Conjoint Survey Case Study:
The data is from a case study on conjoint analysis presented by SAS:
http://support.sas.com/rnd/app/da/market/mraex.html
We look to demonstrate how MetaCause can be used to analyse conjoint survey responses.
In this study, seven respondents were presented with 18 choices of car tyre, each having different properties, and asked to rank them in order of preference.
Aim: The aim is to establish the influence of each of the following properties on consumer preference: |
 |
- Brand Name - RollsAhead, GoodThreads, TireMax
A good brand desirable for manufacturers, but how important is it to customers when considering a purchase?
- Installation Cost - £7.50 or no cost.
An important consideration for customers buying a tyre, is the indirect costs. But how important are they? Is selling a specialist tyre with extra fitting costs a good idea? Purchase Price - £45, £60, £75
Is cheapest always best? Is the most expensive tyre always best? We will study the importance of pricing and its place in the buying process.
- Thread Mileage - 40k, 60k and 80k Miles
Investment for the future? How important is last-ability in the tyre market? Are customers looking for long running tyres or are they concerned about getting a puncture before the tyre wears out?
Out of the 54 possible combinations, 18 were created using an orthogonal array to best study each of the effect of each property.
|
|
|
|
|
|
|
|
|
|
|
Product Features
|
Respondent Responses
|
| Brand |
Installation Cost |
Price |
Mileage |
1 |
2 |
3 |
4 |
5 |
6 |
7 |
| TireMax |
7.5 |
45 |
40000 |
17 |
17 |
7 |
14 |
13 |
15 |
14 |
| TireMax |
0 |
75 |
40000 |
18 |
15 |
17 |
17 |
17 |
18 |
17 |
| TireMax |
0 |
60 |
60000 |
15 |
10 |
9 |
9 |
9 |
8 |
1 |
| TireMax |
7.5 |
75 |
60000 |
16 |
11 |
15 |
12 |
14 |
16 |
6 |
| TireMax |
0 |
45 |
80000 |
13 |
3 |
1 |
1 |
2 |
3 |
9 |
| TireMax |
7.5 |
60 |
80000 |
14 |
4 |
8 |
4 |
5 |
9 |
8 |
| GoodThreads |
0 |
60 |
40000 |
11 |
12 |
13 |
15 |
16 |
12 |
15 |
| GoodThreads |
7.5 |
75 |
40000 |
12 |
14 |
18 |
18 |
18 |
17 |
18 |
| GoodThreads |
0 |
45 |
60000 |
9 |
5 |
4 |
7 |
4 |
4 |
3 |
| GoodThreads |
7.5 |
60 |
60000 |
10 |
7 |
10 |
10 |
11 |
6 |
2 |
| GoodThreads |
7.5 |
45 |
80000 |
7 |
2 |
2 |
2 |
3 |
2 |
10 |
| GoodThreads |
0 |
75 |
80000 |
8 |
6 |
12 |
5 |
6 |
7 |
11 |
| RollsAhead |
0 |
75 |
40000 |
5 |
16 |
6 |
13 |
12 |
14 |
13 |
| RollsAhead |
7.5 |
60 |
40000 |
6 |
13 |
11 |
16 |
15 |
13 |
16 |
| RollsAhead |
7.5 |
45 |
60000 |
3 |
8 |
5 |
8 |
7 |
5 |
4 |
| RollsAhead |
0 |
75 |
60000 |
4 |
9 |
14 |
11 |
10 |
11 |
5 |
| RollsAhead |
0 |
60 |
80000 |
1 |
1 |
3 |
3 |
1 |
1 |
7 |
| RollsAhead |
7.5 |
75 |
80000 |
2 |
18 |
16 |
6 |
8 |
10 |
12 |
The Analysis
MetaCause analyses the parameters presented to each visitor and learns from the behaviour of each of the individual visitors. It learns which parameters are important in influencing the reaction of the consumer, and how.
Other parameters! - These are often forgotten about! In every study there are hundreds of parameters that are not considered. Most have little or no effect (such as what a customer has for breakfast) and some have more of an effect (such as road quality, type of vehicle, and economic climate).
MetaCause's advanced pattern recognition keeps the effects caused by both other parameters in the study and external parameters, and can detect if a pattern repeatable outside of the analysis. This is automatically considered in the Importance Weighting in the reports, for every study.
The Results
The results generated by MetaCause are shown below:
- Click for full HTML results file -
 |
As we can see from the results, the 'LOW Thread Mileage' has the highest importance weighting at 60.6%. This is a very strong correlation and indicates that a low thread mileage is very undesirable to the customer (Factor to Avoid). Also note that 'HIGH Thread Mileage' is strongly desirable (Optimal Factor).
This marks Thread Mileage as the most important factor in the analysis, as there is big reward for getting it right(High is strongly desirable), and a big penalty for getting it wrong (Low is strongly undesirable).
A 'LOW Installation Cost' is found to be strongly desirable (an optimal factor) with an importance of 53.9%, however although the reverse is true, 'HIGH Installation Cost' is a factor to avoid, the importance is significantly smaller at 42.3%. |
This indicates that a Low installation cost is desirable, but the High installation cost is not significant enough to deter customers. The reverse can be seen for the purchase price. A 'HIGH Purchase Price' is very undesirable, but a Low price is not necessarily a strong advantage.
Brand name does not feature as a very significant factor in the analysis |
- Back -
|
