SkipCaptcha
212
edits
Changes
Jump to navigation
Jump to search
Line 73:
Line 73: + + + +
no edit summary
:::Although the P-value for the third experiment was not equal to or less than the arbitrary 0.05, but was instead 0.0823, this is still not consistent with the null hypothesis. Consider the alternative perspective- if you're 91.77% confident in deviation from the null hypothesis (based on P=0.0823), then conversely you could say that you are 8.23% confident that the null hypothesis was verified. To have been consistent with the null hypothesis, the P-value for the third experiment would have been in the ballpark of P=0.95. Thus a P-value of 0.0823 is nowhere near supporting the alternative hypothesis.[[User:Kallium|Kallium]] 09:01, 30 July 2008 (EDT)
:::Although the P-value for the third experiment was not equal to or less than the arbitrary 0.05, but was instead 0.0823, this is still not consistent with the null hypothesis. Consider the alternative perspective- if you're 91.77% confident in deviation from the null hypothesis (based on P=0.0823), then conversely you could say that you are 8.23% confident that the null hypothesis was verified. To have been consistent with the null hypothesis, the P-value for the third experiment would have been in the ballpark of P=0.95. Thus a P-value of 0.0823 is nowhere near supporting the alternative hypothesis.[[User:Kallium|Kallium]] 09:01, 30 July 2008 (EDT)
::::The null hypothesis does not have to be proven; it is assumed to be true by default. The burden is on the experimenter to '''disprove''' the null hypothesis. The 0.05 point is a historical accident and is arbitrary, but hallowed by long use. There may be only an 8% chance that the null hypothesis is correct in the third replay, but that's enough. That doesn't detract from the fact that the combination of replay experiments was highly significant.
::::It is interesting to note that the results for the second and third replays were presented in such a way as to minimize the number of times that an individual cell mutated to Cit+; the data are presented as "''independent'' Cit+ mutants". It is mentioned that in the third replay 7 of the 8 plates gave rise to multiple Cit+ colonies, which was attributed to a two-mutation process; one of which took place during the population growth phase and one that took place after the bacteria were plated. Had the results been reported as 'number of Cit+ colonies formed', the mutations would have 'appeared' to scale better. I'm not arguing that the data should have been presented that way - I think that the inter-replay scaling is a red herring. It would also have made it impossible to include the first replay, since in that setting it was impossible to state just how many times the Cit+ trait arose independently in those cultures - almost certainly just once per culture, but not provably. What matters is that the intra-replay pattern of Cit+ mutation was the same across 3 different experimental setups. --[[User:Brossa|Brossa]] 19:05, 30 July 2008 (EDT)
:4)It was not 'an error' to include later generations in the replay experiments, because those generations provide information that bears on the contingency hypothesis- they can indicate if the contingency 'window' closes. It would be a valid request that the authors provide an explicit description of the techniques used to exclude the presence of occult Cit+ cells from the samples used in the replay experiments. At some point you will have to accept or reject that it is ''possible'' to do so, routinely, using basic laboratory techniques. If you accept that it is possible, the only way to prove that the routine procedures failed is to repeat the experiment with more rigorous controls and show a different result. Otherwise your objection boils down to "because I said so". If you reject categorically that it is possible to screen all Cit+ cells out, then all of microbiology comes into question, because these are truly fundamental techniques used across thousands of labs for many, many years.
:4)It was not 'an error' to include later generations in the replay experiments, because those generations provide information that bears on the contingency hypothesis- they can indicate if the contingency 'window' closes. It would be a valid request that the authors provide an explicit description of the techniques used to exclude the presence of occult Cit+ cells from the samples used in the replay experiments. At some point you will have to accept or reject that it is ''possible'' to do so, routinely, using basic laboratory techniques. If you accept that it is possible, the only way to prove that the routine procedures failed is to repeat the experiment with more rigorous controls and show a different result. Otherwise your objection boils down to "because I said so". If you reject categorically that it is possible to screen all Cit+ cells out, then all of microbiology comes into question, because these are truly fundamental techniques used across thousands of labs for many, many years.