PakMediNet Discussion Forum : Biostatistics : p-value
p-value
I would be highly obliged if any one can guide me how to calculate p-value from clinical data.
Looking for the response.
Thanks in anticipation.
Posted by: shafeezPosts: 3 :: 25-04-2002 :: | Reply to this Message
Re: p-valuep-value (significance) can be calculated using many formulas in the statistics. For your study, type of formula depends on your variables and nature of the study. Choose an appropriate formula and apply the test. This can be easily done in SPSS.
Posted by: docosamaPosts: 333 :: 26-04-2002 :: | Reply to this Message
Re: p-value
A P-value is always related to a hypothesis test; it is the probability of obtaining a result as or more extreme than the one observed, if the null hypothesis is true. Some people like to think of the P-value as the probability that the observed result is due to chance alone. The P-value is calculated after the statistical test has been performed; if the P-value is less than alpha ( alpha is the level of significance chosen before the test is performed and gives the probability of incorrectly rejecting the null hypothesis when it is actually true ), the null hypothesis is rejected. Traditional values used for alpha are small because you do not want to reject the null hypothesis when it is true and therefore, most people like to choose a value of 0.05.
So whatever study you are conducting, you should have already calculated the number of cases required for you to a reach a level of significance and once that number is reached, select an appropriate test according to your study design and see if the level of significance can be reached.
[Edited by chameed on 04-26-2002 at 05:46 PM GMT]
Posted by: chameedPosts: 173 :: 26-04-2002 :: | Reply to this Message
How to calculate p valuechameed has correctly explained the method of determination of p value in the language of statistics. However the newcomers to this field simply cannot follow the statistical terms. In my experience hypothesis testing is the most difficult portion to explain to the young medical postgraduate students. Why a hypothesis? Why a null hypothesis? Why to formulate a null hypothesis with the sole aim of rejecting it? Why an alternate hypothesis? These are the questions which come into minds of most of the students. And these are difficult to explain to a student with no background knowledge of biostatistics. Therefore my advice for all serious students of "p value" will be to start reading a book on biostatistics. Best such book, in my opinion, is "Statistics at square one" by TDV Swinscow. This book is based on series of articles published in British Medical Journal. The book is available on Internet and its link is already available at PakMediNet. Concentrate on the solved examples and also try to solve them yourself. Moreover, carry out the exercises given at the end of each chapter. You will automatically learn to calculate p value.
Posted by: saeedkhanPosts: 7 :: 04-05-2002 :: | Reply to this Message
Re: How to calculate p value
quote:
saeedkhan wrote:
chameed has correctly explained the method of determination of p value in the language of statistics. However the newcomers to this field simply cannot follow the statistical terms.
A statistical question was answered in statistical language.
quote:
In my experience hypothesis testing is the most difficult portion to explain to the young medical postgraduate students. Why a hypothesis? Why a null hypothesis? Why to formulate a null hypothesis with the sole aim of rejecting it? Why an alternate hypothesis? These are the questions which come into minds of most of the students. And these are difficult to explain to a student with no background knowledge of biostatistics.
All of the questions raised regarding hypothesis testing can be easily expalined with one simple example. The problem is that some one has to be dedicated enough to take the time and explain to the young.
quote:
Therefore my advice for all serious students of "p value" will be to start reading a book on biostatistics. Best such book, in my opinion, is "Statistics at square one" by TDV Swinscow. This book is based on series of articles published in British Medical Journal. The book is available on Internet and its link is already available at PakMediNet. Concentrate on the solved examples and also try to solve them yourself. Moreover, carry out the exercises given at the end of each chapter. You will automatically learn to calculate p value.
The person who came here looking for answers, probably has been through all the routine above.Posted by: chameedPosts: 173 :: 08-05-2002 :: | Reply to this Message
Re: p-value
Do u All think that a small article in one of medical journals trying to explain statistics in simple language will be worth an effort? Or enough is already written about this?
