A Peek Behind the Curtain into the World of Data Analysis

One of the more mysterious steps in the market research process is the data analysis stage. Oftentimes, we use phrases such as multiple regression, factor analysis, cluster analysis and multidimensional scaling do describe data analysis, but what do all of those mean and how are they used? 

While those phrases may seem intimidating or complicated at first glance, they are simply used to describe the techniques data analysts utilize to interpret a raw data set. These raw data sets are usually large files filled with hundreds, if not, thousands of responses all displayed as numbers. In the market research world, these numbers usually correspond to certain, preselected responses to different questions. These files contain all survey questions asked of respondents, verbatim open-ended responses, and incomplete surveys. It's a data analyst's job to sort through these data sets using one or a combination of the following common methods: 

  • Multiple Regression - This method, in essence, answers the question of how one variable changes, when another is intentionally changed. For example, this method can answer how sales revenues changed based on placement of advertisements, advertising budget, etc. 
  • Factor Analysis - This method identifies how underlying variables relate to each other. In every data set, there are bound to be correlations between two or more variables. Factor analysis determines which variables have the strongest correlations. A market researcher can use factor analysis to find the best combination of factors that are attractive to customers. 
  • Cluster Analysis - This method separates data into separate, homogeneous groups based on alike traits. Common traits are based on demographics, but can be as specific as an analyst wants them to be. This method is useful to separate consumers into market segments. 
  • Multidimensional Scaling Arguably one of the more abstract methods of data analysis, multidimensional scaling is useful for comparing competing brands and products. For example, different types of air fresheners can be compared based on scent strength, scent type, and longevity. The competing products would be put onto a perceptual map, with the distance between these brands highlighting the dissimilarities.    

These are just some of the tools our data analysts use to interpret data from each and every study we conduct.