Researchers recognise the value of the following suggestions:
Keep questions as short as possible. Ask one question at a time.
Pilot questions before finalising the questionnaire to ensure children understand what you are asking and that the response options fit their answers.
Ask children to respond to affirmative not negative statements (disagreeing with a negatively phrased statement is a cognitively complex task).
Always balance the number of positive (e.g. agree, agree a lot) and negative response options (e.g. disagree, disagree a lot).
It can put children at their ease if you preface a statement with an introduction that says, "Some children agree with this, and others do not. What do you think?".
Always separate out the scale midpoint (e.g. 'partly agree, partly disagree') from the 'don't know' response, and ensure the latter is always recorded.
For attitudinal questions, think carefully if you wish children to answer on behalf of children in general or themselves in particular.
Reverse the direction of some questions to reduce response bias: for example, if saying 'yes' to some questions means you like the internet and saying 'yes' to others means you don't like it, one may minimise the effect of children's tendency to agree with statements presented to them.
If item lists are provided as response options (e.g. lists of media used, lists of activities) then always end with an 'other' option. If you have the resources to hand code these, then ask the child to specify what the 'other' is.
Pitfalls to avoid
The pitfalls are implicit in the above advice, and in essence are the same for children as for adults. If a survey questionnaire is too complex or confusing, uses difficult words, has inappropriate response options, doesn't provide a 'don't know', 'other', or 'I don't want to say' response option where needed, asks leading questions etc, you may not know this from the survey administration until you come to analyse the answers. A 'don't know', 'other', or 'I don't want to say' response option may increase the data quality, as it will reduce the amount of default (or misleading) selections. If the survey is administered as a pen-and-paper survey, children will write rude answers if they don't like or don't understand the questions! Large amounts of missing data also provide a clue that you've got something wrong.
Questions to consider
Is this a topic that can be well addressed using a survey? Do you know the kinds of answers that children are likely to provide? Have you piloted the survey and do you know how long it takes? For young children, will there be someone present to help them or answer their questions? Should this topic instead be addressed using qualitative methods? If you ask open-ended questions, are you sure you have the resources to code their responses?
Fowler, F. J. (1993). Survey Research Methods (2nd ed.). Newbury Park; London; New Delhi: Sage Publications.
Frey, J. H. (1989). Survey Research by Telephone (2nd ed.). Newbury Park; London; New Delhi: Sage Publications.
Examples of good attitudinal questions
From 'UK Children Go Online', questions to low or non-users included: "How much do you agree or disagree" that - "I'm missing out by not using the internet and email (more)"; "I can find out all I need from books"; "The internet helps people get ahead in life"; "I sometimes feel left out when my friends talk about the internet"; "The internet makes it easier to keep in touch with people"; "I would like to use the internet more in the future". Response options: Agree a lot/Agree a little/Neither agree or disagree/Disagree a little/Disagree a lot/Don't know.
From Internet 'Parents & Teens 2004 Survey': Do you agree or disagree or don't know (NB no scale midpoint provided) that "If a child isn't using the internet by the time they start school, they will fall behind their peers"; "Most teens are not careful enough about the information they give out about themselves online"; "Teens who use the internet to stay in touch with their friends have better social lives than teens who don't use the internet to do this"; "Teens waste a lot of time online, when they could be doing more important things"; "The internet helps teens do better in school"; "Too many teens today use the internet to cheat on their schoolwork"; "Most teens do things online that they wouldn't want their parents to know about".
The 2005 National Center for Missing and Exploited Children survey (Finkelhor, 2006) asked a simple question: "How important is the internet in your life, on a scale of 1 to 5, with 1 being not at all important and 5 being extremely important?". (Range 1-5) Don't know/Not sure/Refused/Not ascertainable/Not applicable.
(Sonia Livingstone and Panayiota Tsatsou, UK)
A researcher's mistake
In my 'mobile phone' questionnaire, I realised that it is not a good idea to have too many sub-questions under the same question as this confuses respondents. For example, a question about 'use of camera on the mobile' was subdivided to no less than 14 subsequent questions which sometimes confused the respondents. The lesson to be learned is to have fewer questions, and not too many sub-questions. Each sub-question has to be worded so as not to leave any space for misinterpretations or variable answers.
(Liza Tsaliki, Greece)