0

I am doing a biology experiment for the first time, and even though the teacher did not require me to do this, I calculated the p-values for the data on my experiment.

They all are above 0.25. From what I have read online, this is beyond acceptable.

However the goal of my biology experiment is to prove that if I water my plants 35% less there should be no statistical difference between my plants and the pants that are watered as normal.

Wouldn't it, in this case, mean that the data that I obtained from watering my plants less is very similar to the data from the controlled sample and therefore we would want a p value that is close to 1?

Tim
  • 108,699
  • 20
  • 212
  • 390
Laura Salas
  • 3
  • 1
  • 1
    Could you try describing your procedure in greater detail, step-by-step? How did you calculate the p-value? What exactly is the hypothesis that you are trying to test? You want to see if there is a difference in plant's growth given the experimental condition, where for one of the conditions you watered them less? – Tim May 30 '19 at 07:54
  • I used excel's T-Test; selected a sample size of 10 data points for 2 arrays, set it to 2 tails and of type 2. Yes, I basically want to confirm that watering them less does not have a significant impact on the data. I apologize if I sound so lost, like I said, I only learnt about this a couple hours ago. I will gladly provide any other info that might be of help! – Laura Salas May 30 '19 at 08:14
  • 1
    No statistics is better than bad statistics. You have to be very very careful with statistics - the assumptions behind the tests, how your data was collected, how the results can be interpreted. Do not just apply tests that sound like they should fit your problem - the results are likely to be misleading! – rinspy May 30 '19 at 08:25
  • Consider, for example, what it is exactly that you want to test. "No statistical difference" - what exactly do you mean by that? If you gathered a billion samples watered at 100%, and a billion samples at 65%, and even if you managed to keep all other variables perfectly controlled, I bet that you would find a possibly very very tiny, but nonetheless significant (in the sense that it is real, reproducible and caused by the difference in watering) difference between the two samples. – rinspy May 30 '19 at 08:31

1 Answers1

2

In the comment you said that

Yes, I basically want to confirm that watering them less does not have a significant impact on the data.

In such case, you want to run an test, for example two one-sided tests (see questions tagged as ), as standard hypothesis test won't work for such scenario. You can find more details in Why do statisticians say a non-significant result means "you can't reject the null" as opposed to accepting the null hypothesis?, Does failure to reject the null in Neyman-Pearson approach mean that one should "accept" it?, and Is it possible to prove a null hypothesis?, to give few examples of related threads. You should start with reading the threads, especially the last one seems to answer your question.

Tim
  • 108,699
  • 20
  • 212
  • 390