Below are some of the
questions you are likely to ask yourself when planning a
1. What are the main things I need to
think about in designing the survey?
Your survey results may be used
to inform important policy decisions. It is therefore important
to do the survey well! Important things to consider are;
- The survey method to use –
some methods can give better quality data, but may cost
- Representativeness – your
sample of responding people should be a representative
sub-set of the
whole population. Choose appropriate sampling methods and
safeguards to avoid getting a biased sample.
- Getting the scope and
objectives of the survey clearly defined and agreed.
- The questions to ask – try
not to ‘re-invent the wheel’ when designing questions. Use
tried and tested questions where available; and ones that
allow you to compare your area to the national picture or to
2. How big
a sample size should I have?
Sample size usually depends on a
few key factors;
- The ‘accuracy’ you want to
obtain from the survey
- The variability in the
- The need for results for
various sub-groups of the population.
For general population surveys,
many people say that you should choose a sample of 1000
respondents so that you get data to within +/- 3% of the true
population figure. So, for example, if I use a survey of 1000
people in a Trust and find that 30% of them smoke, I can be
fairly sure (95% confident) that the result is within +/- 3% of
this – so, between 27% and 33%.
If however, I want to be able to
look at the proportion of people who smoke in more detail and
find the percentage of each age group, for men and women who
smoke, then, with 5 age groups I may only get 100 men and 100
women in each age group (1000 respondents, 10 cells). This
gives us less accuracy – but in this case is probably still ok.
A general ‘rule of thumb’ is to
have ‘cell sizes’ of a minimum of 30-50.
3. How much will it cost?
The cost of the survey depends on many factors, including;
- The sample size
- The data collection method
- The length of the
- Whether you use external
In general, the larger the
survey, the more the cost…But, a survey
twice as large should cost twice as much – there are some fixed
costs, and some variable costs.
As an example, a postal survey of
1000 residents could cost from £6000 upwards depending on the
number of reminders issued, questionnaire length etc.
How should I choose my sample of who to include
in the survey?
Sampling based on probability
(where you choose people to include from a list or sampling
frame) is best – but may not always be possible. Cost or the
availability of a suitable and accessible sampling frame may be
If your chosen survey method is
to collect data by conducting interviews with residents, then it
is expensive to select individual names at particular addresses,
visit the house, find them in and conduct the interview.
An alternative is sampling
not based on probability. The main thing to remember here
is to fix quota controls to ensure you get a sample that
looks like the population (a representative sample).
What If I can't afford
to carry out a Lifestyle Survey?
Lifestyle surveys are
an important source of data at a local level and it is
important that resources are made available to conduct
them. Partnership working, for example thorough
local Strategic Partnerships, provides a good
opportunity to Share the cost of a survey.
However, if resource cannot be found,
there are a number of options:
- There are a small
number of indicators available at a local level from
the population census, although obviously far fewer
can be obtained from a Lifestyle Survey:
- There are a large number of indicators available at
regional level, again from Government sources: