Statistical reliability
Our key aim is to provide statistically reliable information
on which both strategic and tactical decisions can be based
with confidence.
Sample size has an important role to play in ensuring the findings
are representative of the whole population whilst meeting the needs
of clients in terms of reliability of data and, of course, budget.
For quantitative research techniques based on statistical reliability
the calculation of confidence intervals is used to ensure that the
sample is truly representative of the whole. Knowing the reliability
of the study means that findings can be defended on
the basis of logic rather than emotion.
The box below shows the confidence interval with different sample sizes.
By selecting the most appropriate confidence interval (often referred
to as the margin of error) you can be certain that the findings cannot
be dismissed on grounds of dubious methodology.
Population |
Statistical
Reliability |
+/- 5% |
+/-10% |
25 |
24 |
20 |
50 |
44 |
33 |
100 |
80 |
49 |
500 |
217 |
81 |
1000 |
278 |
88 |
5000 |
357 |
94 |
For example: a response rate of 81 out
of a population of 500 is needed for a +/ 10%
confidence interval. If the mean of all the respondents' scores
is 4 on a scale of 1 to 5, the actual range will be between 3.6
and 4.4 (on 95% of occasions). At the more rigorous
+/ 5% (217 responses) the range narrows
to between 3.8 and 4.2. |
