We've written before about the reasons for including demographic questions (including questions about length of employment, department, etc.) in a questionnaire. There are at least three instances when you should consider omitting them. These are:

1. When the sample size is so small that including the questions provides no added value. Perhaps the most important benefit you derive from including demographic questions is the ability to compare the data from different demographic groups. When the sample size is very small (certainly fewer than 50 responses would qualify as being too small), you really can't slice the data and see statistically significant differences. Thus, you derive no benefit by including the questions, yet you risk losing responses (or obtaining less honest responses) because in a small organization, employees are more apt to believe that they can be identified with such questions.
2. When there have been issues about management trust in an organization. If you know that there have been problems regarding the employees' trust in management, it may be best to abstain from asking questions which employees might fear can be used to identify them.
3. When the survey is too long. If you have added considerable length to a survey in the course of asking traditional organization-performance questions, a good place to cut additional length is the demographic section. However, before you do this, you might want to take a second look at the questionnaire to determine whether cuts can be made elsewhere. If neither of the first two conditions mentioned above exist, the demographic questions will enable you to have an increased depth of analysis.