Hi all
i need help in one silly question
Can any one help me understand the meaning of p value?
if we say p value = 0.05 then what does that means.
what is confidence interval?
If we say 95% confidence interval then what does that means.
I read certain article where it says p value is error estimation . If we are dealing with trancriptome data then in this data what could be error. meaning people select pvalue < 0.05 as significance what this significance means???
When running a test you are running a risk of type I error (accepting null hypothesis when it is not true, false positives, alpha) and type II error (failing to reject the null hypothesis when in fact it is false, false negatives, beta). P value is the smallest probability (alpha) at which your test is still significant. P-value of 0.05 basically means you are running a 5% risk that your test will be significant by chance and this value is commonly accepted as sufficient for a single test. So when running a test and returning the P-value of e.g. 0.003 (which is smaller than 0.05) you accept the value as significant since the probability of running a type I error (false positives, basically your P value) is smaller than prior selected (0.05).
Confidence intervals give you a range of values that would be significant at a preselected confidence level (1 - selected P value * 100 --> 1 - 0.05 * 100 = 95%).
False Discovery Rate correction deals with the issue that while P=0.05 for one test is ok, if you do a lot of tests (for different genes or whatever), then ~5% of your results are going to get called as significant but are actually a false positive.
FDR corrections modify the calculated P-value to make it more strict based upon the number of tests you are doing. There are several different options (see here: https://en.wikipedia.org/wiki/False_discovery_rate ), but usually whatever pipeline you're usiong will have an option to inculde it, or it will do it automatically. This should be used as your P-value.
The P value of < 0.05 means that the chances of a gene/transcript not being expressed is less than .05. So you can confidentially say that transcripts with value <.05 is actually expressed.
Pvalue > .005 that there is a high chance that transcript might not be expressed at all.
Pvalue=.05 the transcript can be wither expressed or not expressed at all.
In simple words, pvalue helps you deduce conclusion between null hypothesis(no expression) and alternate hypothesis(expressed).
You are opening such a can of worms! Just few days ago I posted this ASA discusses limitations of p-values what are the alternatives? Oh, if you get confused about p-values be reassured that at least you are in good company.
Is this an April fool?
...Mmm, no, no April fool here, why? (Hope there is no misunderstanding going on...)
sorry I meant the original question
i was not knowing abt ASA discussion on P value. But still people uses this.