Sample size should be as close to 100% of the target work group (function, business unit) as possible.
We are measuring cultural behaviours, so a large sample of the work group is necessary.
In practice we find about 80% completion rate is good.
The smaller the number of people in the target group to be surveyed, the closer to 100% is required. If a work group we are reporting on has 10 people, we need 8 to complete. If it has 100, then 70 is sufficient.
We like to have work groups on which we report to have at least 10 people.
Large groups can also be sampled in way that provides high levels of confidence and low error rates.