There seem to be too standards as to how to collect concourses:
do it as a random sample, just gather the universe of statements as indiscrimiantely and broadly as you can (this is hard enough already because there is no known universe list of statements as in sampling people as per #229 (there is no list of all possible things to be said, this thing does however exist for people, such as voting registrars etc).
you can do sort of a saturation sample, where you add items only if they are new, or better and you edit items so that they are easier to understand or whatever (like I do per a deliberative standard).
The question is however, whether this line is as sharply drawn as one might think it is.
In all cases to we need items that reasonably invite affect as per the condition of instruction.
It would, for example, not make sense to invite people to tell me how much they agree/disagree with (an extreme case):
A is B
AND
A is not B
simply because this is a contradiction in terms.
asfijhgiuhasd
Also would not make much sense as an item, at least it's hard to think what condition of instruction would make sense here.
So from that vantage point, you always have to edit to have items that correspond to your condition of instruction.
But that's not really random sampling.
You don't - in random sampling of people – edit out people who don't conform to your theory.
There seem to be too standards as to how to collect concourses:
The question is however, whether this line is as sharply drawn as one might think it is.
In all cases to we need items that reasonably invite affect as per the condition of instruction.
It would, for example, not make sense to invite people to tell me how much they agree/disagree with (an extreme case):
simply because this is a contradiction in terms.
Also would not make much sense as an item, at least it's hard to think what condition of instruction would make sense here.
So from that vantage point, you always have to edit to have items that correspond to your condition of instruction.
But that's not really random sampling.
You don't - in random sampling of people – edit out people who don't conform to your theory.