eLab 2.0

Design Your Own Movie Theater Using Conjoint Analysis

(Thanks to suggestions provided by the WELL Movie Reviews Conference.)

Software.  In order to run the conjoint analysis applet, you will need a browser capable of supporting version 1.1 of Java. 

Screen resolution. Your monitor must be set to a minimum resolution of 800x600, or else you will not have enough screen space to comfortably view the applet.

Imagine you are designing a new movie theater. Assume you have identified five attributes that are important to movie-goers when they select a theater - ticket price, line-of-sight, seat comfort, audio/visual equipment, and concessions. There are different values each of these attributes can take. For example, suppose that you are considering charging ticket prices of $10, $8, or $6, and are also considering large vs. small screens in your theater. It's pretty obvious to you that most people will prefer a $6 ticket to a $10 ticket, and that most people would prefer a big screen to a small screen. But, what you don't know is how much people will trade off on ticket price in order to see a movie on a big screen.

Shortly, you will use conjoint analysis to understand how you, yourself, make tradeoffs among five key attributes that describe a movie theater.  Conjoint analysis is a marketing research technique that helps us understand why consumers prefer certain products over other products.  Conjoint analysis assumes that a product (such as a movie theater)  can be expressed as a bundle of attributes.  There are certainly more than five attributes that can be used to describe movie theaters.  However, to keep things simple, we will only consider the following five attributes:

The Conjoint Analysis Task

Below you should see a list of 18 movie theaters that are described by various combinations of the above five attributes.  (If you do not see the list of 18 theaters, please verify you have the required configuration for the Java applet to work.) 

Please sort the following 18 movie theaters from "most preferred" to "least preferred" theater: 
 

1. Clicking on a row, and holding the mouse button,  will highlight that row in red.
2. Once a row is highlighted, you can drag it up and down in the list with your mouse.
3. Releasing the mouse button drops the row into a new position.

How to Interpret Your Results - Basics

After you click on the Calculate button, the tables and figures will be automatically filled in for you.  Here's how to interpret them. 
 

1) The first table you see shows your Relative Importance for each of the five attributes.  These percentages sum to 100%.  The greater the relative importance, the more an attribute influences your preference for movie theaters.  For example, if your relative importance for Ticket Price was 40, and your relative importance for Food was 10, this would imply that Ticket Price was four times as important as Food in determining how you ranked the 18 movie theaters.

2) The second table shows the Part-Worth for specific levels of each attribute.  These have been scaled to range from a low of 0 to a high of 1.  The greater the part-worth, the more positively a level of an attribute influences your preferences.  You can compare part-worths within levels of a given attribute.  For example, if your part-worth for a $6 ticket was .8, for a $8 ticket was .7, and for a $10 ticket was .2, you would conclude that you had a slight preference for a $6 ticket over a $8 ticket, but that your preference for either a $6 or an $8 ticket was considerably greater than for a $10 ticket.  Note:  you cannot compare part-worths across levels of two different attributes.  Thus, you cannot directly compare a part-worth of .8 for a $6 ticket with a part-worth of .8 for gourmet food and conclude that both are equally important.


3) Most people find the five plots of part-worths much easier to interpret than the table of part-worths.  You interpret the part-worths in the figures exactly the way you interpreted them in the table.

Advanced Interpretation

One question we like to answer with conjoint analysis is how much a particular attribute is worth, in terms of dollars.  You can get at this indirectly in conjoint analysis, by comparing differences in part-worths across Ticket Price and a second attribute.  Remember, we said you can't directly compare part-worths across levels of two different attributes.  However, you can compare differences in part-worths across levels of two different attributes.  Here's how it works. 

Look at the difference between your part-worth for a $6 ticket (say it is .8) and an $8 ticket (say it is .7).   In this example, the difference is .1.  Now, suppose your part-worth for gourmet snacks is .6 and your part-worth for plain old hot dogs and popcorn is .4, so that the difference is equal to .2.  This means that the step up from dogs and popcorn to gourmet snacks (.2 units) is the same as two times the difference between a $8 ticket and a $6 ticket (.1 unit).  In dollar terms, you'd be willing to pay twice the difference between $8 and $6 tickets (i.e., $4) in order to be able to purchase gourmet snacks instead of hot dogs and popcorn.  You can compare other attributes as well, to get an idea of how much a good sound system is worth to you, big seats with cup holders, etc. 

Instead of comparing a $6 ticket with an $8 ticket, you could also compare an $8 ticket with a $10 ticket.  Psychologically, the two dollar gaps between $6 and $8, and between $8 and and $10, are not necessarily equal, even though both are two dollar differences.  The part-worths show the psychological difference corresponding to the dollar difference.  In our hypothetical example, we said the part-worth for a $10 ticket was only .2.  Thus, the difference in part-worths between $8 and $10 tickets is .5 (.7 minus .2).  This means that the step up from dogs and popcorn to gourmet snacks (.2 units) is only 40% of the difference between an $8 ticket and a $10 ticket.  So, if we are comparing the two higher priced ticket options, gourmet snacks are only worth 40% of $2, or 80 cents;  if we are comparing the two lower priced ticket options, gourmet snacks are worth two times $2, or $4. 

In this hypothetical discussion, the part-worths for the three levels of ticket price are non-linear.  Identifying such non-linearities in price are an important objective in conjoint analysis.  We're always on the lookout for the point where the price becomes so high that consumer interest drops sharply. 

Consider the following for in-class discussion

a. Interpret your pattern of part worths, and your relative importances.  What does it tell you about your preferences for movie theaters? 

b.  How might a different number of levels of attributes have affected your results?  For example, using five price points instead of three. 

c. Can you identify any other attributes, other than the five used in this exercise, that are important considerations for you in selecting a movie theater? How might the inclusion of these additional attributes have affected your results? 

d. Assume we have part worths and relative importances for all class members.  How can this information be used for market segmentation? For product development?

e. How well do you think conjoint analysis would work for designing a movie, compared to designing a movie theater?

f. Please prepare a one page overhead showing the plots of your part worths and your relative importances. (You will need to print, and then cut/paste/reduce your output to fit). Bring this to class. I will call on class members for examples.