What Statistical Hypothesis Test Should I Use?

What hypothesis test should I use? So, you collected some data and now you want it to tell you something meaningful. Unfortunately, your last statistics class was years ago and you can't quite remember what to do with that data. You remember something about a null hypothesis and and alternative, but what's all this about testing? 

Sometimes it's easier just to give a problem to the Assistant. Especially when it comes to statistics.

Don't get me wrong, I love to analyze data and see what it means...but most of us don't analyze data all day, every day.  And in statistics, as in sports, if you don't use it, you lose it. If you haven't done an analysis in months it's not unreasonable to imagine you might need a little help. 

In that case, you might seek out the Assistant. Specifically, the Assistant menu in Minitab Statistical Software. The Assistant's always ready to guide you through a difficult statistical task if you're not quite sure what to do. 

The 2-Sample t-test

For example, suppose you want to compare two different materials for making backpacks–Cloth A and Cloth B–to determine which would make a more durable product. You sample materials from both suppliers and measure the mean amount of force needed to tear them.

If you're already up on your statistics, you know right away that you want to use a 2-sample t-test, which analyzes the difference between the means of your samples to determine whether that difference is statistically significant. You'll also know that the hypotheses of this two-tailed test would be:

  • Null hypothesis: H0: m1 - m2 = 0 (strengths of the material from both companies are equal)
  • Alternative hypothesis: H1: m1 - m2 ≠ 0 (strengths of the material from both companies are different)

And that if the test's p-value is less than your chosen significance level, you should reject the null hypothesis.

Maybe you're the type of person who remembers all of that stuff, even if you haven't done a t-test in years. If so, good for you–but I could stand to get a little help. Let's see what the assistant can do for me. 

The Assistant and the Hypothesis Test

I'll start by pulling up Assistant > Hypothesis Tests... in Minitab. Up comes this dialog box: 

assistance with statistical hypothesis tests

assistance with 2-sample t-test

Well, I know I have two samples that I want to compare.  But I can't remember if I need a paired t-test, a % defective, or what.  So I'll click "Help me choose."  Now the Assistant gives me an easy-to-follow decision tree that leads me to the 2-sample t-test. 

I can follow the tree straight to its conclusion, as shown on the right.  

A Guided Path to the Right Hypothesis Test

But if I can't remember enough specifics to follow the decision tree from start to finish from the amount of information shown on the right, the Assistant will actually guide me through the process step-by-step so I arrive safely at the right hypothesis test to use.

In this scenario, The Assistant asks one question, then I choose the right option for my situation and proceed to the next question. 

Particularly helpful is the fact that if I forget, for example, the difference between Continuous and Attribute data, all I have to do is click on a button and I'll get an explanation and an example for both. 

In this situation, the questions I need to answer are: 

  • Do you have continuous data or attribute data? (Answer: Continuous)
  • What are you comparing? (Answer: Two means)
  • Are you measuring different sets of items or the same set of items? (Answer: Different sets)

Now I know what test I need to use to compare the two means.  But I'm not sure I know how to do that test.  

Fortunately, the Assistant can help me with that, too.  I'll show you how in my next post



Name: Rahul Thakur • Tuesday, January 8, 2013

Good one. But would like to have an elaborate explaination on what hypothesis test I should use. Don't you think softwares like minitabs make one more mechanical than techically competent ?



Name: Eston Martz • Tuesday, January 8, 2013

Hi Rahul, thanks for your comment. There are many levels of explanation about why specific tests might be best in specific situations, so I'm sorry if this post didn't go the depths you'd hoped. I don't believe using software makes one more mechanical; we could make the same argument about the spell-check function in text-editing software. Using software doesn't minimize the value of technical competence, but the Assistant in Minitab was designed to help people get meaningful information from their data whether or not they have a substantial statistical background. It can guide you to the appropriate analysis, but the question of how to act on the results of your analysis is where technical competence really comes into play!

Name: Irfan • Wednesday, May 15, 2013

very grateful to have such a wonderfl site. Please explain if pulse rate is discrete or continuous variable. URGENT. regards

Name: Eston Martz • Wednesday, May 15, 2013

Thank you for the kind words. This is one of those cases where the distinction between discrete and continuous is a bit grey, i.e. you could make a fraction or decimal of pulse rate, depending on how you measure it. But since pulse rate is typically measured in whole numbers, i.e., counted by individual beats, it fits the textbook definition of a discrete variable.

blog comments powered by Disqus