NLPES Site Map

Question of the Month

October-December 2004:

WHEN CONDUCTING SURVEYS, DO YOU PROMISE CONFIDENTIALITY TO POTENTIAL RESPONDENTS? (If so, in what circumstances do you do so? And how do you word your pledge of confidentiality? If you don't promise confidentiality, why don't you do this?)

Rick Riggs, Kansas

We've always included language in our surveys telling respondents that their survey document will become part of the public workpapers after the audit is released. It's rare but not unheard of that an agency official or a reporter will come look at the workpapers, and we wanted people to be clear that we were NOT promising them confidentiality and their decision on how to respond to the survey should take that into account.

Lately we've come to see this as a problem, because it can't help but have a chilling effect on what respondents are willing to share with us. In the last couple of months, we've been talking with our legislative committee about changing state law to allow us to make all survey documents confidential.

The bill that will be introduced when the Legislature convenes specifies that all survey documents are confidential. If the bill passes, we think it would go a long way toward raising respondents' comfort level.

Sylvia Hensley, California

As a general rule, the California Bureau of State Audits cannot promise confidentiality to potential respondents. All information that we use in support of an audit report is subject to the public records act. Consequently, unless the information gathered through a survey is protected by law or is not used in support of the audit report, it would be available to the public.

Most often the type of information we gather through surveys is not sensitive so our inability to promise confidentiality is not an issue. However, when requesting information that people may be reluctant to provide without the promise of anonymity, we may use other methods to protect the respondent's identity. For example, the survey document might make the respondent's name and contact information optional, allowing each respondent to decide whether to provide this information. Because this may interfere with our ability to follow up with non-respondents and respondents alike, it is not be a viable option when a high response rate or follow up is required. Another method we have employed is to use a code to identify respondents instead of their names. In such cases, we might place an alpha or numeric code on the return envelop provided with the survey document. The survey document itself neither asks for nor includes the respondent's name. The coding on the return envelope, allows us to track the responses and follow up as necessary. Although the mailing list and the completed surveys would be public documents, the documents linking the two are kept in the administrative file, which are not available to the public.

Byron Brown, Florida

We do not promise confidentiality unless there are special circumstances that require or necessitate that the survey be confidential, such as when we are working with information that the law classifies as confidential. We don't usually promise confidentiality because it is a legal term and our work-papers are not classified as confidential by our statutes. The potential exists that we may share information from our work-papers with either the agency or with the legislature. We tell survey respondents that we do not include names or identifiers in our reports, and that we generally present survey results in aggregate so that individual responses are not identifiable.

Perry Simpson, South Carolina

As far as confidentiality goes, our records are protected under our confidentiality law but individuals such as legislators could get access to these records. Therefore, we generally don’t put any identifying information on our surveys. This helps to protect the respondent’s confidentiality. If we don’t know who they are, then we can’t share that information.

Trish Bishop, Virginia

In general, we pledge confidentiality to our survey respondents. In most circumstances, the terms of the confidentiality will be clearly expressed on the front of the survey and in a cover letter that usually accompanies a survey. Confidentiality usually means that survey responses will be reported in aggregate and that no identifiers will be used. We often use the following language or similar language on our surveys: “All results will be confidential, and results/comments will not be attributed to a specific agency or individual.”

Jill Jensen, Connecticut

We do promise confidentiality to our survey respondents. A couple of examples of the language we typically include in the cover letters sent with our survey instruments follow:

As part of the study, each housing authority director is asked to complete the enclosed questionnaire and return it in the envelope provided by July 26, 2004. Some of the questions seek general information and others ask for opinion. The survey results will be compiled so that identification of any responding individual will be impossible.

As part of our study, we are asking the directors of special education for local and regional school districts in the state to complete the enclosed questionnaire and return it to us in the envelope provided by October 15, 2000. The code number that appears on the survey form will be used by committee staff to follow up on questionnaires that are not returned by the due date or that need clarification. Please rest assured your individual response will remain anonymous.

Joel Alter, Minnesota

Our office has authority to classify survey responses as nonpublic data if, in our judgment, the information provided would not have been obtained otherwise. We promise confidentiality in cases where we are asking for information or opinions that we think might be sensitive. For example, we used the following language in one recent survey: “Under Minnesota Statutes §3.979, subd. 3(c), the Office of the Legislative Auditor will classify survey responses as private information that will be protected from public disclosure.”

In many surveys we conduct, we do not promise confidentiality. To help provide accountability for the conclusions we reach in our studies, we try to avoid classifying supporting documentation as nonpublic if this does not seem necessary. (Not that we have many requests from reporters or citizens to see our workpapers…)

In cases where we have decided not to promise confidentiality to survey respondents, I noticed that our practices have differed regarding what we tell potential respondents. On the one hand, a recent survey by our office had the following language: “Upon completion of this project, all information, including questionnaire responses, will be public data (as defined by Minnesota Statutes (2002) §13.03, subd. 1) and available to the public upon request.” On the other hand, we sometimes have not explicitly told survey recipients in the cover letter or survey document that their responses will be part of our public workpapers. In such cases, perhaps we’ve assumed that people should expect their responses to be part of our public records unless they’re told otherwise—just as we don’t routinely start every interview with a statement to the interviewee that their comments to us will eventually become part of our public records unless they request confidentiality. As Rick Riggs from Kansas commented above, I also worry that a statement in each survey stating that responses will be public could have a “chilling effect” on what the survey recipients tell us.

By the way, even in cases where our office does not specifically promise confidentiality to survey respondents, we reserve the right to classify individual responses as nonpublic if the person requests confidentiality, or if we make a judgment that the individual would not have provided the information if he or she thought that it would become public information.

One final comment: Even in cases where we promise confidentiality to the persons we survey, we try to obtain identifying information from the respondents. This allows us to (1) send follow-up letters to persons who have not yet responded to the survey, and (2) contact the respondent if we need clarification regarding their responses.