The results of a survey are said to be representative when they make it possible to draw the general opinion of a reference population without obtaining all the answers from this population. In other words, representative results are usable results that reflect the opinions and feelings of coworkers.
In this article:
📌 General principle
To determine the representativeness of the results it is interesting to keep those 2 elements in mind:
- The diversity of the target populations: the respondents should represent the diversity of the population surveyed, not just a subgroup;
- The participation rate: depending on the number of people surveyed, there is a minimum threshold of responses to be collected in order for the results to be reliable.
To simplify your analyses, you can consider that generally in Supermood surveys, the diversity of your populations is preserved since the survey is sent to everyone and not to a sample.
The results can therefore be considered sufficiently representative for :
- 50 employees surveyed, with a participation rate of 86%.
- 500 employees surveyed, with a participation rate of 35%.
- 1,000 employees surveyed, with a participation rate of 21%.
- 10,000 employees surveyed, with a participation rate of 3%.
As you can see, the smaller the number of respondents, the more essential it is to have a high participation rate in order to have results that are representative of the population.
|
💡 Our recommendation Always remember to analyze the participation rate with the number of employees surveyed. A participation rate of 50% for 500 employees surveyed is not interpreted in the same way as for a population of 10 people. |
📏 Measuring representativeness
Make sure that respondents are representative of the diversity of the population surveyed
Within a population, you can observe different subgroups according to demographic (gender, age group etc) or organizational (department, service etc) characteristics. The opinions and feelings of coworkers can vary from one subgroup to another.
It is therefore essential to ask the question: "Are the coworkers who responded to the survey representative of the population as a whole? “
Suppose you have a group of 500 people, 50% men and 50% women, and you ask a question about diversity and inclusion. With a participation rate of 50%, if you assume that women and men responded equally, your results should be representative.
However, if you had a participation rate of 100% for men and 0% for women, the survey results would not be representative of the entire population: they would not be representative.
|
💡 Tip Whenever there are significant gaps in participation, the Supermood survey report allows you to identify the relevant dimensions of analysis and the groups with the lowest and/or highest participation. |
If no disparities in participation are identified by Supermood, the diversity of your population is well represented by the results and you can calculate a representativeness rate.
If so, keep in mind that some sub-populations may be under-represented, even if the overall participation rate is good.
How reliable are the results obtained?
If your respondents are representative of the population surveyed, you can calculate a representativeness rate and define your margins of error and confidence levels for the results obtained.
| Margin of error | 10 % | 10 % | 10 % | 5 % | 5 % | 5 % | 1 % |
| Confidence interval | 90 % | 95 % | 99 % | 95 % | 99 % | 99 % | 99 % |
| Surveyed population. | Minimum participation rate to obtain representative results of the surveyed population. | ||||||
| 10 | 90 % | 100 % | 100 % | 100 % | 100 % | 100 % | 100 % |
| 25 | 76 % | 84 % | 88 % | 92 % | 96 % | 100 % | 100 % |
| 50 | 58 % | 68 % | 78 % | 86 % | 90 % | 94 % | 100 % |
| 100 | 41 % | 50 % | 63 % | 74 % | 80 % | 88 % | 100 % |
| 500 | 12 % | 16 % | 25 % | 35 % | 44 % | 57 % | 97 % |
| 1000 | 6 % | 9 % | 14 % | 21 % | 28 % | 40 % | 94 % |
| 5000 | 1 % | 2 % | 3 % | 5 % | 7 % | 12 % | 77 % |
| 10000 | 1 % | 1 % | 2 % | 3 % | 4 % | 6 % | 62 % |
Let's say that for a given survey, the results have a confidence level of 95% and a margin of error of 5%. This means that if you were to repeat the exact same survey in the same context and at the same time, the results for your population as a whole would correspond to +/- 5% of the results obtained in 95% of cases.
The scientific community most frequently recommends a margin of error of 5% and a confidence level of 95%.
The margin of error does not take into account all sources of error such as the exclusion of a group that did not answer or potential bias in the formulation of the question. To help you formulate your questions properly, see our article: The structure of a closed-ended question.
🧐 What if the results are not representative?
It is always better to have the opinion of a few than none. So keep acting for the coworkers who speak out!
- A good idea can influence the commitment and satisfaction of the whole group.
- As long as you measure your impact, you will have an idea of how the population as a whole has been affected.
- Communicating and acting on the results proves the company's involvement and therefore the interest in participating in surveys to express its opinion.
Encourage teams to participate and try to improve the participation rate in your next surveys. If you need tips to boost participation, find our tips to:
Comments
0 comments
Article is closed for comments.