'Self-completed questionnaires' are surveys that respondents complete for themselves. The mailed questionnaire is probably the most common form, although self-completed questionnaires may be distributed personally (e.g. to clients by practitioners) and, increasingly, via e-mail. This tool of research is frequently employed in social care research, especially by practitioners, and, therefore, advice relating to its use is given below.

They should be carefully designed to prevent respondents from completing them incorrectly, giving clear directions about what information is wanted. Have a clear idea of the information you require from the research and make sure that you are concise and focused in terms of quality and quantity of questions asked. Practitioners thinking about conducting a survey using questionnaires should ensure that they are not unnecessarily long, since this can be off-putting. Questionnaires designed for self-completion are likely to have few open questions (those that allow participants to give more detailed information, rather than having to answer to two or more tick box responses). Beware of too many open questions, requiring respondents to write in detail, which are prone to stop people from completing a study.

Mailed questionnaires are quick to administer in comparison to face to face or telephone interviews, which require more time (and probably manpower) to carry out. However, be warned, several weeks may pass before a sizeable number of mailed questionnaires are returned, and it is often necessary to send out follow-up letters to people failing to return original forms. Second and third mailouts to the same people are not uncommon, resources permitting.

Mailed questionnaires dispense with the difficulty of researcher influence or bias (conscious or unconscious). For example, people may respond differently to a male or female interviewer and are likely to give socially desirable responses when interviewed or telephoned, regardless of the gender of the person asking questions. However, the lack of direct contact with a member of the research team, in the case of mailed surveys, means that the respondent is unable to clarify what questions mean. Such an approach also denies researchers the opportunity to explore interesting areas in more detail. In addition, there is a greater chance that a respondent will miss out parts of a mailed survey (e.g. they may inadvertently skip over parts, they may get bored, or they may feel that a question is inappropriate or of little interest to them).

Respondents (and researchers) may find mailed questionnaires more convenient because they can be completed at a time and place that suits. One consequence, however, may be that since researchers are unable to control the order in which questions are answered, questions may not be truly independent of one another. In addition, it may be impossible to ensure that the intended person has replied (for example, a busy manager may pass a questionnaire on to one of her/his staff to complete, or a relative may complete a survey intended to seek the opinions of a service user).

The target respondent should shape the way in which the questionnaire is structured and written. Those who have poor or no literary skills, or who do not have English as a first language, may find it impossible to respond to mailed questionnaires. However, for many types of research, reaching these members of society is of great importance and every effort needs to be made to ensure that their voice is heard (e.g. producing versions of the survey in relevant foreign languages).

A big downfall of mailed questionnaires as a research tool is that they are prone to poor response rates and, unless researchers are able to show that there is little difference in terms of demographics and other important variables between responders and non-responders, it is possible that results will be biased (e.g. only those who are very satisfied or dissatisfied with a service might respond, with those who have a more intermediate view failing to do so). If a written questionnaire is delivered on a more personal basis (e.g. after a course, after an assessment) the response rate is likely to be better.

Various techniques have been proposed to help improve response rates from self-completed questionnaires, some of which are described below:

• Include a good covering letter explaining what the research is about, why it is important and how the individual was selected, with a clear assurance of anonymity and confidentiality. If possible, it may help to address each letter personally to the respondent.
• The inclusion of a stamped-addressed envelope is essential. Respondents will not want to be out of pocket for helping with your research!
• Keep a list of all those who respond so that non-responders can be sent a follow-up letter and questionnaire, tactfully urging them to complete and return it as soon as possible. Reassert the importance of the research and their involvement in it and, once again, make clear that they will not be identified in the research. This should be done about three weeks after the original was posted. Then, about three weeks after sending the first reminder, it may be necessary to distribute a second reminder, with another copy of the questionnaire enclosed, in case the first one has been lost. If response rate is still very poor after this, it may be necessary to send out further reminders. But do not overdo it! People may have valid reasons for not wishing to respond.
• Keeping questionnaires clear and concise should encourage more people to return them. Printing on both sides of paper can make a questionnaire appear less bulky, although careful instructions need to be given to ensure that respondents do not miss the questions on the backs of pages. Limiting open-ended questions to the bare minimum is also likely to improve response rates (although topics of relevance and interest to respondents may produce a good response rate regardless of the length).
• Start with basic, easy to answer questions (e.g. age, gender, job title) as this can help to ease the respondent into the flow of the survey.
• Well presented surveys, written in an adequately sized (11 or 12 points), easy to read font (e.g. Ariel, Verdana, Times New Roman) have a better chance of being completed than ones that have not been well formatted. Printing on brightly coloured paper can also help to stop the survey from being lost among other forms and letters that people receive on a daily basis.
• Provide incentives for returning questionnaires "“ this may make people keener to respond. If possible, each person could be paid or receive a small gift as a token of thanks for their time. Alternatively, a draw for a prize that is appropriate for the targeted sample and the topic under research (e.g. a book token for a survey of children's reading habits) could be arranged, involving all those people who returned their questionnaire.
• Consider special needs of those receiving a questionnaire. For example, you might like to take advice about how best to print information so it is accessible to individuals with visual impairments.