Random survey questions assigned to each respondent?

Question

Is there a name for a Polling/Surveying scheme that takes a pool of survey questions but randomly picks one or few of those questions for each respondent, instead of asking every participant all the questions?

There's random-sample voting - but that asks a random sample of the population all of the questions. I want to burden each respondent with just a few of the pool of available questions. I suspect that I can get disproportionately better response/completion rates by asking fewer questions.

I guess part of the value of a survey is that we learn relationships between questions. Technically, it seems that a survey of 20 independent questions would be 20 independent surveys. But 20 pick 2 seems valuable, given enough respondents.

1. Is there a name for that method so I can look it up?
2. Is such a method actually a bad idea that shouldn't be used?

Answer 1

I believe you are looking for missingness by design/planned missingness: http://www.peterlugtig.com/2013/01/planned-missingness.html.

From http://marco-morales.com/documents/wp/PMMI_pub_140920.pdf:

Planned Missingness is the name commonly given to survey designs in which the same target population is queried to answer different sets of questions, thus generating a controlled item non-response.