How Many People Should I Survey – SamplingY
ou need to select survey respondents carefully because their responses will have a significant impact on your survey findings.
There will be two important terms which will need clarity before you choose respondents – population and sample.
Population is the entire set of people who you want to understand.
For example, if you wish to understand, “How many people use smartphone in the UAE, your population would be people living in the UAE [Around 9.5 million, as of 2016].
Your population many be fewer people if you wish to understand, “How many people use smartphone in your family and friends.”
Sample is the number of selected respondents who will participate in the survey.
How many people from the population are enough for your survey?
Since including the entire population in your survey is not realistic in many cases, experts have given guidelines to pick acceptable numbers of respondents. See Table 3.1 below.
For example, if your entire population is 1000 people, you would need 278 respondents at 5% margin of error and 95% confidence level.
Margin of error
Margin of error is the plus and minus figure to allows little flexibility when you compile results of your survey. For example, if your analysis informs that 80% of your family members use a smartphone, 5% margin of error will mean that 75% (80-5) to 85% (80+5) of your family members use smartphones.
From the above example, this tells that you are 95% sure that if same survey is conducted again you will get the same results i.e. “80% of your family members use a smartphone.”
Not Everyone Responds
When you send out your survey, not everyone responds.
The percentage of people who actually fill out your survey is known as the “response rate.”
Response rates vary widely depending on a number of factors such as the relationship with your target audience, survey length and complexity, incentives, and topic of your survey.
For online surveys, in which there is no prior relationship with recipients, a response rate of between 20¬ – 30% is considered to be highly successful. To be on the safe side, keep 10 – 15% response rate in mind, if you haven’t survey your population before.
So, if you need 100 respondents from a large population you need to send it to more than 100 people.
Here is one example:
Respondents needed = 100
Expected response rate = 10%
You need to send your survey to 100 / 10% which means 1000 people.
Did you know! Sample size does not change much for populations larger than 20,000.
Sampling methods for surveys
Sampling methods are classified as either probability or nonprobability. In probability samples, each member of the population has an equal chance of being selected. Probability methods include random sampling, systematic sampling, and stratified sampling. In nonprobability sampling, members are selected from the population in some nonrandom manner. These include convenience sampling, quota sampling, and snowball sampling.
a. Random Sampling
The purest form of sampling under the probability approach, random sampling provides equal chances of being picked for each member of the target population.
b. Systematic Sampling
In systematic sampling, every Nth name is selected from the list of the members of the target population. For instance, the sample will include the participants listed in every 10th from the list. That means the 10th, 20th, 30th and so on will be selected to become the members of the sample group.
c. Stratified Sampling
Stratified sampling involves the use of “stratum”, or a subset of the target population wherein the members possess one or more common attribute. Examples of stratum include mothers, fathers, students, teachers, females, males, etc.
Tip: To generate random numbers, use Excel’s rand functions =rand() or =randbetween()
d. Convenience Sampling
This non-probability sampling method is used when there are only a few available members of the target population who can become the participants in the survey.
e. Quota Sampling
Another non-probability method, quota sampling also identifies strata like stratified sampling, but it also uses a convenience sampling approach as the researcher will be the one to choose the necessary number of participants per stratum.
f. Snowball sampling
It is a special nonprobability method used when the desired sample characteristic is rare. It may be extremely difficult or cost prohibitive to locate respondents in these situations. Snowball sampling relies on referrals from initial subjects to generate additional subjects. While this technique can dramatically lower search costs, it comes at the expense of introducing bias because the technique itself reduces the likelihood that the sample will represent a good cross section from the population.