How can I get more statistical power from my data?

Sample size and power calculations can be used to compute the power of a study (proposed or completed) to detect a difference (or effect) of a defined size. Often you'll be disappointed that you can't get more power while still using a reasonable sample size. Consider this list of approaches to increase power (abridged from Bausell and Li, referenced below ).

The best approach to getting more power is to collect more, or higher quality, data by:

• Increasing sample size. If you collect more data, you'll have more power.
• Increasing sample size for the group that is cheaper (or less risky). If you can't add more subjects to one group because it is too expensive, too risky, or too rare, add subjects to the other group.
• Reduce the standard deviation of the values (when comparing means) by using a more homogeneous group of subjects, or by improving the laboratory techniques.

You can also increase power, by making some compromises:

• Increase your choice for alpha. Alpha is the threshold P value below which you deem the results "statistically significant". While this is traditionally set at 0.05, you can choose another value. If you raise alpha, say to 0.10, you'll increase the power of the study to find a real difference while also increasing the chance of falsely finding a "significant" difference.
• Decide you only care about a larger difference or effect size. All studies have higher power to detect a large difference than a small one.