 | Hello everybody,
I have some questions regarding a specific experimental design and
wonder if you could help me.
Background:
The purpose of the experiement is to find out what factors influence
the selection of stories by a user as relevant. There will be three
levels of relevance (first choice, second choice, third choice), where
the first is most relevant, ect. I provide the user with a set of
stories and a set of situations. I will ask the user to provide a
ranked list of his/her first, second, third most relevant piece of
information based on the situation. The user will select three
stories from the pool of stories for each situation. The story pool
will therefore be reused for each situation.
I want to go for a full-factorial experiment where each factor has a
set of qualitative levels, according to the following list:
time = {morning, noon, evening}
location={placeX, placeY, placeZ}
interest = {topicA, topicB}
Therefore, I will have a 3*3*2 experiment with 18 different situations
for which the user needs to provide 18 judgements by providing three
ranked stories according to each of the situations.
Questions:
1) How many stories do I need to present to allow for staticially
significant results given the fact that the user must select 3 stories
in ranked order for EACH situation?
2) What methods do I have to use to compare interactions in the
judgements given the fact that I get three ranked results and not, as
described in my book, a single direct value which can be compared?
Thank you very much in advance,
KarlK
|
|
