Likert Scale is not for Indonesian!

For those who have conducted surveys using the Likert scale in Indonesia, we might encounter the same issue. Many respondents tend to provide answers that are high on the scale, for example, 4 or 5 on a scale of 5. In fact, on a scale of 10, answers often range from 7, 8, 9, or 10. There are even questions that should ideally be answered with low numbers, such as 1 or 2, but respondents rarely do so. This phenomenon might also be observed in other countries, but it is quite common in Indonesia. In European or American countries, people are more open to giving low scores, and I have also analyzed surveys in Singapore where the responses were more varied.

This becomes an issue because the purpose of the Likert scale is to obtain a diverse distribution of responses from low to high. If all answers are concentrated at the high end, we cannot clearly ascertain the level of interest respondents have in the products we offer. For instance, if many respondents answer 4 or 5, can we conclude that they would purchase our product? It is possible that the product may not actually sell well in the market. This could mislead us into drawing incorrect conclusions. We might feel that our product is ready for launch based on the average responses supporting it; however, these responses often do not reflect reality.

To address this problem, here are three tips I typically use in surveys with the Likert scale in Indonesia:

1. Use Comparative Products: Ask respondents about our product, then compare it with competitor products. This way, we can observe significant differences between products. If there’s no clear difference, add a question about which product they rate higher.
2. Compare with Standards: Utilize norms or standards for respondents' answers in Indonesia. For example, if the average response is 4.78 out of 5, while it visually appears high, we need to determine whether it’s better or worse than the general average norm of Indonesians. This norm can also vary depending on the category or profile of the respondents, such as gender or age.
3. Analyze Top Box: Focus on the highest answers in the Likert scale. For a scale of 5, look only at those respondents who selected the highest score, which is 5. For a scale of 10, we can see what percentage gave a score of 10, or for larger Likert scales, we might also consider those who chose the top 2 boxes, which would be 9 or 10.

This is always a challenge, but we can still find ways to analyze the responses of these generally generous Indonesians with the Likert scale, even though the goal of obtaining a spread of answers isn’t fully achieved.

Is there another method? Yes, there is, which is normalization; we can discuss that in the next article.

Aristo Labare - December 2024