Unformatted text preview: If being right about the difference is more important, use higher significance levels, If missing a difference that really exists is more important to avoid, use might accept a lower significant levels. If the difference in your sample is not statistically significant, you conclude that you cannot tell whether there is actually a difference in the real population There may be one, but the power of your test was too weak to find it It is important to keep in mind that we impose a high standard on significance Typer 1 error ( overconfidence): thinking there is a difference between means where is non Type 2 error ( humility): thinking there is none, even though there is. 1....
View Full Document
- Spring '08
- Statistics, Null hypothesis, Statistical hypothesis testing, Statistical significance, Statistical power