P-Value and Hypothesis Testing

P Value in Arabic

Well, p-value?

This magical value (p-value) that determines the fate of hypotheses. We usually find this value in studies, it is a principal concept to understand studies, so, what does it mean?

To better understand this concept, let’s assume that a study is comparing between two drugs for the treatment of headaches, the first drug is (X) and the second is (Y). (30) individuals were recruited in this study and separated into two equal groups with (15) individuals in each group. Individuals in the first group used (X) and the second used (Y), when they had headaches. The research then, measured the effect of these two drugs in each group. Assuming that the researched found (13 out of 15) individuals in the first group  had a relief and (3 out of 15) in the second group had a relief. Looking at the numbers, there’s an obvious difference between the groups. The individuals who had a relief are more in the first group then the ones in the second. But, can we consider this as a significant difference where we can generalize these results on the population? Here comes p-value.
We cannot be certain that (X) is better than (Y) in relieving headaches before we perform the statistical analysis. After performing the analysis, we found that p-value is (p=0.0001). So, what does this indicate?
The (P) in p-value means probability, which is the probability of the headache not being relieved after using the drug. Usually we set a value for (P) before starting the analysis, so we can use this preset value as a cutoff point. This preset cutoff point is (0.05). If the statistical analysis revealed a p-value of less than this cutoff point, then we can conclude that there is a statistically significant difference between the two drugs and that (X) is better than (Y).
While if we assumed that the p-value from this statistical test was (p=0.08), i.e. p>0.05, then we can say that the difference is not statistically significant and we cannot generalize the results of this study to the population. So it is not wise to state that (X) is better than (Y) in treating headaches and we cannot prescribe this drug for patients.


Why did we chose (0.05) as a cutoff point for p-value?

There is a consensus among statisticians that (0.05) i.e. (5%) is the standard cutoff point in many cases especially in healthcare. Which means that in order to generalize results from a study to the population, we need the probability of untreated headaches -for example- is less than 5% or that the probability of treatment of headaches happening by chance is less than 5% (p<0.05).


This concept is immensely important to understand research, and understanding it, can make understanding research as a cool breeze. You can read the Arabic version of this article here.