What percentage of the population must participate in a survey for validity? What are the risks if only 50% of the population responds to the survey?
We don t usually survey everyone. We will survey a department, or a project team, or those involved in a cross-company alliance. We survey anywhere from 20 to 800 employees in a typical situation. We try to get everyone to participate because getting a good cross-section in your distribution list is important. If you re only getting 50% of the population participating, and particularly if the key players are missing then we keep pressing for more data. We prefer to have participation of 85% or higher of the group we have decided to survey. Again, it depends on what you re surveying. If you are mapping a subject matter expert network, full participation is not so critical. If you are doing a survey where you are looking for confirmed ties [A says he/she works with B, and B says he/she works with A] then participation will have to be high or you will see many unconfirmed ties.
