Types of analysis

Design-based analysis

• In this approach randomness is induced by the random selection of samples or the assignment of samples to a subset
• Choice of a statistical model can be used for model-based inference

Model-based analysis

In this approach randomness is because of the innate randomness in the measurements (in the case of surveys – these are the responses)

---

Transcript

So two major types of data analysis: design-based analysis, in this case, the approach is that the randomness is induced by the random selection of the participants or the assignment of the participants to a given subset.  That's where the randomness came in.  And then you're going to choose a statistical model to do model-based inference. So you're going to say, "Okay! Here is my model, let us fit it to my data" But the randomness came from the selection of the people, [hence] you believe that there's a model there so [it is] design-based analysis.  In the model-based analysis, the idea is the randomness occurs because of the innate randomness in the measurements themselves.  So we have a model, we applied, for instance, a set of surveys to a population, we got a set of responses, we believe that there is randomness in there - but the randomness is inherent in the measurements - we couldn't remove the randomness. There's going to be randomness no matter what we do.