Nesting of interviewers for mixed effects regression

[Marcel Schmelzer]

Say I have 5 interviewers and 10 participants.

Each of the 5 interviewers interview 2 participants. There is no overlap (i.e. each participant is only interviewed by one interviewer).

Each interviewer interviews each participant twice.

Is it correct to say that interviewers are nested within participants? Then use a regression similar to:

mixed y x1 || participant: || interviewer:

Many thanks!

[Clyde Schechter]

No, you have it backwards. The participants are nested in interviewers, and recurrent interviews ("each interviewer interviews each participant twice") are nested in participants. The appropriate code would be

mixed y x1 || interviewer: || participant:

[Marcel Schmelzer]

Thanking you very much for your input Clyde. To help my clarity of understanding, is this because of the lack of overlap? If each participant was by more than one interviewer would it be as I originally was writing Code:
mixed y x1 || participant: || interviewer:

[Clyde Schechter]

No, it's not because of the lack of overlap. If each participant were interviewed by more than one interviewer, you would not have nesting at all--that would be a multiple membership model or crossed effects, and the code for that would be different still.

The direction of the nesting comes from the fact that each participant is interviewed by only one interviewer, whereas each interviewer interviews multiple participants. That defines the nesting as participant within interviewer. The nesting would be interviewers within participants (and the code you propose in #1 and #3 would be correct) if each participant were interviewed by multiple interviewers and each interviewer only interviewed one participant.