Have an idea?

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

What do you think about this design test?

Dear all,

I would like to ask your opinion on the design test.
This is my first quantitative survey and, of course, the first time I use MaxxDiff. So I have many doubts.

You can find the results of the design test below.
I think everything is ok.
I tested the design with different number of items per set.
I got the lowest std dev  with 5 items per set.
If I use 4 items per set, I will need more tasks and this is not what I want.

Do you think this design is good enough?

Thanks in advance.

Kind regards.

DESIGN TEST

Number of Items (Attributes): 9
Number of Items per Set: 5
Number of Sets per Respondent: 9
Number of Versions: 300

 
One Way Frequencies:
 Item  Times Used
-----------------

    1    1500
    2    1500
    3    1500
    4    1500
    5    1500
    6    1500
    7    1500
    8    1500
    9    1500

Mean = 1500
Std Dev. = 0

 
Two Way Frequencies:

 Item|       1       2       3       4       5       6       7       8       9

-----+------------------------------------------------------------------------

    1|    1500     750     750     750     750     750     750     750     750
    2|     750    1500     750     750     750     750     750     750     750
    3|     750     750    1500     750     750     750     751     749     750
    4|     750     750     750    1500     750     750     750     750     750
    5|     750     750     750     750    1500     750     750     750     750
    6|     750     750     750     750     750    1500     750     750     750
    7|     750     750     751     750     750     750    1500     750     749
    8|     750     750     749     750     750     750     750    1500     751
    9|     750     750     750     750     750     750     749     751    1500

 
Off Diagonal Elements (not adjusted for prohibitions)
Mean = 750
Std Dev. = 0.33333

 

Positional Frequencies:

          | Pos.       1       2       3       4       5

----------+---------------------------------------------

Item     1|          300     300     300     300     300
              2|          300     300     300     300     300
              3|          300     300     300     301     299
              4|          300     300     300     300     300
              5|          300     300     300     300     300
              6|          300     300     300     300     300
              7|          300     300     300     300     300
              8|          300     300     300     300     300
              9|          300     300     300     299     301
 
Mean = 300
Std Dev. = 0.29814

Note: Test design tests the current design on file.  It does not reflect changes made to the MaxDiff settings dialog since the design was generated or imported.
asked 3 days ago by Mikael

1 Answer

0 votes
Mikael,

The design test looks great.  Your one-way and two-way balance and positional balance are near-perfect.  

MaxDiff is much more forgiving than conjoint analysis is, and it often works well even with poor designs.  But your design is excellent, so you're in great shape.
answered 3 days ago by Keith Chrzan Platinum Sawtooth Software, Inc. (78,825 points)
Dear Chrzan,

Thank you very much for your comment.

Regards
...