Small handbook for Social Research: Sampling.
When we design researches that investigate people’s preferences, one of the first issues to be addressed is the sampling. If we want to investigate the preferences of a group of people, we need to conduct our research on a representative sub-group, and numerically manageable, of the group as a whole.
As a result, the reliability of the results that we get is closely linked to the representativeness of our sample.
It is evident, therefore, that the sampling is a thorny issue that must be addressed with strategies, necessary to prevent a “contamination” of the data.
When we perform our research online and we appeal to our friends on Facebook or Twitter to participate, perhaps by completing an online questionnaire, we’re not sure of rowing in favor of the reliability of the results, nor we are helping our work of researchers. Especially when the number of responses that we get is not high, the lack of representativeness of our sample is likely to make us take blunders dangerous.
Common sense suggests that it is better to have a limited number of responses and / or investments in our research from people who are sure to be part of the group that we want to investigate, rather than thousands of responses from people that we are not a sure guarantee good index of representativeness.
Some trick: if you decide to collect subscriptions through social networks, at least try to profile the recipients of your invitation using hashtags relevant. If you want to investigate consumer preferences of your product, choose tools to send an invitation less “public” and favorite tools like email, text messages, direct messages. This will help you to better control the exposure of your message and put a stop to the slightest indistinct participation.
Just get a small budget available to create a mini-campaign to join through Facebook or Twitter, that offer profiling tools available to the public broadly that are accurate enough. We make an example: if my sample are part of single women under 35 in Rome, with highly educatation and interested in Fashion and Beauty, the Facebbok Ads profiling tools, for example, can be really helpful and accurate enough. I repeat: do not need big budgets. I do not care to get a response from the general public, but from only one part of it that is, however, truly representative.
Another point that is close to my heart: the anonymity of the answers. Proliferate online questionnaires and studies that ask us to login, to place our email address, etc. Not a good idea. The answers may be distorted resulting from the desire of the interviewee does not look bad, not to expose themselves more or less publicly, to adhere to the views most common and shared, or, on the other hand, they could “exaggerate” in answers for the sake of argument or be caustic / defiant / etc. only for matters of public exposure.
The anonymity of the responses protect us from these kinds of distortions (otherwise lawful and understandable) and has the merit of removing from our research those lovers of “at any cost controversy”.
In the next article — how to get quality data — we will deepen this very type of problem related to the distortions and try to provide some guidance to help us to get reliable data.