Power Analysis and Pilot Surveys

Power analysis and Pilot surveys are often overlooked as an essential part of successful survey and monitoring programmes. Power analysis can improve the efficiency of a survey by ensuring that sample sizes appropriate to test the required hypotheses are collected. There is no point in undertaking a long-term monitoring programme in an attempt to detect change of a vulnerable species or habitat if the sample size could not possibly detect the expected change. This approach can be "inverted"; by setting the amount of fieldwork resourch available first, and then testing to see what degree of change could be detected. This might indicate that a different approach – or more resources are required.

Theoretical power analyses are helpful, but they become especially powerful when combined with pilot surveys. In this way, crucial unknown parameters, such as between-site variance, temporal variance and initial starting values can be elucidated. This allows a power analysis to be parameterised much more precisely, so that required sample sizes and minimum detectable change can be predicted more successfully. BioEcoSS has been involved in several pilot surveys (e.g. Winter Mammal Monitoring, Small Mammal Monitoring ) and always strongly recommends this approach to clients, especially for potentially expensive programmes.

The advent of extremely powerful desktop computers and, latterly, parallel cloud computing, allows computer-intensive techniques to be used for power analysis. In particular, BioEcoSS has developed simulations techniques, running millions of permutations to complete sophisticated scenarios, tailored to specific statistical models.