choice-metrics.com

Hi,

I wish to create a homogeneous pivot design (single design) for the entire sample. The design has 3 alternatives including one reference alternative.

I wish to segment the sample population on the basis of the following data: (The travel time is for the car drivers)

Travel Time (minutes) Percentage
0 - 15 19.05
15 - 30 16.28
30 - 60 20.67
60 - 90 28.92
90 - 120 13.43
120 + 1.64

Now, I am confused between two scenarios:

1. I consider 6 segments, each representing a row in the table. (Assuming that each row/segment of respondents has different reference alternative)

2. Creating three segments: Short (0-30); Medium (30-60) and Long (60 - )


The reason I am confused is: Will having more number of segments improve the design (I mean maximizing the fisher matrix... If my understanding is correct)?


Thanks
Neeraj

If you provide 6 segments, Ngene will be able to optimise over all 6 segments, while if you only provide 3 averaged segments, it will only optimise over these three segments. So providing more information (using 6 segments) should theoretically yield a better design. In practice there is perhaps not so much difference.