Confidence Interval FA

[Broker]

Advertisement - Members don't see this ad

I have just figured out how to work out the CI using the formula:

CI = +/- Z x SEM (standard error of the mean)

But can someone explain to me what is meant by the text in FA:

'If the 95% CI for a mean difference between 2 variables includes 0, then there is no significant difference and the null hypothesis is not rejected. If the 95% CI odds ratio or relative risk includes 1, the null hypothesis is not rejected'

Any help?

 

[Torquer]

If in a 95% CI of a Relative Risk calculation, you get a range, say, [0.8 - 1.5], then you can NOT reject the null hypothesis. Why? Recall that while RR <1 shows negative association and >1 shows positive association, 1 shows NO association. Which is exactly what a null hypothesis proclaims.
So essentially you're saying there's a 95% chance that your RR will be 1 = null

Sames the situation with mean difference between variables being 0 - which is exactly what the null is trying to scream out - that theres no difference! So we fail to reject null unfortunately in those situations too.

hope that helps!

 

[AndreyE]

I'll wait for the answer, but what I know is that if CI goes thru 1.00 (eg CI 0.85-1.01) that means you can disregard the statement this CI supports. So if a comparison of two trials shows CI 0.85-1.01 that means the trial is not informative and can't be trusted. Something like that, correct me for terminology 😛
PS But if it's 1.01-1.15 then you can trust it!

 

[Broker]

Thanks for the help, makes more sense now I relate it to relative risk and difference!

 

[jfgavina]

how do you calculate confidence interval for relative risk or odds ratio?

 

[Torquer]

how do you calculate confidence interval for relative risk or odds ratio?

i remember there was a massive formula. doubt its important, the CI is usually given in the stem of the question anyway.

i suck at math. just have the concepts UW gave😛