Determining Sample
Size
Sampling may also be unacceptable if it causes more disturbance than
would simply measuring everyone. For example, it may disrupt a
lesson more to withdraw six pupils than to test the entire class. Three
rules of thumb for sampling sites are listed below:
| |
 |
|
|
- Sample sizes over 30 have certain advantages, statistically.
- Correlations are very unstable on samples smaller than 50
- 100.
- Often you will simply use the largest sample on which you
can collect high quality data within the constraints of time
and money available.
|
|
![[Skip Navigation]](/images/pixel.gif){kind=tracking}
