Once the different variables in your dataset have been categorized, you're ready to start analyzing the data. Different levels of analysis are possible from very simple to very complex. The more complex the analysis, the more likely you're going to get rich insights about the marketplace, respondents or your competitors. In this lesson, we'll discuss different analysis techniques including descriptive analysis, inferential analysis, association analysis and causal analysis. We'll then discuss descriptive analysis using nominal data and show how you can visually summarize dispersion of nominal data. Once you have categorized the different variables of interest for research, you are ready to start analyzing the data. Obviously, different labels of analysis are possible from very simple, to very complex. These analysis could also vary in terms of number, of quality of insights that you can obtain from these analysis in this two features or link. The more complex the analysis is, the more likely you're going to get richer insights about the market place, about the respondents and about your competitors. And so what we'll do in the rest of this module will be to present various analysis techniques with an increasing level of complexity. In so doing we'll try to discuss the marketing applications of each technique. The analysis techniques we'll discuss for the rest of this module include descriptive analysis, inferential analysis, association analysis, causal analysis. Descriptive analysis has to do with describing the data. Getting why is essential to analysis and why is desperation of these answers. Inferential analysis is concern with how different could the results in my survey be from the overall market or the overall population. Association analysis has to do with understanding the relationship between two or more variables. One example we'll cover here in the association is association between where someone lives and different types of brand preferences. And we'll wrap up this module, with an introduction to casual analysis, focusing on example, discussing the impact of advertising on sales. Descriptive analysis relies on, running the descriptive statistics for the variables. These descriptive statistics are useful to obtain another review of the data. Each variable can be describe on at least two dimensions. One is a dimension of central tendency which is basically the mean value of stronger effect in the data. The other is dispersion which relates to the concept of how different or how much variation there is in data from one answer to another. If my questionnaire respondents provide me with a sense of their preference for brand A over brand B, can I get a sense of how different those answers are? Depending on the scale using the questionnaires and he sees using the descriptive statistics variables vary. In other words depending on the scale of the variable, the central tendency and dispersion measures will vary. Let's begin by discussing nominal data specifically the measure of central tendency for nominal data. Again, nominal data are used to describe or categorize entities of customers or actions. So an example could be which type of TV program do you watch the most? Here, the question is not perennial ranking of the data. It's just what TV program do you like the most? It could be news, it could be reality TV, it could be sports it could be a TV series and so on. So when the results from a media survey may get something like this table, which provides you with different answers for the four TV categories, news, reality TV, sports and TV series. You can see that the news get 27 answers. That means that among respondents surveyed, 27 people said that news is a TV program they watch the most. Reality TV is 13, sports is 94 and TV series 66. In this data, the measure of central tendency being the mode, informs us that sports is the mode of the data because it got the highest number of responses. Measuring dispersion for nominal data allows you to measure the sense of different frequencies between the data. So 14% here is equal to 27 divided by 27 plus 13 plus 94 plus 66 for a total of 200 responses. So, it tells you that fully present of the respondents chose news as the primary type of TV program that they watch. You can also provide or summarize dispersion of nominal data visually. So here is an example. The top graph present data that you've already seen in the table before. And that gives you sense of what is a mode. What is the most frequent answer? And as you can see which is consistent with the table, sport is the most constant answer. Now how is that informative about dispersion? Well, in the next graph, lets say that the survey provided you with the following answers where 50 people they said that news were their favorite types of TV program. 50 people decided that reality TV is their favorite type of TV program. 50 people decided that sports is their favorite type of TV program and so on. And so as you can see here, even though the question is the same, the answers could be different. Another question that you might want to ask yourself is, which of these two graphs informs or depicts more variation? In this case, the graph on the bottom is the one that gives you the sense of more variation. Why? Because each of the action is very likely. That means that everyone you survey is very different in their TV preferences. On the contrary, the graph on the top indicates that one answer gets more answers. And so in that sense, that tells you that the response is quite homogeneous with respect to their favorite type of TV program.
