Maximum Difference Scaling (MaxDiff Scaling) is a research methodology that results in interval scale measurements that are based on comparative judgements. With MaxDiff, survey respondents are shown a set of possible items and are asked to indicate what the best and worst items are (or most and least important, most and least appealing, most and least desirable, etc.) There are many reasons to use MaxDiff in your research methodology, but we are just skimming the surface here today. Keep reading to learn more about MaxDiff, why you should use it, how it works and how to analyze it! And if you're interested in learning more, sign up for our free 30-minute webinar on October 28, 2014 about Everything You Need to Know About MaxDiff.
As mentioned before, MaxDiff Scaling, also known as Best/Worst Scaling, is a question type in which respondents select the items that are most and least preferred. According to the creator, MaxDiff assumes that the respondent will evaluate all possible items and choose the pair that reflects the largest difference in preference or importance. MaxDiff is an effective way to capture the preference of importance of scores for multiple items. It even works well for a large set of items, resulting in very little survey fatigue. MaxDiff has a lot in common with Conjoint Analysis, but it is more applicable to a wider range of research scenarios.
How does MaxDiff work? MaxDiff is an extremely simple question type to analyze and respond to. Respondents are asked to review a set of attributes, generally 3 to 6 at a time (but you could do many more). They are asked to give answers to two extremes: most and least preferred. They may be asked to repeat the exercise multiple times, jumbling up the order of answer options or adding and subtracting attributes. These questions are represented by a simple thumbs up or thumbs down.
The output presented by MaxDiff is called Share of Preference. All preference scores for each attribute must add up to 100%. Survey Analytics uses Logit Modeling for its data output. Logit Modeling calculates utilities for each alternative and presents the preference scores for overall and then breaks down to the most and least preferred scores. Some pros of this model is that it is fast, scaled utilities can be compared, and "share of utility" can be calculated for population and for segments. Some cons include that it is a bit more complex than simple counts.
Do you think you can get the same results for MaxDiff with Rating Scales? Wrong. For example, with a Likert scale, every attribute seems important to the respondent. With very little differentiation, it becomes vulnerable to cultural and personal biases, and the analysis can be hard to quantify. On the other hand, MaxDiff eliminates virtually all bias. Because honest differentiation occurs, it is easier to take and analyze.
So, are you ready to implement MaxDiff Scaling into your research efforts yet? If you're not convinced, or just want to learn more, check out our upcoming webinar on October 28, 2014 that will take you through Everything You Need to Know About MaxDiff in just 30 minutes.
Did you enjoy this post? Sign up for our weekly blog digest.