Last year, I posted an infographic titled “Hypothesis Tests in One Picture”. But formulating a hypothesis statement can be tricky–and you need one to even start choosing tests. That’s why I like this simplified graphic, posted on Kenyon College’s website.
Rather than starting with the type of data, it starts with a hypothesis statement. Do you think there’s a relationship between the data? Or perhaps there’s a difference in variances, distributions, or some other statistic. The starting point in analyzing data is often forming a hypothesis statement. And figuring that out is vital to getting the right answer from your data. With a hypotheses statement, you’ll be in a good place to choose the right test or procedure to analyze your data.
A hypothesis statement is simply a short statement, a reworded question if you will; A question you want the answer to. For example, you might wonder if there’s a difference in the different sections of your data (for example, higher averages), or you might wonder if there’s a trend over time.
The picture below (based heavily on the aforementioned graphic) sums up some basic hypotheses, and leads to the tests and procedures that follow those statements.
More info
What is Numeric Data?
What is Categorical Data?
One sample T Test
Chi Square Test
ANOVA
GLM
References
What Statistic Should I Use
Credit: Data Science Central By: Stephanie Glen
