As M.J. Moroney suggests, statistical analysis is just as much a science as it is an art. It takes a careful examination of the results and understanding of the domain to provide meaningful and reliable results. However, these results are of little use if the appropriate steps aren’t taken to ensure that the data has been managed and prepared properly. Researchers can become eager to quickly start running statistical analysis while ignoring crucial first steps that can impact the results. This installment of CPG basics will focus on those important steps that are often forgotten.
Step 1. Visually Scan the Data
You’ve hunted down participants and took the hours to painstakingly enter all the data into your analysis software. Finally, the data has been entered. Let’s crunch some numbers. Stop! Resist the urge to jump right into data analysis and instead step back for a moment and take time to really look at your data. Scan the data left to right, up and down. Look for glaring irregularities, strange pattern of numbers, or a large amount of missing values. Often researchers waste hours looking at confusing statistical outputs only to realize a mistake that could have been easily caught with a visual scan.
Step 2. Run frequencies
Now that the data has been visually scanned it’s time to start some preliminary descriptive analysis. The frequencies function found in SPSS or the Statistical Package for the Social Sciences (also found in most statistical software programs) is a great way to get an understanding of how your data is behaving. It provides a summary of how often participants select a certain option/variable/item and displays whether there is any missing data for each of the variables. This serves as a great second check on your visual scan. Document any missing values or strange frequencies so that you can refer to it during your data analysis and interpretation.
Step 3. Graphically Display your Data
One more step before the statistical analysis begins, displaying the data in graphs or other illustrative forms. The type of graph that is most useful will vary with the purpose of the data. Generally, histograms, bar graphs, or line graphs are a useful visualization that depicts the distribution and trends in the data. This will give you a better understanding of your data and will help make sense in later interpretation of results.
To learn more about strategies and techniques we use at CPG during product development and research visit our “CPG Basics” blog posts.