Page 1 of 1: biostatistics

drdp · Credits: 1467 My Scrapbook My Reading List 6 Books

which of the following results gives the reader the most info concerning statistical significance,sample size,and strength of association?

a:a relative risk of 2.5 with 95%confidence interval of 2.0 to 3.1
b:a p value of 0.0004 and an alpha of 0.05
c:a relative risk of 5.0 with 95% confidence interval of 0.1 to 9.8
d:a p value of <0.05 and relative risk of 2.5

guest · Credits: 77713 My Scrapbook My Reading List 262569 Books

I think it should be B, as very small p value gives high value of significance and large sample size. 0.5 alpha gives low false positivity!

drdp · Credits: 1467 My Scrapbook My Reading List 6 Books

well i thought the same..!
but the answer given is a
a narrow confidence limit does help but how to compare a and b

guest · Credits: 77713 My Scrapbook My Reading List 262569 Books

Actually b has a shortcoming of strength of association! relative risk is a direct measure of it! low alpha error means good positive predictive value, but then PPV does not give a measure of strength of association! Rest I am not sure!

drdp · Credits: 1467 My Scrapbook My Reading List 6 Books

thanks! and what about confidence interval

guest · Credits: 77713 My Scrapbook My Reading List 262569 Books

Well, the only problem I see with (a) is how to say it has large sample size! All I can say is that confidence interval depends on mean and standard error. Standard error itself depends on standard deviation and sample size. If we keep standard deviation constant then large sample size will give small standard error and hence small confidence interval. It very well fits with option (a) provided the assumption of fixed standard deviation holds!

drdp · Credits: 1467 My Scrapbook My Reading List 6 Books

yeah that makes sense..the q is from bhatia gt..wonder where they picked it up from..thanx neway

allergic2morons · Credits: 8107 My Scrapbook

Meta Analysis of Chi Square Study Identifies Insignificant Similarities Among P Values.

Statistician Melvin Beuberhiem explains that “Recent studies refute the finding that the null hypothesis analysis was not improperly under-powered, and assert that the previous result, as an inverse function of one minus the positive predictive value divided by the sensitivity, was significant in an intention to treat cohort, implying a higher number needed to treat. The zaney thing was, the odds ratio remained unchanged.."
He then confessed epidemiology was “a sham,” concocted by Steven Hawking to score with chicks

sorry friends for being too offbeatish in this forum but couldnt help!!

shuvo · Credits: 4595 My Scrapbook My Reading List 11 Books

The answer should be a. The advantages of of CI's over the statistical tests is , they give an idea of all probable values for the population parameter and in that respect contain much more information than a single p value. They immediately give an impression of the precision of the estimate. As such they make a qualitative appreciation of ( the lack of) statistical power possible. Very high CIs carry a high risk of beta error...

ashy_kumar · Credits: 258 My Scrapbook

ans is a.)
coz..i)R.R= 2.5 is a positive association between the two factors..i..e. the strength of association
ii)C. I. = 95% IS p</=0.05 therefore statistically significant asso.
iii)now main thing is....in a std. normal curve, narrow confidence interval indicates distibution of means in a narrow area..which in turn indicates large sample size..

		Sign In New User? Sign Up Sign in to access your control panel and messenger!
				TechZone \| SpiderNevi \| HowTo? \| Scrapbook!