How Does Conjoint Analysis Work?

So, how does conjoint analysis work? In conjoint analysis, we do not asked consumers to directly tell us the importance of attributes or the value they receive from each level of the attribute. For instance, we don’t go to the consumer and say, “How important is memory to you?” or “How much do you value 4 gigabytes of memory relative to 2 gigabytes of memory?” Instead in conjoint analysis, we ask consumers to perform a realistic task: that of providing overall evaluations of products and then we use those evaluations to infer the consumer’s value system.

For instance, we will show a consumer a computer as a whole and ask them to evaluate the computer. From those evaluations, we would infer the importance associated with the underlying attributes. In providing this evaluation, the respondents must simultaneously consider both the good and the bad characteristics of the product in making the judgment. That is we must make tradeoffs and then they gave us the evaluation of the product. The alternative to conjoint is of course to ask the consumer directly to tell us how important is the attribute and how much they value the level of the attribute.

Now, it has been found that there are several concerns when we ask consumers directly. One concern is that consumers often tend to rate all attributes as being important. So they say, “Memory is important to me, hard disk is important to me and the price is important to me.” And that’s a problem because they are not really making tradeoffs between attributes.

Another concern is that asking consumers directly is an unrealistic task. This is not how consumers go about making decisions about the product. They don’t look at one attribute at a time and think about how important that attribute is. Instead, they are used to evaluating products in a holistic way.

The third concern is that sometimes consumers are unwilling to tell us their truthful evaluation of underlying attributes and here is an example of that last concern. A study was done at a business school where the goal was to understand how MBA students think about the attributes of job offers. So, if you think about job offers, the job could have attribute or future such as the salary associated with it, the region of the country in which the job is located, the functional area in which the job is; if it’s a marketing job or an accounting job or a finance job, how much travel does the job involved, how much opportunity there is to advance in the job and so on and so forth. And the researchers wanted to understand the preferences of the students for these different features of jobs. They divided the students into two groups randomly. The first group was asked to directly tell the researchers how important the attributes were and they were asked to rank order eight features of the jobs. The second group was given descriptions of job offers. Each job was described in all the features of the attributes and then the researchers inferred from that what must be the importance associated with the features of the jobs.

So, let’s think about the difference in the results between these two groups in terms of the importance of features and let’s focus on one particular feature which is the salary. Interestingly, the researchers found in this case that the group that was asked to directly tell the researchers the importance of salary said that salary is sixth most important; it’s not really that important. The other group which was not asked directly; when they inferred the importance of salary, we in fact found that salary was the most important feature. And so, the second group was really doing a conjoint task and we find that when we inferred the importance of certain features, those are actually quite different from the directly stated importance of that particular feature.