topic Re: Sample Size: Number of replications needed per block in Statistical Procedures

Sample Size: Number of replications needed per block

AgReseach7 — Thu, 07 Dec 2017 15:28:41 GMT

I'm designing a trial that will be replicated on 6 ranches.

Each ranch will have 2 treatments (2 types of supplemental feed) and only 2 pastures; only 1 treatment per pasture.

Within ranch and pasture, how do I determine if I need 5, 10, 20, or ? animals.

Basically, how do I account (in my Power/sample size calculation) for the "within treatment" SD?

Data recorded reflects the effect of supplement on:

pregnancy (gave birth or did not give birth; 0 or 1)

blood serum

body weight, etc.

I am mainly interested in pregnancy (0 or 1).

Thanks

ballardw — Thu, 07 Dec 2017 15:37:55 GMT

How many animals will each pasture support?

How much $ is available for the supplemental feed and how many animals will that support?

Do you intend to account for the possibility of loss of animals during the trial ?

Have you determined the statistical tests you will use in the analysis?

AgReseach7 — Thu, 07 Dec 2017 15:54:56 GMT

all good points that will put a ceiling on animal numbers/pasture.

However, for an Anim. Use Protocol, we are looking for a justifiable #.

Currently, I'm stating that there will be a minimum of 10 animals/pasture (up to 40 depending on pasture size, etc.).

On average, 5% of the cows won't calve in normal conditions. The supplement we are using: I would like to see differences

between 95% calving rate vs. 85% calving rate (d = 10).

Don't expect any cow death loss.

Stat test (for blood serum, etc.):

We will collect data (on 3 during the trial) from individual animals, but PASTURE (same as trt in this case) is the exp. unit. Below: ID = animal ID

(I'm not 100% sure the following [repeated/random statements] is correct, but...)

Proc Mixed;

Class RANCH PASTURE DAY ID;

MODEL SERUM N = PASTURE | DAY / DDFM=KR SOLUTION;

REPEATED DAY / SUBJECT=pasture TYPE = xxx;

Random = pasture(ranch)

LSMEANS pasture|DAY / DIFF ADJUST=SIMULATE (REPORT SEED=121211) cl adjdfe=row SLICEDIFF=DAY

thanks for your time.