The use of the telephone has been found to be one of the most inexpensive, quick and efficient ways of surveying respondents. The ubiquity of telephone ownership as well as the use of unlisted numbers are factors that must, however, be considered as part of the sampling frame, even in North America, where the number of households with phones approaches 100%. Telephone surveys also allow for random sampling, allowing for the extrapolation of characteristics from the sample to the population as a whole.

There tends to be less interviewer bias than in interview surveys, especially if the interviewers are trained and supervised to ensure consistent interview administration. The absence of face-to-face contact can also be an advantage since respondents may be somewhat more inclined to provide sensitive information. Further, some people are reluctant to be approached by strangers, whether at their home or in a more public location, which can be overcome by the more impersonal use of the telephone.

On the other hand, telephone surveys are also known to have a number of limitations. The length of the survey has to be kept relatively short to less than 15 minutes as longer interviews can result in refusal to participate or premature termination of the call. The questions themselves must also be kept quite short and the response options simple, since there can be no visual aids such as a cue card.

The increasing use of voice mail and answering machines has made phone surveys more difficult and more costly to undertake. Calls that go answered, receive a busy signal or reach a machine, require callbacks. Usually, eligible respondents will be contacted a pre-determined number of times, before they are abandoned in favour of someone else. The potential for response bias must be considered, however, when discussing the results of a study that relied on the telephone.

The sample for a telephone survey can be chosen by selecting respondents

• from the telephone directory, e.g. by calling every 100th name
• through random-digit dialling (RDD) where the last four digits of a telephone number are chosen randomly for each telephone exchange or prefix (i.e. first three numbers), or
• the use of a table of random numbers.

The importance of randomization is discussed under probability sampling.

Two practices that are increasing in popularity and that raise considerable ethical issues, since the respondents are misled into believing that they are participating in research, are:

1. the survey sell (also known as sugging), whereby products or services are sold, and
2. the raising of funds for charity (also knows as frogging).

Any survey technique that requires the respondent to complete the questionnaire him/herself is referred to as a self-administered survey. The most common ways of distributing these surveys are through the use of mail, fax, newspapers/magazines, and increasingly the internet, or through the place of purchase of a good or service (hotel, restaurant, store). They can also be distributed in person, for instance as part of an intercept survey. Depending on the method of survey administration, there are a number of sampling frame considerations, such as who can or cannot be reached by fax or internet, or whether there is a sample bias.

A considerable advantage of the self-administered survey is the potential anonymity of the respondent, which can lead to more truthful or valid responses. Also, the questionnaire can be filled out at the convenience of the respondent. Since there is no interviewer, interviewer error or bias is eliminated. The cost of reaching a geographically dispersed sample is more reasonable for most forms of self-administered surveys than for personal or telephone surveys, although mail surveys are not necessarily cheap.

In most forms of self-administered surveys, there is no control over who actually fills out the questionnaire. Also, the respondent may very well read part or all of the questionnaire before filling it out, thus potentially biasing his/her responses. However, one of the most important disadvantages of self-administered surveys is their low response rate. Depending upon the method of administration chosen, a combination of the following can help in improving the response rate:

1. A well written covering letter of appeal, personalized to the extent possible, that stresses why the study is important and why the particular respondent should fill in the questionnaire.
2. If respondents are interested in the topic and/or the sponsoring organization, they are more likely to participate in the survey; these aspects should be stressed in the covering letter
3. Ensuring confidentiality and/or anonymity, and providing the name and contact number of the lead researcher and/or research sponsor should the respondent wish to verify the legitimacy of the survey or have specific questions
4. Providing a due date that is reasonable but not too far off and sending or phoning at least one reminder (sometimes with another survey, in case the original one has been misplaced)
5. Follow-up with non-respondents
6. Providing a postage paid envelope or reply card
7. Providing an incentive, particularly monetary, even if only a token
8. A well designed, visually appealing questionnaire
9. A shorter questionnaire, where the wording of questions has been carefully considered. For instance, it might start with questions of interest to the respondent, while all questions and instructions are clear and straight forward
10. An envelope that is eye-catching, personalized and does not resemble junk mail
11. Advance notification, either by phone or mail, of the survey and its intent

Face-to-face interviews are a direct communication, primary research collection technique. If relatively unstructured but in-depth, they tend to be considered as part of qualitative research. When administered as an intercept survey or door-to-door, they are usually part of quantitative research.

The opportunity for feedback to the respondent is a distinct advantage in personal interviews. Not only is there the opportunity to reassure the respondent should s/he be reluctant to participate, but the interviewer can also clarify certain instructions or questions. The interviewer also has the opportunity to probe answers by asking the respondent to clarify or expand on a specific response. The interviewer can also supplement answers by recording his/her own observations, for instance there is no need to ask the respondent’s gender or the time of day/place where the interview took place.

The length of interview or its complexity can both be much greater than in other survey techniques. At the same time, the researcher is assured that the responses are actually provided by the person intended, and that no questions are skipped. Referred to as item non-response, it is far less likely to occur in personal interviews than in telephone or self-administered surveys. Another distinct advantage of this technique is that props or visual aid can be used. It is not uncommon, for instance, to provide a written response alternatives where these are complex or very numerous. Also, new products or concepts can be demonstrated as part of the interview.

Personal interviews provide significant scope for interviewer error or bias. Whether it is the tone of voice, the way a question is rephrased when clarified or even the gender and appearance of the interviewer, all have been shown to potentially influence the respondent’s answer. It is therefore important that interviewers are well trained and that a certain amount of control is exercised over them to ensure proper handling of the interview process. This makes the interview survey one of the most costly survey methods.

Although the response rate for interviews tends to be higher than for other types of surveys, the refusal rate for intercept survey is higher than for door-to-door surveys. Whereas the demographic characteristics of respondents tend to relatively consistent in a geographically restricted area covered by door-to-door surveys, intercept surveys may provide access to a much more diversified group of respondents from different geographic areas. However, that does not mean that the respondents in intercept surveys are necessarily representative of the general population. This can be controlled to a certain extent by setting quota sampling. However, intercept surveys are convenience samples and reliability  levels can therefore not be calculated. Door-to-door interviews introduce different types of bias, since some people may be away from home while others may be reluctant to talk to strangers. They can also exclude respondents who live in multiple-dwelling units, or where there are security systems limiting access.