Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

TURF Analysis using a likert scale

I have a likert scale on 29 items in a survey. I wish to run a TURF analysis to understand reach. I was trying to do the same using the Maxdiffanalyzer.com.  Unfortunately, I'm not sure of how the data should be fed into the tool. Do I need to convert the (Very likely-somewhat likely) scale to 1 and the rest to 0.

What is the Reach method should I use? I use weighted by probability for MaxDiff scores, but I don't think we can use this for integer data.
If I use the First choice method, I'm unable to explain such low reach numbers.
asked Jun 22, 2021 by Karuna Tandon (300 points)

1 Answer

0 votes
Use the "General" data type as you create your MaxDiff Analyzer project.  This is a .CSV file (comma-separated data file) where respondent ID must be in the first column.  You should keep the ratings on the 5-point or the 10-point scale...however you collected the data.  (Don't convert to 1/0 unless it was pick-any data from a multi-select grid.)

I'd use the Threshold method of reach, where you can adjust what the threshold should be.  For example, if you think that either a 4 or 5 on a 5-point scale indicates reach, then you set the threshold at a "4".

Weighted by probability requires that the data are on the logit scale, coming from a MaxDiff experiment and if data estimation was individual-level MNL, such as HB MNL estimation.
answered Jun 22, 2021 by Bryan Orme Platinum Sawtooth Software, Inc. (196,215 points)