Hi Lina,

Sorry, it is no possible. The inverse (5 or 6 points Likert Scale to dichotomic scale) is possible, because we reduce the range of answers.

I suggest you take new data with the dichotomic scale modify.

Or

Use some statistic technic that is possible the analysis with polichoric data. For exemple,if you will use some causal relationship model (Structural Equation Model), especially, Partial Least Square models. The dichotomic variables can be used.

If you want send me more information of you research for i help you

hope you help

Best regards

Dirceu

Hi Lina,

Sorry, it is no possible. The inverse (5 or 6 points Likert Scale to dichotomic scale) is possible, because we reduce the range of answers.

I suggest you take new data with the dichotomic scale modify.

Or

Use some statistic technic that is possible the analysis with polichoric data. For exemple,if you will use some causal relationship model (Structural Equation Model), especially, Partial Least Square models. The dichotomic variables can be used.

If you want send me more information of you research for i help you

hope you help

Best regards

Dirceu

Prof. Dirceu da Silva

I will do , thank you so much

Lina

Hi Lina,

I don’t think this a good idea. What kind of analysis are you considering? There are a lot of possibilities, depending of your research questions: logit analysis, correspondence analysis, SEM…

The questionnaire has been extensively applied and studied in its psychometric properties. You should follow the scoring scheme construction in order to obtain the values from the different scales it provides. Why do you need to convert dichotomous questions (answers) to 5 or 6 point Likert scales? Maybe you don’t want to use the questionnaire but to study something else about it. Anyway, I don’t recommend you to proceed with this conversion.

thanks Prof.Jorge

I want to determine the relationship between these variables, and whether the cognitive factor play a mediating role or not. i will use SPSS for analysis and correlation and regression tests will be applied.

my research is quantitative cross-sectional.

Lina,

I agree with Pr. Dirceu da Silva's suggestion.

best regards.

thanks a lot Prof.Jorge

I agree that converting a yes/no question into a 5- or 6-category Likert response scale is not advisable, as it would mean all previous psychometric work on the validity and reliability of the original question would no longer be applicable. One additional note: if you have data from some yes/no questions and some 5-category Likert questions that you want to add together to create a single multi-question index, you can score the yes/no questions 5/1 (or 1/5) instead of 1/0 (or 1/2), so that the ranges of all questions in the index are the same (and therefore, the variances are similar). Otherwise, the Likert questions will implicitly be weighted much more heavily than the yes/no questions in determining the total score on the index.