Quantitative research

Quantitative research answers the “what” or “how many” questions, qualitative research answers the “why” question. It is often worth combining both types of method in a hybrid approach. Reduced to its simplest terms, quantitative research produces charts with figures covering a fairly large number of users. Qualitative research produces knowledge about the real uses and motivations of a smaller number of users.

Sample size is a critical question as soon as you go beyond purely descriptive analysis. To address certain hypotheses, you have to make sure the measurement conditions are right (by setting up a control group, for example), then that you have enough respondents for the statistical power to be sufficient. If you target too few respondents, the very point of surveying them can be called into question, because you will have no way of generalising. For a sample representative of the French population, for instance, you would count on around 1000 respondents, using the quota method.

Yes, through a more classic questionnaire method. The tracking data collected by analytics tools is a good example of quantitative data. Analysing it lets you establish objectively how many people visit a given page, but it gives you no access at all to the goal the user is pursuing (or to the reason why they leave a page). So that data alone is hard to exploit. Standardised questionnaires exist, depending on what you are trying to measure in the UX of a system. And you can of course build bespoke questionnaires depending on what you want to study.

In statistics, you can run tests fairly simply. What matters is being organised. With a little searching, it is easy to find which tests are suited to comparing two groups with each other, for example. So yes, you can carry out all the descriptive analyses without being an expert. As soon as you start doing inferential statistics, you have to be very methodical about checking the conditions of application and the interpretation of the results.