# Statistics are you friend… really

I’m now at the part of my experiment where I’m analyzing data. Which is the “fun” part. Especially because analyzing data always seems to mean learning new statistics.

I took a stats class during my undergrad, but that was about 10 years ago now! (Wow time flies.) And, to be honest, even if I did remember, the stats I learned during that course would not be helping me all that much. Those courses are a good intro, but that’s all they give you. Of course, you do need a firm grasp on mean and standard deviation and variance and how to compare and understand results.

But experiments will produce results that are more complicated. You can’t just compute a few means and variance and produce an analysis on that. Instead, you need more complicated tests like ANOVA, or confusion matrices or chi-square tests.

However, don’t let all those terms scare you. First of all, it’s expected that you won’t always know what test to use and when. That’s part of why you’re still the student. And, don’t be surprised when your supervisor doesn’t know either. Often, as you approach your analysis in a different way, or collect a different type of data, you’ll have to modify how you analyze and find the appropriate test. If you, or your supervisor don’t know what that is, then the best thing to do, is to find someone who’s an expert in the field. Trust me, searching out the right test on the internet is surprisingly hard. There are so many tests out there, and the information comes across mixed and confusing. An expert can help you narrow down the field to the appropriate test, or a few candidates.

Second, as much as the internet is a mass of confusion when it comes to finding what test to use, it’s a gold mine when it comes to understanding how a test works. Once you know what test you want to use, search out some information on it so you understand what it’s doing. What are the parameters of the test, and what should your values be? What do the results mean? How do you compare the results to make conclusions?

Once you know what you’re doing, make sure you have the right tools for the job. If the statistic is complicated, you will probably need to use some specific statistical software, like SPSS, SAS or R. As a student, I recommend R, since it’s free. The other two are probably accessible through computer labs on campus. However, if it’s a simple test, like a T-Test, calculating means, or standard deviations, it can usually be done using programs like Excel.

So while learning a new statistical test can be frustrating and an exercise in what feels like futility, at the end of the day when things start to come together it’ll feel somewhat more worth while. And you’ll be that little bit more prepared for the next test. Also, in the future, if you keep good notes about what you did (the paper you write is a good place for this), you can just repeat previous methods.