How to Perform Statistical Data Analysis
A popular maxim is that the data speak for themselves. Statisticians, however, know that the data rarely speak for themselves; you have to ask them the right questions. That's what statistical data analysis is about: asking a question, performing the appropriate analysis and examining the results.
Things You'll Need
- A computer
- A data analysis program (popular ones include SAS and SPSS; Excel, while more of a spreadsheet program, can be used for data analysis as well)
- Data set
- A statistics book or guide for reference
Data Analysis: From Preparation to Presentation
The first step of data analysis consists of tasks that need to be accomplished before the data are even analyzed. As complex as some statistical techniques are, the tasks you must accomplish before analysis are even more difficult. You must start by defining your research problem. What are you trying to find out? Then you must develop a research plan that includes drawing a sample and developing appropriate measures.
Having defined your research problem and collected data, you must ensure that those data are in analysis-ready condition. They must be entered into a data analysis or spreadsheet program. Then you should check the data for accuracy and do any transformations necessary to ensure the data can be analyzed using whatever techniques you will be using for your particular study.
Now that your data have been entered and prepared, get to know them better by running a set of descriptive statistics. These are simple summary measures. Suppose, for example, your data examine student performance on standardized math tests at five high schools. You might begin by running a set of descriptive statistics that describe characteristics of the students at each school, such as ethnicity, gender, grade level and socioeconomic status. Data analysis software such as SPSS and SAS, two of the better-known statistical analysis programs, can do this easily. You might also get each school's passing rate on the test and the average score.
Now that you have your descriptive statistics, giving you a better understanding of your data, you can begin running inferential statistics, using methods that relate to the research questions or hypotheses that form the basis of your study. Here's another example. Suppose two of the five high schools in your data had a special tutoring program designed to improve student performance in math. The obvious research question is whether participation in the program improved math achievement. Depending on the nature of your data, this question can be answered with a variety of inferential statistical techniques that vary in quality, from significance testing to analysis of variance to a multivariate linear regression model.
Once you've run the appropriate inferential statistics, you should examine the output from your statistical procedures and write up your results. Think of writing the report as an extension of your analysis; after all, you're presenting the results of your study in such a way that the intended audience will understand. The statistical output from your analysis may be quite lengthy, so you should select the most important results for summary tables and graphics that you use in your report.
Tips & Warnings
- Examine the output from your analysis closely, ensuring that your results are valid. Repeat procedures if necessary.Be sure to report results accurately in your presentation.