What we're going to be looking at in this course are the common tools that are used to analyze survey data. Now surveys get used for a wide range of applications within marketing. It may be to understand consumers political preferences. It might be to understand your brand preferences. It may be used in the design of new products. It may be used to figure out what are the right attributes to be focusing on in the marketing communications. Well think about the last time that you received a survey to fill out. Might have been ten, 20 questions. Other surveys might be 50 to 100 questions. Surveys can be very long and for each respondent that might be 100 individual items that they're responding to. Well as a marketer what we're trying to do are derive insights from those surveys. And I don't really care about how you responded to an individual item. What I care about is what's driving you, what are your beliefs. And the idea is that the individual items on a survey are manifestations of those underlying beliefs. So what we're going to be doing throughout this course is first looking at a tool called factor analysis that's intended to allow us to go from a large number of survey items, narrow that down, retain as much information as possible to identify underlying preferences, underlying beliefs that consumers have. No once we've done that, then we can go about forming market segments using cluster analysis. We can also look to identify individuals that belong to different segments using discriminate analysis. And lastly we're going to look at perceptual mapping as a means of understanding how our brand is seen relative to other brands. So to start out we're going to look at how we identify those underlying constructs using factor analysis. All right, so let me give you an example to start off. Suppose that we're interested in understanding consumer preferences for local retailers verses large national chains. And in this case we've got five survey items that were included. First asking about whether or not respondents agreed with the statement that local retailers have more variety compared to retail chains. Second question, ask whether or not you agree with the statement that the associates at retail chains tend to be less knowledgeable than the associates at local businesses. And the last three questions, questions three through five, get into the courtesy and the level of personal attention that you might expect when you patronize local retailers versus when you patronize national chains. Now if we collected these five responses to these five questions, you might have it from a sample of respondents. In this case we have 15 responses. What we might begin to do is look for patterns among the responses. That is, for when people respond above average to question one, how do they tend to respond to question two? When people respond above average for question three do they tend to respond above or below average for questions four or five. And so the technique that we might default to using is correlation analysis. What correlation analysis let's us look at is is there a pairwise linear relationship? That is do the two items, when one goes up does the other tend to go up? When one goes down does the other tend to go down? That'd be indicative of a positive relationship. Negative relationship would be when one goes up the other tends to go down and vice versa. And if we're dealing with a small number of survey items such as the case here that might be all right. So what we could look at first is the correlation matrix. And we can see along the diagonal, we have ones, that's to be expected because we are taking the correlation between, let say item one and itself, so that's why we're getting the ones along the diagonal. Then we look below the diagonal 0.61, fairly strong positive relationship between items one and two. If we look for other high or very low values of correlation, we might see question three is correlated with question four. Question three is also correlated with question five, and questions four and five are also correlated with each other. Now, in this case we might say, let's identify those items that tend to move together. And it looks like items three, four and five tend to move together and Items one and two tend to move together. Now in this case we happen to get lucky with the correlation matrix. The items that are correlated with each other are directly adjacent to each other. We're dealing with a small enough number of items that we can just stare at the correlation matrix and see which items tend to move together. But what about a lengthier survey? What about a survey that's several pages long if we're dealing with 20, 50, 100 items? Staring at that matrix is going to be very difficult to identify the patterns that exist. All right, so that's where factor analysis is going to come into play for us. It's going to allow us to draw these boxes around items. That tend to move together without us having to do that work. So what factor analysis is going to take as an input is all of the survey responses. It doesn't matter if you have ten survey items, doesn't matter if you have 50, 100, 200 items factor analysis doesn't care about that. What it's going to do is take those individual items the responses from all of the individuals on those items, and identify which sets of items tend to move together. So think of this as correlation analysis on steroids. Well, let me give you a second example that's going to reveal a little bit more about why we need a tool like factor analysis and why we're not going to be able to restrict ourselves to just looking at correlation matrices, right? Suppose we're an auto manufacturer and we want to understand what drives of particular segment of consumers. So let's say we were looking at the young urban professionals or the yuppies. And how do you go about designing branding and targeting consumers with a message that's going to ultimately resonate with them. So one way we might go about trying to understand the consumer is to administer a survey. So one way we might go about trying to understand our consumers is to administer a survey. So let's take a look at the survey that we might administer. So this is a copy of the survey that was administered. Item one of this survey is a purchase intention question. How likely are you to buy this vehicle? One is I'm unlikely to buy it, all the way up to eight or nine being, I'm very likely to buy it. And then we have questions about people's perceptions of themselves. So for example, I'm in very good physical condition. I have more stylish clothes than most of my friends. Life is too short not to take some gambles. So we're tapping into some idea of how you view yourself, your adventure seeking activity. If we look at the next couple of questions, how concerned are about the environment? Asking about the ozone and pollution. Is society today generally fine? Do you tend to volunteer? We scroll down a little bit further, we see questions relating to finances. How family is not too heavily in debt? And scale one being on the low end, eight and nine being on the higher end. I pretty much spend for today, let tomorrow bring what it will. So perhaps I'm not too concerned about my financial well being. Interest rates are low enough to allow me to purchase what I want and tapping into that long term planning aspect question about self confidence and being seen as a leader. I'm usually the first among my friends to try new products. I work hard I play hard. I can do anything I set my mind to. Five years from now my income is going to be higher. So we've got questions about how do you see the environment, how do you see society, how do you see your finances, how do you see yourself. And in total we've got about 31 questions if you include that purchase intent question. And so based on these survey items what could we ultimately do with it? Well, if we could identify those people who are likely to buy a car or expressing interest in this car. And what are the perceptions of themselves, perceptions of the society, the perceptions of their finances are associated with people who are likely to buy this car, right? And so we might afford to say let's run one regression. let's take all of these survey responses as inputs or outcome variable, or y variable, that's can be the purchase intention. And conceptually that makes sense. That's what we're trying to do. We're trying to relate the individual survey items to the outcome of interest. The problem is some of the survey items are going to be highly correlated with each other. And we may run into problems of multicollinearity, if we were to run that large regression. The other problem that we might run into is supposed that we are able to run the regression. Well, what do we ultimately do with it? So, suppose that the government should restrict import or products from Japan is a significant driver of purchase intentions. How do we act on that? That's different from saying that somebody's who is likely to buy this car has a lot of patriotism. Saying that, we're going after consumers or a patriotic that's something that we can design a marketing campaign around. Saying that we're going after people who are against imports, not as clear.
