Questionnaire Design: getting the most out of your quantitative research in Japan

•	Attribute scales. Compared with other markets, a longer scale can yield better results for attribute ratings. So instead of offering respondents the choice of 1-5, a longer scale, such as 1-7 can encourage more variation around the centre point. Also, where a clear-cut reading is important, an even numbered scale (such as 1-6 or 1-10) forces the individual to choose a positive or negative response with respect to a fractional mid-point.
•	Normalizing data. To compare data across different countries, it's usually advisable to normalize for such culturally-driven scaling biases. Obviously, scaling biases may differ greatly according to the region of the world; a well known scaling phenomenon is the use of extremes in some developing markets with Latin cultural influence, such the Philippines and Brazil. Here respondents tend to rate at the ends of the scale, which is exactly the opposite to the typical respondent behavior seen in Japan! Hence, a common approach is to normalize by taking into account the average score in the population (how people tend to centre in a country) and the standard deviation (the extent that they use the full scale when rating).
•	Wording of questions. In terms of the actual wording of the questionnaire, it important to consult with someone who knows the Japanese market well. This is not only a question of arriving at an accurate translation, but gaining some native input as to how questions should be phrased, so as to elicit honest and accurate responses. Furthermore, there are a number of questions that cannot be asked easily in Japan - especially those questions that respondents might feel are 'too personal' and so may make them feel self-conscious and less open to sharing information in subsequent parts of the survey. It is also important to bear in mind that Japan does not have a similar 'social class' system or defined socio-demographic groups as used in the UK and other markets.
•	Questionnaire length. One pitfall that many researchers from overseas often overlook is the fact that a 20-minute questionnaire in English ends up being around 25-minutes in Japanese, due to the expansive demands of the written language. As for a maximum survey length, it really depends on the methodology and the topic. For a fairly interesting product or service, in which the respondent would likely have a depth of knowledge or an emotional involvement, a 30 minutes for an online survey or 25 minutes for a telephone survey is usually the maximum length recommended. However, for those topics that lack a degree of 'star power', a slightly reduced time is advised to maintain a high level of data quality.
•	Involvement. By extension to the above, surveys should ideally keep the respondents interested and involved throughout. This implies that the flow and questionnaire layout should be drafted with the respondents' experience in mind. Variety in the type and length of questions asked can help, whereas page after page of tables will quickly tire them out. This of course impacts data quality and tends to make the culturally driven centre-point scaling bias even more prevalent.