Cannabis Campaigners' Guide News Database result:


After you have finished reading this article you can click here to go back.

UK: Can you produce scientifically meaningful results by studying only a handful of people?

Alok Jha

The Guardian

Thursday 16 Oct 2003

---
Yes, but it depends on what you're measuring.

Scientists in New York announced earlier this week that smoking cannabis
regularly could impair male fertility. In a study of 22 cannabis smokers
and 59 fertile men, Lani Burkman, a researcher at the Buffalo University
school of medicine, found that the sperm from the smokers moved "too fast,
too early". She concluded that the active ingredients in cannabis must be
doing something to affect the sperm.

The story made the newspapers, including this one, but with only 81 people
on the trial, can it really be meaningful? Sample sizes are normally huge -
epidemiological studies, for example, typically use tens of thousands of
people to give statistically sound results.

Roy Anderson, a professor of infectious disease epidemiology at Imperial
College London, says that if whatever is being studied has big enough
effects, a sample size of just 10 people could provide useful results. For
example, height distributions by age normally have little variance over
large populations. If you wanted to check, say, whether a chemical
pollutant is affecting people's height at a certain age (and the effect of
the chemical was big enough), you could probably do it with a relatively
small sample population.

Drug testing require thousands of people however. The exact number depends
on what the trials are meant to show. If doctors want to claim that their
latest medicine can cure a certain percentage of people with a particular
disease, for example, there are specific calculations that will work out
how many people should be involved in a trial to prove the claim.

 

 

 

After you have finished reading this article you can click here to go back.




This page was created by the Cannabis Campaigners' Guide.
Feel free to link to this page!