Posted by: shahabPosts: 2 :: 21-02-2003 :: | Reply to this Message
Re: Re: p-value
quote:
shahab wrote:
Do u All think that a small article in one of medical journals trying to explain statistics in simple language will be worth an effort? Or enough is already written about this?
Posted by: anwer_khurPosts: 30 :: 22-02-2003 :: | Reply to this Message
Re: Re: Re: p-value
Dear Anwer Khursheed Sir,
It will be a great help if you can guide us about basics of statstics relevant to clinical studies.specifically,p-values, confidence interval,and when a data is statistically significant.
quote:
anwer_khur wrote:
quote:
shahab wrote:
Do u All think that a small article in one of medical journals trying to explain statistics in simple language will be worth an effort? Or enough is already written about this?
Dear Dr. Shahab:
Just read your message. I am a biostatistician. If you get this message please reply so that we can discuss the problems of P_values.
Regards.
Professor Anwer Khurshid
Department of Statistics
University of Karachi
Karachi PC75270
Posted by: majorirfanPosts: 5 :: 24-02-2003 :: | Reply to this Message
Re: p-value
Dear All User of Statistics:
I have written a short article on p-values:
"Role of P-values in Decision Making", which has been published in Proceedings of Eighth Statistics Seminar. It will (hopefully) answer all (most?) of the questions. Those who want the reprint please send me their mailing address. I will send the reprint by mail. Unfortunately, I dont have the electronic version right now.
For any other bio(statistical) problems, dont hesitate to contact me.
I beleive our combined efforts (doctors and (bio)Statisticians) will uplift the standard of statistical research in Pakistan which is definitely in very poor state (You may not agree with me but this is a a fcat)
Regards.
Professor Anwer Khurshid
Department of Statistics
University of Karachi
Karachi PC75270
Posted by: anwer_khurPosts: 30 :: 24-02-2003 :: | Reply to this Message
Re: Re: p-value
Dear Prof. Khursheed,
I shall appreciate receiving a reprint of your article. I am Director Research, Pakistan Medical Research Council at Fatima Jinnah Medical College , Lahore. Since I teach Research Methodology to Postgraduates as well Faculty of all Medical Colleges on behalf of College of Physicians & Surgeons Pakistan, your article will be a great assett for me.
My postal address is as follows.
32 / G, Gulberg III
Lahore.
Sincerely
Dr. N.Rehan
quote:
anwer_khur wrote:
Dear All User of Statistics:
I have written a short article on p-values:
"Role of P-values in Decision Making", which has been published in Proceedings of Eighth Statistics Seminar. It will (hopefully) answer all (most?) of the questions. Those who want the reprint please send me their mailing address. I will send the reprint by mail. Unfortunately, I dont have the electronic version right now.
For any other bio(statistical) problems, dont hesitate to contact me.
I beleive our combined efforts (doctors and (bio)Statisticians) will uplift the standard of statistical research in Pakistan which is definitely in very poor state (You may not agree with me but this is a a fcat)
Regards.
Professor Anwer Khurshid
Department of Statistics
University of Karachi
Karachi PC75270
Posted by: nrehanPosts: 6 :: 15-03-2003 :: | Reply to this Message
Re: p-value
Now you can download this article from PakMediNet (PDF Format) at
Role of P-values in Decision Making:
http://www.pakmedinet.com/files/anwer1.pdf
This has been recently done.
Thank you to Prof. Anwer
[Edited by docosama on 03-16-2003 at 03:40 AM GMT]
Posted by: docosamaPosts: 333 :: 16-03-2003 :: | Reply to this Message
p-value
A friend of mine sent me an interesting e-mail with a question. Let me first share a part of his e-mail and then I will ask you my dilemma. He wrote:
“Boyfriend of my office manager claims that he knows an ancient secret formula by which he can predict the gender of a fetus. We have to tell him the first name of the mother and her LMP only. Last week we planned to test his claim. We had him predict the gender of the fetus prior to ultrasound (when no one knew the gender of the fetus) and then compared his predictions with the ultrasound findings. We had 10 ultrasounds scheduled that day. To our surprise, he predicted gender of 9 out of 10 fetuses correctly.
