Why use this guide?
Choosing the right sample for your research can help you ensure your research results are reliable and accurate.
As a general rule of thumb, the more responses you receive, the more reliable your results will be. Use this guide to understand how many people you need to hear from for your survey research to be representative.
What is included in this guide?
This guide includes five sections:
- Understanding sample size
- How to calculate your sample size
- Who to approach in your space
- Make sure you have a balanced approach
- Common types of bias to look out for.
Understanding sample size
Sample size is the subset of the population that is invited. It is a sample, because you’ll not collect data from every person in the population (a census).
Response rate is the proportion of responses that you receive.
Your sample size and response rate will likely be different. This may be due to a number of reasons – e.g. you don’t have an attendee’s email address, an attendee may choose not to respond to your email. This means that your survey responses will only represent part of the group of people (the Population) you want to hear from.
Calculating a sample size and target response rate in advance means that you can design an approach and select an appropriate incentive. Once you hit your target response rate, you can be confident that your results are representative of your population.
How to calculate sample size
There are a number of online sample size calculators and a quick web search will give you plenty of options to choose from. If you’re using an online survey tool, it may have a built-in calculator.
Online sample size calculators will ask you to enter three things:
- Your population size
- Your confidence level
- Your margin of error
Follow the instructions and you’ll have your sample size!
Who to approach in your space
As well as the sample size you’ll need to know your expected number of visitors. Below is an example of how to work out who to approach.
You work at a gallery and you have written a short survey. You want to approach gallery attendees with the survey at the end of their visit – while the attendee’s experience is fresh in their mind.
To work out who to approach, you first need to work out the number of responses needed per day. You can do this by making a quick calculation (outlined in the table below):
|Minimum sample size||400|
|Divided by Number of fieldwork days||25|
|Equals Number of responses needed per day||16|
To achieve a minimum of 400 responses a minimum of 16 responses is required each day over a 25-day period.
The next step is to calculate who to approach. You can do this by making a quick calculation (outlined in the table below):
|Number of expected visitors per day||96|
|Divided by Number of responses needed per day||16|
|Equals Nth person to approach||Every 6th person|
This means you need to approach every sixth person. In this example, the gallery expects most attendees to visit in pairs. In this case you should then approach every third group (every sixth person). To keep your count simple (approaching every sixth person), allow only one person from each group to respond to your survey. This example is based on one person who will be conducting the research.
Make sure you have a balanced approach
To make sure you have a balanced approach, you’ll need to consider three things:
- The number of fieldwork days you need
- The time of day you’re collecting information
- The days of the week you’re collecting information.
The days and times you choose should be generally representative of a normal week, month or exhibition period. For example, if you’re open six days a week, from 10am until 5pm, you would want to approach people on a range of different days (such as Monday, Wednesday and Saturday) and a range of different times (such as 10am-1.30pm one day and 1.30pm-5pm on another).
Common types of bias to look out for
Sampling frame error
Before you contact your database, check to see if it is representative of your population. For example, if you’re looking to hear from only those who live within 50 kilometres of your space, check your database’s postcodes. If there aren’t people in your database with postcodes within 50 kilometres of you, think of alternative ways to reach these people, such as through local community groups or networks. Unfortunately, if you haven’t already collected this sort of information from your database, you may not know if your database is representative of your population until after you collect their responses.
Check to see if your respondents differ from non-respondents. For example, you may find that you have only heard from one group of your database – only parents with children because your incentive was free tickets to a kids shows. This sort of thing will bias your results and lead to non-response errors.
Offering an incentive that is a cash equivalent and that appeals to as many people as possible will help to avoid non-response errors.