I was told that I have to change a dichotomously-measured variable into an ordinal one. Is such a thing usual or not? Authorized or not? How can I justify such a change in my research paper? That variable I mean to measure has often been assessed dichotomously, and given to the fact that it has not been yet measured in an ordinal way, I'm in doubt whether it is possible or not.
Hi
Have you collected data or not? If not, then it is okay to change your scale. But if you have collected data, then it wont be advisable. Different studies adopt different methodologies, and that in itself, is a gap in research. That can justify the change in scale.
Hope that helps.
Are you talking about a single variable, or a set of variables that make up a scale? If it is the latter, then there are advantages to having several ordinal variables to combine into a scale.
If you are working with just one variable, then a lot depends on whether it is an independent or dependent variable, as well as the kind of statistics that you intend to do in analyzing that variable.
Can you give us more details?
Thank you dear Daglous and David.
Dear David, there are actually 6 scales in my questionnaire, in which one of the mediating constructs has always been measured dichotomously (yes/no) in the literature, at leat I haven't been able to find an ordinal scale of that variable, and the other five variables are measured using an ordinal scale. As you might have guessed I intend to analyze the data using SEM, more precisely covariance-based SEM.
To sum up, one variable, which is a mediating variable, is measured with binary responses, and the rest with 7-point Likert scale.
Could you please kindly provide me with a reference which elaborates on what you said in your first paragraph?
Thank you so much for your help and information David. I'm waiting for your kind support.
I'm not sure what you are asking for, because combining several ordinal variables into a scale is a very standard practice in SEM. Any book that provides an introduction to SEM should cover the idea that combining several variables into a scale will increase the reliability of that scale in comparison to any of the individual variables.
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers