Power Analysis to Determine the Number of Samples Needed for Trend Detection

Background

Power analysis of trends is a useful tool for finding an effective, cost-efficient sampling design. One of the fundamental questions related to aquatic resources monitoring is whether conditions are actually getting worse or better over time. The ability to answer this question depends on three important parameters: the amount of change, the variability in the particular response (both across and within time periods), and the amount of confidence in trend estimates. Power analysis helps researchers analyze the probability that they will correctly identify a trend, given a specified sample size and magnitude of change.

The power analysis tool developed by SCCWRP provides an easy way for researchers to weigh the costs of increasing sampling across sampling periods (e.g., across years and/or within years) against the benefit of increased resolution for detecting changes. Thus, power analysis helps ensure both the scientific value of the study and its cost-effectiveness.
 

Status

This is an ongoing project.

Methods

This statistical tool is based on the statistical model proposed by Fryer and Nicholson (1993) for detecting linear trends over time. The Fryer and Nicholson model assumes random variation among and within time periods. The program uses Microsoft Excel for Windows, and will run on Excel 2000 or better.

The most common use of the program involves observing the test power associated with different numbers of time periods and replicates within time periods. Results are presented as table data and a plot. In addition, the program will also compute power for any combination of the number of time periods, number of within-time period replicates, and percentage change.