Statistical versus Practical Significance

If you're reading this in order, you're about to finish Social Statistics 3600.

The big problem that entry level statistics student face is the issue of statistical verses practical significance.

In social and the behavioral sciences, statistical significance occurs when we have P < .05.  David Moore, the author of your text, says that we place too much emphasis on the P value and not enough emphasis on confidence intervals.   He is correct.  If we merely note that research is statistically significant, we are likely to make an incorrect decision from the research and make a harmful decision for or about a client system.

As a safeguard, Moore recommends the following:

  1. always graph your findings before testing them.

  2. exam the confidence interval (it estimates the size of the effect/change).

  3. seek out outliers (these can destroy a significant result).

  4. note and assess influential observations in regression analysis.