MaxDiff analysis Vs standard rating scale questions
Researchers have found out that rating questions are susceptible to user scale bias, scale meaning bias, and it lacks discrimination. In addition to that, ranking questions have limitations of their own like being over bias, difficult for respondents to evaluate, limitation in testing number of items, ordinal data obtained limits the analysis and it does not allow ties.
Constant sum questions also have the same limitations and loopholes. Furthermore, researchers have seen that when presented with constant sum questions, respondents try to make the task of evaluating all the items easier by engaging in the response strategies.
Having understood the problems and limitations of different survey question types; especially with rating scale questions, researchers prefer to use the Maximum Difference Scaling or MaxDiff survey as we call it. We can call MaxDiff analysis as a trade-off analysis technique allowing researchers to conduct pairwise multiple comparisons.
By effectively using the MaxDiff question, researchers could ask the respondents to select the most and least preferred or important points from the given list of answer options they are interested to test for greatest difference amongst items.
- If we compare the MaxDiff and standard rating scale questions, then the MaxDiff question will show greater discrimination amongst items and between responses received for the items.
- In a typical MaxDiff analysis example you will notice how easy it is to apply a robust trade-off technique compared to other rating scale questions.
- Compared to the rating scale questions, the resulting item score is easy to interpret in the MaxDiff survey as it is placed on a common scale of ‘0’ to ‘100’ points and summed up to 100.
- Being simple to understand and easy to reply even for adults and kids, a MaxDiff survey yields more reliable data.
- In a MaxDiff survey, you directly ask respondents to make a choice and not to express their strength or preference using some numeric scale. Hence, there is no scope for scale use bias.