Which test to use to calculate the sample size? I do not know which test (or 2 Sample T-Test, two proportions, etc.) is one that applies most to the formulation of a sample size of our experimental model.
Our problem is the following:
For the hypothesis of a factor explaining why the blood pressure measurements of digital blood pressure monitor are less accurate than the gold standard sphymomonameter mercury.
Our hypothesis is as follows:
The repeatability of measurements of blood pressure using a digital monitor Lumiscope 1100 blood pressure decreases in the presence of an electromagnetic field generated by a cell phone Samsung SCH-U740.
Our method:
We intend to record the blood pressure of a sample of the population of some. For each person, three additional pressure of blood measures will be taken without the presence of an electromagnetic field and more than three blood pressure measurements will be taken to the presence of an electromagnetic field generated by a cellular phone.
The standard deviation of blood pressure measurements Lumiscope 1100: + / - 3mmHg.
Your assumption is one way. It is better to let the two sides until the measures have been taken. If the average mobile phone data show a smaller value, making this assumption.
You can average of 3 measurements of each person and use the average value as a single measure for both measures. Because measuring the same person is used for two samples, a paired t-test (which assumes normality) can be used.
If the proportions are used, differences in the proportion should be used as the person whose blood pressure is taken using these two different instruments is the same.
Look under dependent t-test (source explains the test)
In general, the larger the sample, the smaller will be the error.
My comments are suggestions on what I know.
Posted on March 4, 2010.
