Again, Dillman’s principles reign supreme here as well. However, there are some recommendations that need to be made specifically for children and adolescents that will be shown below.

Response Options in General

1. Use an equal number of positive and negative options. If you have a 4-point scale of “Strongly Disagree” to “Strongly Agree” then the middle two categories should have one disagree item and one agree item, not two disagree items or two agree items. If there is a middle to the response options, you should be able to cut them in half and have both halves look identical yet opposite in valence.
2. State both sides of the attitude in question stems. Again, if you have a disagree-to-agree scale, you should provide that in the question itself. You should not ask “Do you agree…” but rather “Do you disagree or agree…”. Furthermore, you should have the response options in the question match the order presented in the response options. Like a number scale, I always put the disagree options on the left and agree options on the right.
3. Provide mutually exclusive and mutually exhaustive options. Mutually exclusive means that you should only be able to choose one Mutually exhaustive means that there is an option that fits your situation. For instance, when asking “What is your age?” The following response options would be inappropriate: “10-15, 15-20, 20-25, 25-30.” If you were 15, 20, or 25 then you would have two options you could choose (meaning it is not mutually exclusive). If you were below the age of 10 or older than 30, you would have no options (meaning it is not mutually exhaustive). You can add an “other” category to cover your bases. Also, if you know the ages of the respondents, you can limit categories; for instance, if you are surveying high school students, you can safely assume nearly all of your respondents are between 12 and 18 (though there may be younger/older students in special circumstances).
4. Avoid check-all-that-apply question formats. If you really need to know if all the options apply to your participants, ask each option as a separate question. Otherwise, you may deal with primacy (focusing on the first set of options), recency (focusing on the last set of options), and satisficing effects.

Recommendations for Children and Adolescents

1. Use four or fewer response options for children and seven or fewer response options for adolescents. The fewer the response options, the easier it will be for your participants. Even seven might be too high for adolescents.
2. Completely label response options. Sometimes, evaluators like to provide the end options and leave the middle options blank. Some argue this makes the scale more interval rather than ordinal, which has important implications for data analysis. However, this can be difficult to interpret for children and adolescents. Semantic differentials are one notable scale type that does not have middle options, but research suggests that children do not do well with these types of questions (Helwig & Avitable, 2004).
3. Visual response options can be useful, particularly for very young children. These can include using frowny/smiley faces to indicate feelings towards something. This has been used often in illiterate populations who cannot read well and are instead surveyed through in-person techniques rather than pencil-and-paper or online techniques.

What about even or odd response options?

Inevitably, someone always asks me this question. Should there be a “Neither disagree nor agree” option in the middle of “Disagree” and “Agree?” Unfortunately, my answer is always it depends. Primarily, it depends on whether a middle option makes theoretical sense and whether you want to allow participants to choose that option. Sometimes, you want to force the respondent to choose one side or the other, in which case you should choose an even number of response options.