Some Consumers have a positive attitude towards a product but when it comes to actual purchase, they purchase a substitute instead of that product. How one can analyze the two different data sets related to attitude and the actual behaviour?
Two three steps you can take:
1. Collect the data on Likert scale
2. Apply factor analysis to reduce the number of variables
3. Then you can apply t-test
4. You can also compare compare by using only mean scores
5. Other technique, you find fit.
u can find mean std devn then apply ISM nd Fuzzy Micmac analysis..in order to find out relations between the variables..
Attitudes are what people say in response to obtrusive questions in which the interest of the researcher is transparent. A multitude of response biases make attitude measures only measures of what people say - not predictors of what they will do / choose in any choice situation. Clearly when consumers do not buy what their obtrusively measure attitudes predict - then the attitude measures are NOT VALID. They have not measured the TRUE attitudes of the respondents.
If you measure (by observation not obtrusive questioning) what people choose and do not choose, in a variety of choice situations, you might be able to infer their TRUE attitudes. This is the method of the Natural Sciences.
I argue this viewpoint about the invalidity of attitude measurements in:
John P. Liefeld, Consumer Research in the Land of Oz, Marketing, 2003, pp 10-13.
And verify it in:
John P Liefeld, Consumer Knowledge and Use of Country of Origin Information at the Point Of Purchase, Journal of Consumer Behaviour, Dec. 2004, Vol. 4, Issue 2, pp 85-96.
and,
John P. Liefeld, How Surveys Overestimate the Likelihood of Consumer Confusion, The Trademark Reporter, Vol. 93, July-August 2003, No. 4, pp939-963.
We always observe irrationality in consumers' purchase behavior. And what they think and say may not be related to what they really will buy in actual buying scenario. I think Daniel Kahneman's research work is a good one in this topic.
But for statistical tools: the unfortunate answer is exactly what John has rightly put it. The important question is to understand what kind of research methodology and sampling framework should be used. Sometimes inferences by observations can be misleading too: if you conduct a research with small sample size - and that is the case for real market research work in reality - then it may not be generalizable - and if you conduct a research with large sample size, you are making it experimental which the true attitude may be distorted by artificiality.
There is a growing demand for researchers in the industry to conduct observation-based research with larger sample sizes in the market, hoping to obtain statistical significance and generalization.
I've found it useful to examine the attitude-intention link in this respect (having no data on actual behavior) and used Soderlund and Ohman's tripartite intention view for measurement of intention (on 7 point likert scales). Here is their article on the proposed tripartite intention.
Also, for the model, I used a (SEM)PLS approach.
Dear Vishal,
This is quite an interesting field you are researching in. I would start by identifying the variables that trigger the change in purchasing behaviour, like in-store promotions, discounts, or location related to similar products (pricing, features...). And then collect data on some form of Likert scale (do define the scales clearly). An event time line might be useful to determine how the different factors effected the decision making process. Then make use of the statistical tools mentioned by Mr. Kundu.
Then to make it more interesting you can divide the group also into male and female consumer purchase attitude distortion.
Which product gives the consumer the highest perception of bang for consumer's buck?
Regards,
Cornelius
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence