Message-ID: <9e37e45abdd247dcb725ad2003a1e2ce@GBDCVPEXC04.corp.lgc-group.com>
Date: 2019-03-19T17:06:43Z
From: S Ellison
Subject: high p values
In-Reply-To: <CAJhui+vu9Dy2T7nVmc8S58woLijgA7hLiXt_8WG5odcKHs57tw@mail.gmail.com>
> This is my function:
>
> wilcox.test(A,B, data = data, paired = FALSE)
>
> It gives me high p value, though the median of A column is 6900 and B
> column is 3500.
>
> Why it gives p value high if there is a difference in the median?
Perhaps becuase a) because you are testing the wrong data or b) there isn't a significant difference
a) You are probably not using the data you think you are. Check ?wilcox.test; the 'data' argument is specific to the formula method. That needs a formula as the first argument, not a numeric vector. What you've done is apply the default, and 'data' has been ignored. So A and B are whatever was lying around in your current environment, not what is in 'data'. ('data' is a terrible name for a data frame, by the way, as 'data' is an R function).
After that:
- How many data points do you have in each group?
- How much do the two groups overlap?
If the answers are 'not many' or 'lots' (in that order), and especially if both apply, you can't expect a significant test result.
S Ellison
*******************************************************************
This email and any attachments are confidential. Any use...{{dropped:8}}