It is seldom possible to survey the entire population. To do this is usually very expensive unless it is a very small and trivial population. Therefore, samples are usually obtained. Greater accuracy is obtained from larger samples. Again, expense is an issue. Also, no matter the size of the original sample, an extremely high participation rate is essential. Most professional sampling organisations require at least a 95% participation rate. In order to achieve this level various sampling techniques are utilised. The simple random sample is very seldom a useful tool. Stratified sampling and cluster sampling are usually employed. Cluster sampling is especially useful since the costs associated for each participant are much lower, and, therefore, a much larger sample can be achieved. This counteracts the inaccuracy associated with this technique. Finally, valid data is essential.