4.3. Issues of anonymity and ethical approval

Considering the issue of anonymity versus confidentiality is important before a questionnaire is disseminated. In an ‘anonymous’ type questionnaire, the subject is totally unknown to the survey organiser, while in a ‘confidential questionnaire’ all the data is known to the survey organiser, but kept confidential. Box 3 gives an example of how anonymity can be preserved when postal surveying is used, using experience from surveys in Scotland.


Box 3. Example: Preserving anonymity in a postal survey in Scotland.

The membership records of the Scottish Beekeepers’ Association (SBA) provide a well-organised sampling frame of the target population. The help of the SBA Membership Convener was obtained. He is the only person with full access to those records. He was asked not to supply the survey organisers with the full records (which he would not have been permitted to do in any case), but only to supply the list of “Short Reference Numbers”, each of which  uniquely identifies one of the members, along with the associated postal code. Before  supplying that list, he was asked to remove from the list those ineligible to participate in the survey, including for example members not resident in Scotland, institutional members (such as libraries), and those who had declared themselves unwilling to participate in surveys. (The opportunity to opt out of surveys is available to new members when joining the SBA, and the opportunity was given to all existing members of the SBA to opt out via a short article published in the SBA’s regular publication for members, prior to the first survey using the SBA records for sample selection).

The postal codes were abbreviated in the list supplied, so that while preserving the broad geographical location of each potential survey participant, it was not possible to identify any particular address. The postal codes were used to assign each member on the list to a particular geographical region in Scotland, so that the sample selected could be stratified on a geographical basis. This was done by dividing the country into a number of regions related to the administrative areas used by the SBA, in order both to give greater precision in estimation and to ensure greater geographical coverage in sampling and therefore hopefully a more representative sample.

Then a stratified sample of the agreed size was selected from the list, using a sampling function available in the R software for objective random selection. Each questionnaire sent out was put into an envelope on which was written a questionnaire number provided by the survey organisers, also written on the questionnaire in the envelope. These envelopes were then sent to the Membership Convener along with a key file linking the questionnaire numbers to the corresponding “Short Reference Numbers”. This enabled the Membership Convener to print the appropriate address on each envelope and to mail out the questionnaires, without the organisers knowing the identity of the selected members.

In fact, the majority of participants in the Scottish surveys have willingly provided personal contact details as part of their questionnaire return. (For example, in the survey in 2011,  85% out of 94 respondents did so; there was a 47% response rate).

 With the increase in use of email / webmail, using these means of communication make it virtually impossible to guarantee total anonymity, as the respondent’s name – or at least the email address – is automatically included in their reply, although satisfactory survey software packages include the option of suppressing from the recorded responses all means of identifying individual respondents. While the perceived possibility of lack of anonymity may raise levels of non-response or compromise the validity of responses to any sensitive questions in an email questionnaire, the ease of access to a worldwide population of beekeepers, the low administration costs and its unobtrusiveness to respondents generally outweigh this negative effect. It is also a simple matter to issue reminders by email.

However, it is important that the level of confidentiality of the questionnaire is clearly outlined to participants. Hence, the covering letter with the original questionnaire should clearly state that a reminder will be forthcoming if no response is received. The availability of this option to issue reminders is important since research indicates that reminders increase the response rate (Campbell and Waters, 1990; see also the Scottish example below on reminders and incentives). Furthermore, email software allows the dispatcher of the questionnaire the option of notification when the recipient has opened the message.

The use of questionnaires raises the question of personal liberty and ethics and, because of this, many research institutes/universities require the survey organiser to acquire ethical approval prior to disseminating questionnaires. Since questionnaires related to honey bee research are primarily concerned with generic rather than personal information, it may be possible to acquire multi-annual ethical approval in advance, thus allowing the annual dissemination of the questionnaire. However, this will be specific to different institutes and thus clarification on the ethical requirements should be sought locally before a questionnaire is disseminated.