Should I admit that he knows such a formula? “
In other words, boyfriend of my friend’s office manager had 9 successes and 1 failure out of 10 trials. A 2-tailed p-value is 0.02 (ideally, I should be using a 1-tailed p-value which is 0.01). Thus we can reject the null hypothesis that boyfriend can’t predict the gender and will have to accept the alternate hypothesis that he can predict the gender of an unborn fetus.
My dilemma is that, based on my understanding of the Medicine, it is not possible to predict gender by knowing only the first name and LMP of the mother. How can I combine my understanding of the Medicine and this p-value of the claim of that boyfriend? Any suggestions?
Posted by: rqayyumPosts: 199 :: 10-05-2006 :: | Reply to this Message
Re: p-valueQuite interesting .. I still believe that there is a lot ahead of allopathic medicine, "Science of Formulas & Mathematics". Statistically the person has proved that his power of prediction of a gender is significant (p<0.05).
Posted by: yasirPosts: 90 :: 11-05-2006 :: | Reply to this Message
p-value
Yes, statistically, p-value is significant (if we consider a p-value of <0.05 significant). However, does it make sense? I mean, unless we believe in meta-physics, or occult ‘science” we can’t believe it. Can we? Can this small experiment refute all the other evidence?
A p-value of 0.02 means that if we repeat this experiment 100 times, just by chance, twice we will reject a null hypothesis wrongly, in other words, false positive rate is 2%. Is it possible that this was one of those two instances?
If someone makes a claim, similar to the above claim, how many of us will believe him? I would probably say that there is less than 1% chance that he is making a correct claim. My question was how can I combine my skeptic view of his claim to this experimental data?
Let me give an example. Let’s say that you receive a hemoglobin level of 14.0 on a patient from a high-quality and very reliable lab. However, your clinical examination tells you that patient’s hemoglobin level is much lower, probably 8 or 9. What will be the next step? Most likely, you will depend on your clinical examination assuming that this hemoglobin level was due to some ‘error’, manage accordingly, and send repeat hemoglobin levels. If the result is again 14.0, you will, most likely, change your earlier impression.
This is how we all accumulate knowledge. We evaluate each new piece of data in light of our previous knowledge (or opinion). If new data contradicts our previous knowledge, we update our previous knowledge/opinion. If our previous opinion is very strong, new data will weaken our opinion. However, if our previous opinion is weak or data is overwhelming, we may change our opinion.
Back to my question, can I combine previous knowledge with this new data? If yes, then how?
Posted by: rqayyumPosts: 199 :: 13-05-2006 :: | Reply to this Message
Re: Re: p-value
am a young doctor from kenya,am presently doing private studies on research methodologies
can anyone share nice titles in biostatistics especially tackling issues of hypothesis formulations,testing.
thanks.
Posted by: giokoPosts: 2 :: 08-03-2007 :: | Reply to this Message
Re: Re: Re: p-valueI am not sure I understand your question. Could you explain in more detail?
Posted by: siddiqiPosts: 19 :: 13-03-2007 :: | Reply to this Message
Re: Re: Re: p-value
Dear Anwer Khursheed Sir,
It will be a great help if you can guide me about the my results.im sending u the results of my study can u help me sir
Posted by: DRKHUHROPosts: 4 :: 28-09-2007 :: | Reply to this Message
Re: Re: p-value
Dear sir, i shall be highly obliged if you send me a copy of your article on p-value
thanks
Posted by: samreenqmcPosts: 140 :: 09-10-2008 :: | Reply to this Message
Re: p-value
Dear Dr. Samreen:
Thnaks for your interest in my article. Just go to Research Guidelines of PakMediNet and you will find it. Hope it will give some answers to your questions.
Anwer
Posted by: anwer_khurPosts: 30 :: 09-10-2008 :: | Reply to this Message