5
$\begingroup$

I am currently undertaking a meta-analysis of the expression profile of specific biomarkers in a specific disease. During the extraction process, I extracted the name of the biomarkers and their corresponding p-value for each included study. The p value indicates how “good” this biomarker correlates with the disease.

In some studies, only p values in the form of “p < 0.05”, “p.<0.01”, “p<0.001”,”p>0.05" were provided. Unfortunately, the program I am using (which conducts a p-value based meta-analysis) doesn’t recognise those p-value forms.

One of my supervisor suggested me to convert p<0.05 to 0.05, p<0.01 to 0.01 and p>0.05 to 0.5. However he couldn’t give me a rationale for this.

Could someone provide me a rationale on how I should/could convert these p values? I think that reviewers might ask on which basis this conversion was done.

$\endgroup$
  • 2
    $\begingroup$ Maybe you could convert these into categories. p < .001 = "Highly Significant" p < .05 = "Significant" and p > .05 = "Not significant" and then do whatever analysis you were interested in using a categorical method. The other option is to do as mdewey suggested and convert your values in the most conservative way possible, so it doesn't seem like your fiddling with your data for a certain result. $\endgroup$ – RNM Aug 17 '16 at 15:27
  • $\begingroup$ Thank you for your input @Robert Montgomery. I will keep a note of your suggestion! Best regards. Markus $\endgroup$ – m.stevensson Aug 18 '16 at 19:30
  • 1
    $\begingroup$ conversion proposed is likely to provide us with the unjustified meta-analytic results. $\endgroup$ – Subhash C. Davar Apr 27 '18 at 4:39
  • 1
    $\begingroup$ This way the p-values are overestimated causing the p-value for the meta-analysis to also be overestimated. When doing meta-analysis it would be much better to have the exact p-values. $\endgroup$ – Michael R. Chernick Apr 27 '18 at 5:31
5
$\begingroup$

If you convert $p<0.05$ to $p=0.05$ then your analysis will be conservative but you will at least have been able to include all the studies. Similarly for $p < 0.01$ and so on. The problem comes from the ones which say $p > 0.05$ as the only safe option here is to convert them to $p = 1$, your suggestion of $p = 0.5$ cannot really be justified. Be careful with the method you use as some of them do not allow for $p=0$ or $p=1$ so you would need to use a value very close to the boundary.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.