Statistics Help @ Talk Stats Forum - Biostatistics
http://www.talkstats.com/
Epidemiology and biostatistics, public health research. GLM, logistic regression, survival analysis, clinical trials.enWed, 16 Apr 2014 21:33:33 GMTvBulletin60http://www.talkstats.com/images/misc/rss.pngStatistics Help @ Talk Stats Forum - Biostatistics
http://www.talkstats.com/
Appropriate hypothesis test for beta or fractional logistic distributions
http://www.talkstats.com/showthread.php/55075-Appropriate-hypothesis-test-for-beta-or-fractional-logistic-distributions?goto=newpost
Wed, 16 Apr 2014 14:41:47 GMTHi! Thanks to previous suggestions I have determined that my bounded data sets are fit well by beta and/or fractional logistic distributions. I am...Hi! Thanks to previous suggestions I have determined that my bounded data sets are fit well by beta and/or fractional logistic distributions. I am wondering now, what the appropriate hypothesis tests would be for such distributions? I have been using Kruskall-Wallis and the Mann-Whitney/Wilcox ranksum tests, are these appropriate? Does anyone have any other suggestions? Also, I tried a bootstrap test which seemed to give exceedingly liberal p-values and am not comfortable with this...

Thanks!
]]>Biostatisticsphayneshttp://www.talkstats.com/showthread.php/55075-Appropriate-hypothesis-test-for-beta-or-fractional-logistic-distributionsPoint Prevalence over 5 years when only given data for 1 year
http://www.talkstats.com/showthread.php/55071-Point-Prevalence-over-5-years-when-only-given-data-for-1-year?goto=newpost
Wed, 16 Apr 2014 10:56:49 GMTHi all,

I'm abit confused about a tricky sort of Prevalence calculation.

So here's my problem-

Sample of 500 people. In 1 year 5 participants will develop incurable blindness.
Assumptions - No blind people in population to begin with and perfectly even occurence of blindness throughout the study year.

Calculate - Incidence? Prevalence after 1 year? Point Prevalence after 5 years?

5/500 -> incidence & prevalence both= 1% in this population in this year

So Now - If i need to calculate the point prevalence @ the end of 5 years (without any further info) of this 'exact' cohorts blindness is it simply 25/500 = 5% prevalence
OR
a bit more complicated & using "incidence" rate of 1% to predict future numbers

New Cases- @ 1% incidence per year
5/500 1st year
4.95/495 2nd year
4.9/490.05 3rd yr
4.85/485.15 4th yr
4.8/480.3 5th year

5+4.95+4.9+4.85+4.8
Therefore is the Prevalence 24.5/500 or 4.9% at the end of the 5years?

Apologies if this seems a tad simple it's just doing my head in.

Cheers in advance for any guidance
]]>BiostatisticsRekanshahttp://www.talkstats.com/showthread.php/55071-Point-Prevalence-over-5-years-when-only-given-data-for-1-year