In adapting a questionnaire with the options "agree", "uncertain", "disagree", my questions is if it would be preferable to include the options of "Partially agree" and "partially disagree".
What are the benefits and the losses of each model of answers?
It all depends on the potential respondents, so if you have homogenous sample (e.g. students share a lot of similarities like age, educational level etc..) then its less likely to use 3. By all means, i never used 3, usually go either for 5 or 7
Without citing specific papers, I'd add that more response options leads to a more "sensitive" scale, and this is a good thing. More option choices in the item produces a measure with more variance, which is therefore better able to show statistical relationships with other measures. Conversely, "too many" response options leads to annoyance in your respondents, because it's cognitively taxing to make very fine-grained judgements about their beliefs.
It depends on the issue/ topic of inquiry. But as a general rule if the respondents(expected) are mature and have stable thought etc, perhaps 4 point scale in Asian culture and five in European and American cultures is OK. But for young and immature three point is OK.
The scale in a questionnaire (3 or 5) options for answers depends on the nature of study, the researcher, and the purpose of questionnaire.
For me, I believe 5 options is better than 3, because it gives more options, and the answers can be more accurate and precise. 3 options is very limited with choices.
It all depends on the potential respondents, so if you have homogenous sample (e.g. students share a lot of similarities like age, educational level etc..) then its less likely to use 3. By all means, i never used 3, usually go either for 5 or 7
It probably may depend on the type of result needed to be reached at. But for the sake of discussion, if more options are given then spread out responses will be offered and that takes away a focused and targeted investigation and make good conclusion to the intended goal. If we offer 5 distractors for a question of only one solution, then we are opening spaces of more students to select wrong answers while narrowing the number of distractors, we narrow the search space and therefore the possibility of more students to choose the right answer.
I prefer the questionnaire to contain 5 options (strongly agree, agree,no idea, disagree, strongly disagree). The first indicates firm support & the last indicates firm rejection, as you know. One may agree with some reluctance & one may have no idea about answering a certain question. After all, we are looking for an honest scientific objective response from those we are asking.
I strongly vote against odd-numbers (!) - the problem you will get is that most people vote in the middle if they have no real optinion about it, or less time whatever.
If you take 4 possibilities there must be a decision, if I vote more for or against it. what you need is a trend where the answer is going.
So at least 4 or even 6 possibilities :-)
Dear All,
I personally do not like multiple choice tests. Unfortunately – depending on the subject – one can get relatively good scores with limited knowledge but with logical thinking. Another trouble is that cheating among students is quite easy when using tests like those. Thus, more options may increase the reliability of a test.
As a survey respondent, I can tell you that a 5 point is significantly better than a 3 point question. There are very *few* instances where someone holds a view that isn't shaded - given a choice of absolute a, neutral, and absolute b - I will most often be neutral, but only because I don't believe 100% in absolute a or absolute b. As a result, I belive that the survey results become skewed.
I piloted a questionnaire using 11 points (0-10 scale) in an attempt to increase responsiveness. The results came out with too many ceiling effects. So we went down to 5 points, which worked better in that regard. We then played with different alternatives, such as trichotomising and dichotomising the responses. Dichotomising is useful for binomial logistic regression analysis. Problem then is to decide which way to merge the middle/neutral option. As mentioned above, it might be better to just have even number of points, to avoid people selecting neutral. We ended up merging the neutral up in some items and down in others, depending on whether the question were phrased negatively or positively. With a positive response set, you would normally merge it to the not agree side. If you don't know or don't want to take a stance, then you don't agree. I do think these issues are quite complex and it's easy to get skewed results, so important with careful thinking in the construction phase and (maybe several rounds of) pilot testing.
Five alternatives Likert type with neutral option " absolutely agree to absolutely disagree" preferred more. The construct validity shows more stable in factor analysis.
Assume you have 5 persons to ask and 5 options provided for the survey. Each option is selected with only one person, as a one to one correspondence. Then what is the conclusion the inquirer makes ? unless indeed the questionnaire has a purpose of knowing whether respondents might offer a spread out view on the subject under investigation if they definitively do not know the answer or else if the question is just to simply collect views and how spread the responses can be.
But for exam questions as Dear Andras said, although all types of questions are pedagogically recommended, as each type asks particular skills of learning, open questions are the good types of problems that require, aside from knowledge retained, deep understanding, imagination, analysis and logical reasoning to come up to a could be solution.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire