Hi guys,
I have two two constructs that are not opossing to each other from the content perspective (so I can't use the same scale) but logical reasoning reveals that you can't aggree to both constructs contemporarily. In order to reduce amount of time that is necessary to answer the questionnaire I would like to post an introducing question that can be answered either "yes" or "no". If particpant is answering "yes" he will only fill out the one construct while the other has to be filled out in case that the answer is "no". (Thus, the constructs are parallelized in the questionnaire). To keep my sample big enough and on base of the logical exclusiveness of both constructs I would like to score "don't aggree" automatically on each item of the construct that the participant is not answering. Is this fine with PLS?
Otherwise the particpant will also have to answer the questions of other construct, nevertheless that he (from the logical perspective) will always choose "I strongly disagree").
Hope you can help me, and hope that the automatical scoring is fine with PLS because otherwise I would fear that particpants break up the questionnaire.
thanks a lot
Patrick
automatical scoring of constructs that exclude each other
PLS only knows/cares about having data--you can enter "don't agree" or whatever else you want for the other construct.
But it sure doesn't sound right, conceptually, unless you have something similar to my example below.
All respondents either ate dinner at home or went out; if they stay home they answer questions about preparation and cooking time, if they go out they answer questions about travel time and waiting time to get seated and served.
In this example the construct might be "time to wait before eating"--measured differently by the two different groups.
IF you had the same number of indicators in each situation, I could see PERHAPS using one model for everyone.
But note that you don't have two constructs, it's just one. I can't off-hand think of a situation with two truly different constructs where this would make sense. If you care to share more details perhaps I and others could be of more use.
But it sure doesn't sound right, conceptually, unless you have something similar to my example below.
All respondents either ate dinner at home or went out; if they stay home they answer questions about preparation and cooking time, if they go out they answer questions about travel time and waiting time to get seated and served.
In this example the construct might be "time to wait before eating"--measured differently by the two different groups.
IF you had the same number of indicators in each situation, I could see PERHAPS using one model for everyone.
But note that you don't have two constructs, it's just one. I can't off-hand think of a situation with two truly different constructs where this would make sense. If you care to share more details perhaps I and others could be of more use.
John J. Sailors, PhD
Associate Professor of Marketing
The University of St. Thomas
Opus College of Business
Minneapolis, MN
Associate Professor of Marketing
The University of St. Thomas
Opus College of Business
Minneapolis, MN