|  | |
SSC Quickhelp | | |
Sample Size Calculation
Apart from corrections, there are three main factors that influence sample size:
-
Variability: the amount of difference within the population. If the population is very homogeneous
with respect to the characteristic you want to investigate, you don't
need a big sample size to establish a reasonable estimate of the characteristic. In statistical terms,
this factor is measured by the standard deviation of the
population characteristic (or a good estimate thereof).
-
Confidence: the amount of risk you are willing to take to reach false conclusions. This factor
comes in two tastes. The first is to reach the conclusion that your
research hypothesis is true, while in fact it is false. The second is the risk you take to say that
the research hypothesis is false, while in fact it is true. These factors
are statistically translated as alpha and beta. The smaller you want the risk to be, the higher a sample
size you need.
-
Tolerance: How big a difference actually makes a difference for you? Suppose you have two mean
values for a certain population characteristic: 10.2 and 10.23.
Does this make sense, translated to your real world? Will you act differently for both groups? The smaller
the difference you're interested in, the more samples you
will need.
The way you need to interpret these three factors depends on the type of research you want to perform.
The next section handles the subject of the analysis type.
next Type of analysis
|