How to Read an Article p-value
|
p-value
The p-value is a standard that tells us about the strength of
the data being presented. It is not a value which weighs the methods
or statistical method as being appropriate, but if you determine
that these variables are OK, then we look at the p-value. It is
a percentage, and reported in its raw form (0.1, 0.5 or 0.02 for
examples).
We are all aware by reading an abstract of the intent of the authors.
They may be trying to prove that aspirin is a good pain medicine,
that oxygen helps with carbon monoxide poisoning or that driving
on the left side of the road in the United States will increase
the probability of a motor vehicle collision. Their hypothesis
has a counter-argument known as the null hypothesis. Using the
examples above, the null hypotheses would be that aspirin makes
no difference in pain reduction, that oxygen makes no difference
in carbon monoxide poisoning, and that driving on the left side
of the road makes no difference in the United States.
The null hypothesis basically says that the tested variable has
no impact. The p-value is the probability of obtaining the data
the study has shown if the null hypothesis is true. Another way
of looking at this is if the p-value is 0.1 in our aspirin question
and our study did show a definite benefit with aspirin for pain
control, then if the null hypothesis were true (no pain mediation),
then you have a 10% chance of obtaining the data set of aspirin
showing no benefit. This isn’t bad, and the lower the p-value,
the more likely that our data is correct. In medicine, we like
to see the p-value at 0.05 or less (5% or less).
