sure they knew what the instructions were.   The names of respondents were not collected on the survey or provided to the author, so that the anonymity of the questionnaire was maintained. 

After a few days, the receptionists stopped handing out surveys, citing a range of excuses.  Consideration was given to paying them a small incentive to get

the surveys completed, but it was felt that this would not encourage the collection of high quality responses, and that survey responses might be fabricated.  During a couple of visits, the author and colleagues personally distributed questionnaires to customers and increased the number of responses gained.

It was felt that overall, a reasonable sample of users was gained, representing those who visit both locations at all different times of the day and of the week.

It was originally planned that the survey would be administered to a group representing the general population as well as to users of the Internet cafe, so that the demographic characteristics of the user population could be contrasted with those of non-users.  This exercise proved to be logistically too difficult, and was abandoned.  For this reason, the first question on the survey became redundant.  Although it was not removed from the survey, its results were not analyzed.

Problems with the survey

There were some problems with the responses to the survey.  Some respondents did not take care in filling out their survey.  It is believed that the demographics questions were probably answered more accurately because:

a.) They appeared earlier in the survey, before respondents got bored.

b.) They were easy-to-answer factual questions rather than opinions that they

had to think about.

Most of the problems occurred with the questions which concerned how the

Internet is used.  These problems are dealt with in more detail below.

Collation of results

Since the demographic questions on the survey were multiple choice, the answers did not have to be translated. Indeed, that is one reason why the questionnaire was multiple choice in design.

Answers were collated on a spreadsheet.  Since not every question was answered by every respondent, "blank" was adopted as a possible answer, so

that the total number of responses to each question was the same as the total

number of completed surveys collected.

These results were then graphed.  These are presented in Chapter 5.

References:

CREATIVE RESEARCH SYSTEMS (2006) Sample Size Calculator. [online] Last accessed on 7/21/2006 at URL: http://www.surveysystem.com/sscalc.htm

KOLKO, B.E., WEI, C.Y. AND SPYRIDAKIS, J.H (2003) Internet Use in Uzbekistan: Developing a Methodology for Tracking Information Technology Im-

plementation Success Information Technology and International Development 1 (2), 1-19

RAINIE, L. and HORRIGAN, J. (2005) A Decade Of Adoption: How The Internet Has Woven Itself Into American Life. [online] Last accessed on 7/23/2006 at URL: http://www.pewinternet.org/PPF/r/148/report_display.asp