In recent years with the increasingly world-wide introduction of the Internet, the use of online questionnaires has increased dramatically. However in Thailand, there has been only very limited systematic research on web-based design in Thailand, including for Thai undergraduates who are the biggest group of Thai internet users. The particular characteristics of the Thai language (e.g. no capital letters, no break between words, Thai script etc.) present some interesting challenges for online Thai surveys. This experimental study investigated web-based survey design principles based on an English language background trial at a Thai university with individual interviews and focus groups with the use of think aloud and other research techniques. The findings of two types of web usability tests revealed that the scrolling web-based format was the most suitable for conducting survey s and that such surveys are most likely to attract higher response rates when endorsed by a trusted organization, when instructions are short, simple and specific, when closed and dichotomous questions provide sufficient answer options and when matrix and semantic differential questions are limited. Research also indicates that the font, Ms Sans Serif of size "-1" or 14 pixels in Thai, is the most appropriate for the Thai language, as is a simple progression bar, three-point rating scales and an artistically decorated survey form.
The approximate ideal length of an effective on-line survey is about 20 questions, taking about 5 - 10 minutes to complete. The short and potentially sensitive demographic questions are best obtained just before respondents complete the questionnaire. Thai undergraduates adequately understand check boxes, option or radio buttons, and drop-down menus; therefore a help section may only be necessary when the survey is more complex than a general survey.
The study also examined the most attractive invitation method, comparing pop-up windows, message banners and advertising marquees in a 22-day trial on a Thai university website where 3,848 survey forms were completed, representing 22.7 per cent of those who entered the survey web-site. The characteristics did not differ from the actual university web-users profile - 58% were female and 32.8% were university students. The most effective invitation method was a message box when users clicked on any link on the homepage since it is a new method with no restriction from the browser with an effective grasp on the attention of the users. The most significant reasons influencing participants' decisions about the questionnaire were the same factors effecting decisions to participate in surveys generally: topic of survey followed by the importance of the survey content. The third reason influencing users' decisions to participate in this survey was the invitation method. In future, the third factor may become the ben efit accruing to the respondent.