Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Interpretation of Two-sample t test with equal variances?

From   Nick Cox <>
Subject   Re: st: Interpretation of Two-sample t test with equal variances?
Date   Wed, 20 Mar 2013 16:58:41 +0000

I wouldn't want to start with box plots here. For two groups, there is
space for more detail than a box plot gives, except capriciously in so
far as points are more than 1.5 IQR from the nearer quartile.

I grew up, as it were, on box plots, including David's own writings 30
or more years ago, but I think they are oversold in total.

I have often seen -- even in otherwise good textbooks -- people
talking about t-tests or ANOVA and then giving box plots -- and not
even commenting on the mismatch, i.e. that the box plots are showing
medians etc., not means etc.

A more directly relevant graph would, I suggest, be a dotplot (sensu
-dotplot-) or even -stripplot- (SSC). You can show means on such


On Wed, Mar 20, 2013 at 4:43 PM, David Hoaglin <> wrote:
> Despite the tolerance of Student's t for departures from assumptions,
> comparing the (mean) ages in the two outcome groups is not the right
> approach.  For initial exploration, side-by-side boxplots of age (by
> mode_delivery) would give an indication of skewness, as well as
> presence of unusually low or high ages.  Apart from giving some points
> undue influence, skewness in the distribution of a predictor (per se)
> is often not important.  If the predictor could be transformed, to
> straighten a nonlinear relation with the outcome, the transformation
> might (as a by-product) reduce or remove the skewness.
> David Hoaglin
> On Wed, Mar 20, 2013 at 12:16 PM, Gwinyai Masukume
> <> wrote:
>> Thank you so much everyone. Appreciated.
>> David - it was indeed a very helpful discussion.
>> Nick - indeed those are means of maternal age. you are significant.
>> yes, the mother's ages are skewed. what do you mean by student's t
>> test works well even if you lie to it?
>> Carlo - it seems all the relevant independent variables have not been
>> included, the very low pseudo r2 is bizarre to me.
>> Thanks again.
>> Gwinyai
> *
> *   For searches and help try:
> *
> *
> *
*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index