37
edits
Changes
Jump to navigation
Jump to search
Line 2:
Line 2: + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
More on p-values
Thanks for your interesting entry on the p-values. I look forward to reviewing it further.--[[User:Aschlafly|Andy Schlafly]] 09:55, 4 March 2009 (EST)
Thanks for your interesting entry on the p-values. I look forward to reviewing it further.--[[User:Aschlafly|Andy Schlafly]] 09:55, 4 March 2009 (EST)
:Anyone with Excel can calculate the p-values for the Blount el al. [http://myxo.css.msu.edu/lenski/pdf/2008,%20PNAS,%20Blount%20et%20al.pdf] experiments in minutes. The following table shows what the experiment one data should look like once entered into the spreadsheet. The Excel row numbers and column letters are shown. Once the data is entered into Excel, just type “=CHITEST(A1:B12,C1:D12)” into any empty cell of the spreadsheet to find the p-value of the experiment.
{|class="wikitable" style="text-align:center"
!
!A
!B
!C
!D
|-
!1
|0
|6
|0.333
|5.667
|-
!2
|0
|6
|0.333
|5.667
|-
!3
|0
|6
|0.333
|5.667
|-
!4
|0
|6
|0.333
|5.667
|-
!5
|0
|6
|0.333
|5.667
|-
!6
|0
|6
|0.333
|5.667
|-
!7
|0
|6
|0.333
|5.667
|-
!8
|1
|5
|0.333
|5.667
|-
!9
|0
|6
|0.333
|5.667
|-
!10
|1
|5
|0.333
|5.667
|-
!11
|0
|6
|0.333
|5.667
|-
!12
|2
|4
|0.333
|5.667
|-
:The first column is the number of mutations observed for each generation. The second column is the number that didn’t mutate. The third column is the expected number of mutants per generation under the null hypothesis. The null hypothesis from the paper is that the mutation rate is constant for all generations. The estimate of the mutation rate under the null is simply the total number of mutants divided by the total number of trials (4/72 for experiment one). Thus, column three is the number of trials per generation times the expected null hypothesis mutation rate (6*4/72 = 1/3 for experiment one). Column four is the expected number of trials that don’t mutate, which is also the number of trials minus the expected number of mutations.
:The best reference I’ve seen on chi-square tests is ''Mathematical Statistics with Applications'' by Wackerly et al. Chapter 14 of that book discusses the chi-square test in general and Section 14.4 shows specifically how to analyze contingency tables such as this one. If you don’t have access to that book, Wikipedia gives an alright explanation of the process on the page for “Pearson's chi-square test” under the heading “Test of independence”. [[User:SJohnson|SJohnson]] 12:22, 4 March 2009 (EST)