|
In most research projects, the number of
people that must be sampled is determined
by the amount of error that is willing to
be tolerated. Errors in survey research
come from three distinct sources.
1. Missing data from non-interviews and
from item non-response.
2. Inaccurate information given by respondents
or recorded by interviewers.
3. Sampling error.
The first two sources of error are independent
of the number of people in the sample. The
third source of error is very much dependent
on the number of people in the sample.
Interviewing all
of the people in the target population will
eliminate all sampling error. Practical
considerations, including cost, prevent
gathering information from everyone in the
population. Therefore, all surveys that
draw a sample of people from the total population
are subject to sampling error. How
much sampling error is acceptable?
The table below illustrates the sampling
error associated with different sample sizes.
The sampling error should be read as follows:
| If the survey was conducted 100 times,
we could be confident that 95 times
out of the 100, characteristics of the
sample would reflect the characteristics
of the targeted population within plus
or minus the number of percentage points
shown in the table below. |
FOR EXAMPLE: If 2500 people
from the targeted population are surveyed,
and 55% of those 2500 people answer YES
to a given question, we can be confident
that between 53% and 57% of the total population
from which the sample was drawn would have
answered YES, had they been asked the same
question.
Sample
Size |
Sample
Error |
25 |
20.0% |
50 |
14.2% |
100 |
10.0% |
150 |
8.2% |
200 |
7.1% |
250
|
6.4%
|
300 |
5.8% |
400 |
5.0% |
500 |
4.5% |
600 |
4.1% |
800 |
3.5% |
1000 |
3.2% |
1200 |
2.9% |
1500 |
2.6% |
2000 |
2.2% |
2500 |
2.0% |
3000 |
1.8% |
4000 |
1.6% |
5000 |
1.4% |
|
