Statistical power

[allemory]

How can I use the power to find the n if I have 2 treatments measured over 10 years...once each year? I am confounded. Any help or redirect would be greatly appreciated.

[sanctus]

And I am confused on what your question actually is :-)

Could describe the problem a bit more in detail, i.e. what is n, etc?

[allemory]

Sorry. The research involves a relationship between chemically treated limbs and non chemically treated forelimbs in mice. Looking at Muscle mass changes over time. I have data from over 1 year and I wanted to look at a ratio from each comparison. Some measures at time zero...others (in different mice) at one month intervals throughout the year. I just need to know what n would be required to give a power around .80.

 

treated to untreated ratio

 

time zero...one month...two months...three months...and so on

 

Problem for me is the data does not track individuals...so some of the mice were treated last week...some were treated in June...others last November...

 

any of that help?

[Tylor Lothar]

Experimental data will estimate its difference after data has been collected. The difficulty with formulae for Sample Size determination is that calculating the value of 'n' (Sample Size) requires knowing the standard deviation, which is of course not available until a sample has been selected. Instead, prior information about standard deviation (or an estimate) may be utilized as a basis for a reasonable "guess" for the standard deviation. Alternatively, for a population distribution that is not too skewed, dividing the range (difference between the larges and smallest values) by 4 often give a rough idea of what the standard deviation might be.