I want to study the awareness and adherence level of physicians in Yemen to hypertension treatment guidelines based on JNC8.
I designed the questionnaire with three possible answers (agree, disagree, and i don't know). Does it ok to have such questionnaire?
yes, it can be. But in your study design, the subjects are well educated, i.e. physicians. Hence, it is better to use five options (strongly agree, agree, undecided or don't know, disagree, strongly disagree).
Hello Abdulsalam,
Yes, you can design with three answers and more too if it is needed. like given in the previous answers (Mild, Moderate, severe, extremely severe)
It depends on how you want to score the overall scale that you will create from the total set of items. If you score "Don't Know" as missing data, you will not be able to include those items when you calculate the overall scores.
In my opinion design of question depends upon objective your research, that includes statistical method(technique) to be employed.
If your aim is just to know whether or not physicians are aware of hypertension treatment guidelines and whether or not they adhere to them then you may use Y/N/IDK scale. But keep in mind the point by David above. You can then use frequency distribution for understanding your data and chi-square for comparisons, if so required.
And, if your aim is to test a model, then data should be collected on 5 point scale or 7 point scale (you can decide the scale point wording depending upon your question statements). Mukhtar has suggested one solution. In this approach you can then create scale by merging more than one questions (also termed as items) to have an abstract variable.
I would have selected second option.
Dear Mukhtar Ansari, Sakthivel Govindaraj, David L Morgan and Ali Farooq
thank you so much, it is appreciated.
Yes, a lot depends on your goal for the measure. If it is to determine level of knowledge, then you could score each item as correct (+1), don't know (0), and incorrect (-1).
But if you want to create some kind of scale that can be used in more complex analyses, then a 5- or 7-point LIkert score would be more appropriate.
Does it OK to design a questionnaire with three possible answers ?
I concurred with the comments shared by David Morgan & Ali Farooq - whether 3-point or wider range like 5-point or 7-point depending on your research / measurement goal. In the event that your goal is to test your research model, then 5 or 7-point is better because:
You can also find out the advantages & disadvantages of narrower e.g. 3-point scale vs wider range e.g. 7-point scale in the following RG link:
https://www.researchgate.net/post/Which_Likert_scale_should_I_prefer_that_of_1-4_1-5_or_1-7