0
$\begingroup$

I have two data sets, i.e., the number of flowers measured in field A and in field B. Each data set has about 40 samples. I want to do a test to check the hypothesis that field A has more flowers than field B. Which test I have to use.

I used Kolmogrov test and lillietest in Matlab, to check if my data is normally distributed, and I also plot the histogram. It seems that the data doesn't have normal distribution.

Improve this question
$\endgroup$
4
  • $\begingroup$ Welcome to our site, Mina! You have come to the right place: a search for keywords related to your question--mean, test, and normal--turns up over a thousand posts you can refer to. In looking those over you will quickly discover that "Student t-test" appears a lot. Narrowing the search produces many answers to your question. I have chosen one of those as a duplicate. Re the non-normality, take a look at stats.stackexchange.com/questions/51827. $\endgroup$
    – whuber ♦
    Sep 19, 2014 at 14:35
  • $\begingroup$ unfortunately the question is not the same. Because I can not use t test since my data is not normally distributed. And in the question you mention, t-test is being asked! $\endgroup$
    – Mina
    Sep 19, 2014 at 15:58
  • $\begingroup$ The applicability of the t-test has been addressed in many questions and comments, Mina. Exploring those searches should turn up much information. The bottom line is that non-normality of the data does not preclude using the t-test. What matters is the normality of the sampling distribution of the mean rather than the normality of the data distribution. Assessing the sampling distribution comes down to evaluating how and how much the data depart from normality and considering how much data there are. $\endgroup$
    – whuber ♦
    Sep 19, 2014 at 16:01
  • $\begingroup$ A directly relevant thread is at stats.stackexchange.com/questions/9573/…. The accepted answer looks particularly worth reading. I found this with a search on t-test normality sampling distribution and many of the other (500+) hits look like they might be of use to you, too. $\endgroup$
    – whuber ♦
    Sep 19, 2014 at 16:03

0

Reset to default