Hi,
I have a data, mix of items with different levels and coded as below:
4-levels > (agree-3, neutral-2, Disagree -1)
3-levels > (Yes-3, NotSure-2, No-1)
2-level > (Yes, No) ?
So how should I code 2-level items ( Binary) by keeping in view the coding for 4-level and 3-level items, since I am planning to mix all of them for same analysis (Factor analysis/Scale development).
Should I code them as
(1) Yes = 1, No = 0 ?
OR
(2) Yes = 2, No = 1 ?
Thanks
Often, dummy coding (0 and 1) is useful because it leads to an interpretable zero point (in your case, the score of zero would mean group/response "no"). Having a meaningful zero point can be advantageous for certain types of statistical analyses such as regression, in which means and intercepts may be relevant. Having a meaningful zero point can make regression results more interpretable, for example, in moderated regression with interaction terms.
For a factor analysis, it may not matter because we often ignore means and intercepts in factor analysis altogether and focus instead on covariances and correlations for which the type of coding is often arbitrary/irrelevant.
Dear Saeed Shah ,Measurement with the Binary data type can also be done like this in the following example:
1 = like very much, 2 = like, 3 = do not like (three scales can be classified to a scale of 1), while 4 = dislike more, 5 = strongly dislike (can be classified on a scale of 0), so 1 is a scale that tends to agree on statements, events, opinions, etc.), while 0 is more likely to disagree or reject the statement. A measurement scale that has only two answers, is called a nominal scale (eg 1 = male, 2 = female) but binary data cannot be equated with a nominal scale, because the binary data scale must contain the number 0 (zero).
so from me, and sorry if this is not pleasing, thank you, Best Regards.