MaxDiff-Interpreting Results

To access the MaxDiff Analysis Reports, Go to:

Login » Surveys » Analytics » Choice Modelling » MaxDiff Analysis

MaxDiff analysis estimates the values (utilities) of each product from the given respondents choices. We use multinomial logit model which gives us the probability that a product will be chosen from a group of products. P(prodi)= exp(Ui) / (exp(U1)+…+exp(Un))

We have improved the MaxDiff reporting and it is much easier to understand the most prefered attribute based on the share of preference. For each attribute you will see the percentage of times one attribute is preferred over others by the respondents. You will also see the best and worst percentage for every attribute.

To filter the data, click on filter icon and select the data filter from the drop down. The analysis will be filtered according to the segment selected.

We support an Excel and SPSS datafiles. To export click on the icons. We provide 3 step detailed analysis. The first sheet gives an overview of share of preference, Best and Worst percentage for each attribute.

The second sheet provides the raw data with step wise selection made by the user.

• 0 - Attribute was displayed but not selected by the respondent
• 1 - Attribute was selected as Worst
• 2 - Attribute was selected as Best

1. Raw Utility (Best)

   U(b) = Ln(B(i) / 1 - B(i)))

2. Raw Utility (Worst)

   U(w) = -Ln(W(i) / 1 - W(i))

3. Raw Utility

   Raw Utility [U(i)] = Average(U(b) , U(w))

4. Scaled

   U(Scaled)(i) = U(i) - Average(U)

5. Transformed

   U(Transformed)(i) = EXP(U(Scaled)(i))

6. Preference Share

   Preference Share = U(Transformed)(i) / SUM(U(Transformed))