Organizational Diagnostics

Ensuring Confidentiality

Return to Articles listing

A survey process that ensures employees of their confidentiality is essential to a successful survey project.  It preserves the integrity of the data because respondents feel safe to express their true opinions.  It also protects respondents from the risk of retaliation from others in the organization. Unfortunately "humanity" and "common sense" are normally distributed variables:  In a company with 100 managers, a few of them will think that abuse and humiliation are appropriate ways to defend their performance.

Violations of confidentiality can have a lasting effect in corporate cultures, undermining the credibility and effectiveness of companies' use of surveys as assessment tools.  We have run across companies who have had to cancel surveys because of previous violations of confidentiality, even though the original violation was 5 years ago!

Make confidentiality a clear policy

The cover page of the survey should highlight a clear statement of corporate intent to preserve confidentiality.  This pledge should state that surveys will never be used to identify respondents individually.  The survey should clearly state that data summaries for small groups -- we suggest anything fewer than 10 respondents is too small -- will not be provided to anyone in the company.  Respondents in groups with fewer than 10 respondents are still included in data summaries for the organization overall.  Surveys should also have no identifying characteristics such as identification numbers, color codes, or mailing labels.  We have had respondents use rubber gloves for fear that the paper might be treated to retain fingerprints!

The demographics section of the survey should be at the end.

Typically a survey will ask the respondent to identify their job type, site, department, gender, or other demographic information.  When demographic questions occur at the end of a survey, respondents can answer them already knowing what types of information they have provided the company.

At the beginning of the demographics section, respondents should receive an explanation about the purpose of the questions: to verify that all levels of the organization are adequately represented and to enable identification of significant differences in the opinions among different GROUPS of employees, not employees individually.  Additionally, respondents should be advised to skip any questions if they feel more comfortable doing so.  The respondents' comfort and confidence in their confidentiality is more important than complete data.

Collection procedures should be designed to ensure confidentiality and published for all.

If surveys are collected in-house, overt procedures should be in place to assure respondents that no one in the company will ever see their individual responses.  The contents of collection boxes should be inaccessible to anyone within the company.  Trusted proxies or witnesses should be on hand to ensure that surveys are delivered directly into the hands of analysts.

Additionally, respondents should be given an option to return the survey directly to the organization conducting the analysis.  If they complete the survey on paper, an addressed and stamped envelope should be included.  If the survey is web based, respondents should have an option to print a hard copy to complete and mail directly.

Be forewarned that email surveys on sensitive topics are likely to arouse suspicion about the likelihood of the email system being hacked into by concerned managers.  Whether it is likely or not, the mere prospect of it is enough to produce a chilling effect on candor and participation.

Open-ended comments should be retyped and edited to remove the occasional identifying information.

Retyping removes the chances of comments being recognized through distinctive handwriting.  We recommend having one trusted point of contact within the company review the comments.  They are more likely to recognize names and events that, if published, would identify the author or cause damaging speculation about who they author might be.  For example, references to unfair treatment in a performance review may inadvertently identify the author, since only the respondent and the manager mentioned would understand the comment.

Article Text