Similarities and Differences between Conjoint Analysis and Bracket™: When to Use which Method
When measuring consumer preferences, rating scales of importance rarely produce the sort of clear differentiation desired. Choice techniques are better, but which types are appropriate for which situations? Two common techniques are conjoint analysis and prioritization tools. This paper will outline the appropriate uses for each.
Conjoint analysis is widely used to find the optimal configuration of products, services, messages and packages. In Conjoint, respondents are not asked about each attribute or product feature one at a time. Instead, respondents are presented with a combination of features making up an entire product and asked to make a choice between two products with an additional “none” option. Since the products are essentially bundles of features, the conjoint choices are similar to choices consumers make every day. But is conjoint always the best way to measure consumer preference? Of course not.
First of all, some problems are too small (say a product with only one or two features) to warrant a conjoint and others are too large to make it practical. While mathematically conjoint can handle a great many features, from a practical standpoint it isn’t always a good idea. Our research has shown that as the number of features grows respondents focus more and more on the most basic of features…often price.
Second, not every problem demands the ability to optimize every feature. In some cases all you want to do is determine which features are most important.
In situations where the list of features is long or where the goal is to simply determine which features are important techniques like Max-Diff or our proprietary Bracket™ are quite effective. Bracket™ allows you to start with a very long list of features (50 or more) and effectively find the ones that will drive preference. It's similar to Max-Diff (see white paper on Max-Diff), but it uses a tournament structure to successively eliminate the "losing features" early on in the testing process. This makes the process more engaging and faster for the respondent...requiring about half as many choices as Max-Diff.
Examples of Each from the Real Estate World
To better illustrate the differences between conjoint and Bracket™, TRC conducted two studies related to preferences of home buyers. The quantitative study examined factors consumers consider when buying a home. The study employed both techniques, conjoint analysis and TRC’s proprietary Bracket™, each was used to gather data about different groups of factors (or features) that play a role during the home buying process. The study surveyed recent and soon-to-be homebuyers.
For the conjoint analysis study we focused on the five features most often cited in real estate listings. Each house feature involved several levels that respondents could choose from. Conjoint analysis, specifically Discrete Choice, was used because the technique is able to measure both the importance of each of the 5 features as well as the trade-offs respondents are willing to make. The study was able to identify whether homebuyers were willing to give up a bedroom to get the right price or if they were willing to pay more in order to obtain the desired number of bedrooms and bathrooms.
The survey framed the conjoint exercise by asking homebuyers to imagine they were shopping for a home and to assume it is locted in their ideal location. The respondents were shown 2 home listings side by side, plus an "I wouldn't choose any of these" option.
The study found that the top three most important factors when shopping for a new house are number of bedrooms, price and house condition.
Effective Use of Conjoint Simulator
TRC also used the conjoint simulator to reveal that, overall, homebuyers are less interested in a "gut job" compared to "move-in-ready", which is not a surprise.
Homebuyers' preference for a "gut job" plummets as their income level decreases. However, at the $150,000 home price point, share of preference dropped more dramatically from "move-in-ready/some work required" to "gut job" compared to higher price points. It's likely because those shopping at lower price points probably have less disposable income available for major repairs.
When the conjoint simulator was set up at $150,000 and $300,000 home price points and house condition by those with an income of less than $75K vs. those making $75K or more, the tool revealed that preference for a “gut job” is much lower among those making less money (25% vs. 47%).
This illustrates how a conjoint simulator enables further analysis and can uncover findings not immediately apparent.
The Bracket™ study focused on 13 other features, such as a lot size and school system; that are also considered and play a role in the house-buying decision. With this type of exercise we are focused on the feature overall rather than the individual feature levels. A respondent who indicates that "School System" is important is likely to be indicating that they prefer a good school system, but we cannot be sure of this or in fact how good a school system they desire. For each respondent, the tool randomly asssigned the benefits being evaluated into pairs. Participants then chose the preferred attribute from each pair, that attribute moved on to the next round. This process continued until only one 'ultimate winner' remained.
By combining the choices made by a group of participants the tool was able to not only rank the attributes but identify how much more important each attribute is than the next ranked one. The 13 features were: proximity to work, proximity to family, school system, size of lot, fenced-in yard, dedicated parking (i.e. garage, drive-way, parking spot,) basement storage, wood flooring, open floor plan, updated kitchen, updated bathroom, master bedroom and pool.The study revealed that size of lot, school system, updated kitchen and dedicated parking are given high consideration when buying a home. Conversely updated bathroom(s), proximity to family, open floor plan and wood flooring were given less consideration. This is surprising given how televised house hunting and remodeling shows place a high emphasis on open floor plans and hardwood flooring.
Application of Results
So while the five features studied in the conjoint are the most often referred to in house hunting, a realtor armed with the Bracket™ results would also know to talk about things like lot size, school system, kitchens and parking. However, one would still need to include them in a conjoint analysis to find out how good the school system needs to be in order for the attribute to be a make-or-break factor in the house-buying decision.
The Key Is the Objective
As you can see both techniques provide powerful and useful information. The right choice depends on the application.
A realtor who is trying to decide which houses to show their clients would do just fine with only the Bracket™ results.The realtor already knows specifics on what a buyer desires (in a good school district for example). Since the realtor is unlikely to find a home that meets all of the buyer’s criteria they can instead decide which criteria are most important and focus on them.Also worth noting that the Bracket™ data can be cut by segments so the realtor could look at results for people similar to their customer (age, income, family size, etc.).
But what if the realtor was selling the house and wanted to know how to price it? Knowing which features are important is not enough. In that case, they would need to know what the market might pay for a similar house. Only the conjoint results, expressed through the simulator, can help with a broad problem like that.
It is worth noting that in some cases it might make sense to combine the two. As noted earlier it is impractical to include a large number of features in a conjoint (for example, these two studies considered 18 different features of a home), but you certainly want to make sure you include the important ones. Do a Bracket™ study first to tell you which features are important drivers and then do a conjoint study which includes all of the features found to be important (along with levels for each). This allows you to keep the choices relatively simple while focusing on the features what will drive decisions.
As with any market research effort, careful planning and an understanding of the advantages and limitations of the available tools are critical to success.