Look, we get it, sampling may not be the most exciting part of a research project. But when it comes to the success of and confidence in your research initiatives, having a solid sampling plan makes all the difference. At GutCheck, many of our clients don’t necessarily want to “know how the sausage is made” when it comes to filling their studies with respondents. However, what they do want is confidence that their vendors will provide the most representative or balanced samples possible, so they can confidently base decisions on the results. For those of you who want to take a deeper dive into the why and how around GutCheck’s sampling strategy, we are here to help!
Because there is no fully enumerated universe of online users from which people can be randomly selected, samples from surveys conducted online fall into the category of non-probability samples—because we are in some sense surveying those who are accessible. But this doesn’t mean that we are just surveying whoever clicks first on an email or notification that a survey is available. Rather, we ensure sufficient respondents within specific quotas around various demographic variables, based on census data, enter our surveys. This allows us to maintain our speed and agility, while getting our clients the answers they need quickly, without sacrificing the quality of our research. This also allows our clients to confidently project the results of their surveys among a small sample of their audience, across the entire population of their audience.
For qualitative studies, where we work with smaller sample sizes and try not to “size” an audience, we will use a largely targeted sample, targeting everything possible from demographics to even brands used. With qualitative work, we focus more on providing respondents who give quality, in-depth responses, rather than try to make these small groups representative of the overall population.
For quantitative work, where we work with larger sample sizes and try to “size” an audience, we need to ensure a balanced and relatively representative sample enters our studies so that a representative sample of our actual audiences comes out in the final data. To accomplish this, we first ask our sample providers to supply only those respondents who match age, gender, or other demographics for the audience of interest. Then, we use click/start balancing quotas based on census targets to ensure that the representative sample of respondents will begin the survey. Through qualification screening questions, we can confidently tell you that the demographics of those who qualify and complete your survey will be representative of your audience of interest on the census balancing variables. You will then be able to confidently project the survey results and audience across the wider population. This process of funneling a representative sample of a population into a survey, and the representative audience subset that comes out in the final data, helps minimize or even eliminate sample bias and sampling error.
Sampling error occurs in statistical analysis when a sample is unrepresentative. This can happen for a variety of reasons, including not understanding the audience that should be surveyed, non-response to survey invites, and improper balancing within the demographics of the audience of interest. For example, a non-representative skew toward females or toward a certain age group versus census targets is evidence of sample bias.
If you take anything away from this blog, it should be this: The design and implementation of a proper sampling plan is paramount. Without a sample that accurately represents your audience, you should be very careful in making assumptions that your survey results will ring true when projecting to a wider population such as your customer base. Further, making important business decisions based on unrepresentative data should be avoided as it can steer you in the wrong direction. So, choose your research suppliers wisely and make sure they have solid sampling practices in place before you begin your studies!
To learn more about GutCheck’s sampling practices and how we are delivering agile audience intelligence to help inform confident decisions at any stage in the development process, talk to our research experts.