Revision Date: 2005-09-06
| Neither one is very good. In general, the number of degrees of freedom is based on how much data you have. The more data, the more degrees of freedom you tend to have, and the more confident you feel that the result reflects accurately the information about the data being evaluated (a more powerful test). (For the Chi-square test, the number of significant bars you have in your histogram is the key.) However, the confidence level is (or can be) the same for both situations. For example, in the two cases you cited:
Chi-square calculated = 0.44 with 1 degree of freedom could be compared to a Chi-square table value of 0.00+ at the 95% confidence level.
Chi-square calculated = 4.01 with 2 degrees of freedom could be compared to a Chi-square table value of 0.10 at the 95% confidence level.
In both cases, since Chi-square calculated > Chi-square table value, you would reject the hypothesis that the data can be considered normal and be 95% confident that you were correct in that decision.
Gordon Constable, Ph.D. | |
