G
Gerwyn
Guest
A number of studies sponsored by our friends in psychiatry have quoted r values in their claims re causative factors in Me/cfs
for a start r referrs to correlation not causation if there is a strong correlation r must be greater than 0.8 any thing less than 0.5 is weak the lower the weaker.It is quite possible to get a very small pvalue but a very weak correlation
They also can quote confidence intervals----this is actually difficult to describe think of it as a measure of how confident you are that the result is representative of a population and its relatively safe to generqalise from it.In science this normally means that the scatter of your data must be no more than 5% either side of your mean or average value
If you look at the data published by Heim et al claiming a six fold increase in cfs symptoms in those suffereing from self reported child abuse the r value was0.3
and the confidence intervals varied about 70% on each side of the mean that was somehow skewed If I continued to produce fairy stories I,d probably be sectioned----could I suggest the same treatment here!
Saying that cfs symptoms are 6times more likely to occur in patients with CFS than healthy controlls with such a small r value is a little like saying that the symptoms of depression are 6 times more likely than in people who dont have depression!
for a start r referrs to correlation not causation if there is a strong correlation r must be greater than 0.8 any thing less than 0.5 is weak the lower the weaker.It is quite possible to get a very small pvalue but a very weak correlation
They also can quote confidence intervals----this is actually difficult to describe think of it as a measure of how confident you are that the result is representative of a population and its relatively safe to generqalise from it.In science this normally means that the scatter of your data must be no more than 5% either side of your mean or average value
If you look at the data published by Heim et al claiming a six fold increase in cfs symptoms in those suffereing from self reported child abuse the r value was0.3
and the confidence intervals varied about 70% on each side of the mean that was somehow skewed If I continued to produce fairy stories I,d probably be sectioned----could I suggest the same treatment here!
Saying that cfs symptoms are 6times more likely to occur in patients with CFS than healthy controlls with such a small r value is a little like saying that the symptoms of depression are 6 times more likely than in people who dont have depression!