[R] Automatically assign symbol for statistical significance to x given value of x

[Marc Schwartz]

On Jul 20, 2009, at 10:36 AM, Tormod Bøe wrote:

> Dear r-help users,
>
> I am using Sweave and Latex to create tables with output from several
> statistical test. As an example: I have a grouping variable "group"
> with two levels ("x" and "y") which I compare on variables ("a" and
> "b").
>
> I have created a table in which means, standard deviations, and the
> statistic and p.value resulting from a t.test is printed. The table
> looks something like:
>
>  group(x)                  group(y)
> mean        sd        mean      sd     t-value p-value  sig.
> mean(x,a)  sd(x,a) mean(y,a) sd(y,a)    t.a      p.a
> mean(x,b)  sd(x,b) mean(y,b) sd(y,b)    t.b      p.b
>
> What I would like to be able to add to this table is an indication by
> means of asterisks ("*") the level of significance of the p.value(s).
>
> I am after an expression to put in column "sig." which evaluates the
> p.value and prints *, ** or *** depending on it's value (e.g. if it is
> less than .05 print "*", but if it is less than .01 print "**", but if
> it is less than .001 print "***", else print " ").
>
> I am able to use a simple function like if (p.value<0.5) print("*")
> else print(" "),
> but I have not been able to work out how to nest the "if"  
> expressions correctly.
>
> Any help would be greatly appreciated.
>
> Best regards,
> Tormod Bøe

The simple answer is to use symnum(), which is the function the  
generates the symbology for those tables in R where the 'significance  
stars' are included. For example, when printing a linear model summary  
where printCoefmat() is used. See the last example in ?symnum for more  
information.

The philosophical challenge that you will face is that a lot of people  
are not favorably disposed to the use of this symbology. So the use or  
non-use of them may be dependent upon your own feelings and those of  
the consumers of your output. See ?options and note  
'show.signif.stars', which a lot of folks, myself included, set to  
FALSE in .Rprofile.

HTH,

Marc Schwartz