Re: number of alternatives?

[Pierre Abbat <phma@phma.optus.nu>]

On Wednesday 09 September 2009 11:23:32 Jorge Llambías wrote:
> Personally, I would prefer using the same number throughout. Four
> choices makes it easier to construct the questions, five makes the
> questions more discriminating (it makes guessing a little more
> difficult), but on balance I think it's better to have more questions
> with four choices than fewer with five choices.

Based on my experience taking tests, I prefer five. But I think that should 
just be a maximum, and questions can have anywhere from two (true and false) 
to five choices.

> Which could instead be something like:
>
> zo ci'onrai ma dunli lo ka valsi
> A. zo cikrai
> B. zo cimrai
> C. zo cinrai
> D. zo citrai

With three words, there are five ways all, some, or none can be the same. So 
we could ask:
1. cnoblo
2. conlo'i
3. coiblo
A. ro di'u dunsi'u
B. le pamoi cu frica le remoi .e le cimoi
C. le remoi cu frica le cimoi .e le pamoi
D. le cimoi cu frica le pamoi .e le remoi
E. ro di'u ficysi'u

I think that questions should be organized into sections, so that instructions 
can be given once per section. So we have:

Section
{Instructions
 (Total number of questions to ask in this section)
	Question group
	{Story, or the like
	 (Number of questions in this group to ask)
		Question
		Question
		Question
		...
	}
	Question group
	{Story, or the like
	 (Number of questions in this group to ask)
		Question
		Question
		Question
		...
	}
	...
}

You pick at random a subset of the question groups whose numbers of questions 
add up to the section's number of questions, then pick at random questions 
from each section group.

Pierre