UAMS.EDU

One-size-fits-all statistical analysis? Hmmmm….

We see a great many protocols with statistical analysis sections that read as follows:

Data will be analyzed using SPSS version 20 (SPSS Inc., Chicago, IL).  Baseline characteristics will be described as means and standard deviations for continuous variables and percentages for discrete variables.  Continuous variables will be compared using Student’s t test and categorical variables will use Chi square test.

While this may be appropriate for some studies, please note that we’ll take a look to make sure that the data your study proposes to collect fits this data analysis language. We’ll check to see that you have actual continuous and discrete variables, and that you can justify the need to collect all the information you’re proposing to collect. Many times a protocol will describe the collection of a great deal of information that looks like it would be helpful to test for associations between baseline characteristics and outcomes, and then the data analysis section will make no mention of tests of association.  The IRB can’t approve your protocol without knowing specifically why you need all of your proposed data. So make sure the plans for data use and analysis are clearly and accurately described.