ANOVA using dataset with means and SE

[filipepaula]

Hi all,

I am analysing relative RT-PCR data. The mean normalized relative quantities of the various genes for each sample analysed, as well as the corresponding standard errors, are best calculated using a simple spreadsheet due to the complexity of the calculations (see for ex. The Plant Cell 2009;21:1031-3 for details).
As such, I end up with a table of relative quantities and standard errors for each gene/sample.
As suggested in the literature, it is appropriate to perform an ANOVA for differences between samples / groups of samples.

The question is: how to perform an ANOVA (or any other statistical test) in STATA by feeding only means and SE's, instead of the whole dataset of observations, as is needed to do when, like in this case, the calculations for means and SE are done outside of STATA?

Thanks in advance,

[Bruce Weaver]

Take a look at John Gleason's aovsum program.

net install sg132.pkg

Here's an example from the help file using a list of SDs:

. aovsum, n(23 12 14 19) m(11.5 23.0 29.8 38.4) sd(2.11 4.07 3.43 4.62)

     Groups |    Summary of Response variable
    (cells) |        Mean   Std. Dev.        Obs.
------------+------------------------------------
          1 |        11.5        2.11          23
          2 |          23        4.07          12
          3 |        29.8        3.43          14
          4 |        38.4        4.62          19
------------+------------------------------------
      Total |   24.813235   11.453267          68

                        Analysis of Variance
    Source              SS         df      MS            F     Prob > F
------------------------------------------------------------------------
Between groups      7971.57809      3    2657.1927    208.08     0.0000
 Within groups         817.303     64   12.7703594
------------------------------------------------------------------------
    Total           8788.88109     67    131.17733

Bartlett's test for equal variances:  chi2(3) =  11.8087  Prob>chi2 = 0.008

For a list of SEs, replace sd(list) with se(list).

[Dave Airey]

I often analyze qPCR data by delta delta ct methods in Stata. I don't reach for Excel. I perform the analysis on the delta ct estimates inside Stata or R. See this for help.

https://www.nature.com/nprot/journal...t.2008.73.html

IIf you are not doing relative expression calculations, there is a paper out there describing some ANOVA code for summary statistics.

http://citeseerx.ist.psu.edu/viewdoc...=rep1&type=pdf