17
edits
Changes
Jump to navigation
Jump to search
Line 272:
Line 272: + + + + + + + +
Significance of E. Coli Evolution Experiments (view source)
Revision as of 16:39, March 25, 2009
, 16:39, March 25, 2009no edit summary
|0.22
|0.22
|}
|}
Applying the chi-square test to the Blount data violates the parameters under which the chi-square test produces accurate results. The number of 'expected mutants' is below one in all cells, and the chi-square test is not to be used when expected cell numbers are less than five. <ref>http://www.okstate.edu/ag/agedcm4h/academic/aged5980a/5980/newpage28.htm</ref> <ref>http://www.wellesley.edu/Psychology/Psych205/chisquareindep.html</ref><ref>http://www.graphpad.com/www/Book/Choose.htm</ref>
The total number of mutants, four in the first replay and eight in the third, fall below the lower cutoff for the chi-square, which is not felt to produce reliable results for n of less than twenty.<ref>http://faculty.chass.ncsu.edu/garson/PA765/chisq.htm</ref><ref>http://www.basic.northwestern.edu/statguidefiles/gf-dist_ass_viol.html</ref>
Under conditions of low n and low expected cell counts, the chi-square test is too conservative and results in estimated p-values that are too high.<ref>http://www.graphpad.com/www/Book/Choose.htm</ref><ref>http://mysite.du.edu/~jcalvert/econ/chisquar.htm</ref><ref>http://www.basic.northwestern.edu/statguidefiles/gf-dist_ass_viol.html</ref>
Furthermore, the chi-square test assumes that there are no relationships between categories; it should not be used when the same underlying population is tested repeatedly over time, as is done in the Blount experiment.<ref>http://faculty.chass.ncsu.edu/garson/PA765/chisq.htm</ref><ref>http://www.okstate.edu/ag/agedcm4h/academic/aged5980a/5980/newpage28.htm
</ref> Thus the chi-square p-value of 0.004 calculated for replay experiment 2 cannot be interpreted as supporting Blount's conclusions, since this low p-value was derived through the inappropriate application of the test.
==References==
==References==