[R] how to calculate the mean of a group in a table

[David Winsemius]

On May 7, 2011, at 2:24 PM, tornanddesperate wrote:

> Hi its me again
>
> I don't mean to get on your nerves, but the use of R proofs to be a  
> bit more
> complicated than envisaged.
>
> I would like to calculate the mean of a group of values, here
> "wage_accepted". The group is determined by the stage and period, so  
> in the
> end there should be a column with the values of the wages in period 1
> stage1, period 1 stage 2, period 2 stage1... Unfortunately, I  
> haven't much
> of a clue on how to program this. Could you tell me how the function  
> should
> roughly look like – if-else, loops included or not – in the end ?
>
>   treatment session period stage wage_accepted type
> 1          1       1      1     1            25  low
> 2          1       1      1     1            19  low
> 3          1       1      1     1            15  low
> 4          1       1      1     2            32 high
> 5          1       1      1     2            13  low
> 6          1       1      1     2            14  low
> 7          1       1      2     1            17  low
> 8          1       1      2     1            12  low
>
> I am also very grateful if someone could point out a good internet  
> tutorial
> resource for R functions such as if-else and loops and how to use  
> them in
> conjunction with tables. It should contain simple examples if  
> possible.

I see Pete Brecknock has given you two workable answers with `by` and  
`aggregate`. According to the tapply help page, those functions are  
both based on tapply, and so it should be no surprise that tapply is  
also workable:

  tapply(d$wage_accepted, list(period=d$period, stage=d$stage), mean)
       stage
period        1        2
      1 19.66667 19.66667
      2 14.50000       NA

But results are returned as a table, which I find more usable since I  
can then use the inheritance of tables from matrix properties and  
divide tapply-summary calculations by the results of table() done with  
the same marginal factors. On the other hand aggregate can deliver  
multiple column results where tapply can only deliver the results from  
one column. Their argument requirements are somewhat different, eg.  
tapply and by do not need the name of the FUN argument but aggregate  
does, and I keep needing to refresh my memory on those.

-- 

David Winsemius, MD
West Hartford, CT