Re: Women talk only 3.4% more than men

On Jul 9, 7:15 am, Peter Brooks <Peter.H.M.Bro...@xxxxxxxxx> wrote:
On Jul 9, 1:02 am, Lance <LanceG...@xxxxxxxxx> wrote:



On Jul 8, 4:02 pm, Peter Brooks <Peter.H.M.Bro...@xxxxxxxxx> wrote:

On Jul 8, 12:52 pm, Lance <LanceG...@xxxxxxxxx> wrote:

On Jul 8, 10:23 am, Dave Smith <d...@xxxxxxxxxxxxxxxxxxxxxx> wrote:

On 7 Jul, 14:33, Peter Brooks <Peter.H.M.Bro...@xxxxxxxxx> wrote:

On Jul 7, 10:46 am, Dave Smith <d...@xxxxxxxxxxxxxxxxxxxxxx> wrote:

On 6 Jul, 11:56, Peter Brooks <Peter.H.M.Bro...@xxxxxxxxx> wrote:

Here's a strange article from IOL. The title claims that men talk just
as much as women and the text claims that the difference is 'a wash in
statistical terms'. However, simple calculation shows that women talk
3.4% more than men, which is a real enough difference, even from a
sample size of 400 - we aren't given the standard deviation, which
might also be interesting.

Why do journalists do this sort of thing?
" ................................................
The results showed that while the women got through an average of 16
215 words, the men were no slouches, with a daily average of 15 669 -
a wash in statistical terms................."

I think the percentage difference is quite small. My criticism would
be that maybe an unwarranted generalisation is being made from a
sample of students.

Yes, I agree it's a small sample and a big generalisation as well. The
difference is, as you say, 'quite small', but that's rather different
from 'statistically insignificant'!

When I did research many years ago, we used to say a result was not
statistically significant rather than that it was statistically
insignificant. It might be better to talk about statistical
reliability rather than statistical significance.- Hide quoted text -

- Show quoted text -

I think a better approach is to add in talk about "effect size".
Statistical significance can occur even when the size of the
difference between groups is small. So, in addition to significance,
we need to say how large an effect (the difference between the groups)
really is. Thus there is good reason to think that men are better than
women at maths, but the effect size is small (and the difference
really only matters when one gets to the extremes of the
distribution).

In this case perhaps the 3.5% difference is significant, but it seems
to me to be a small effect size. Presuming a normal distribution of
talkativeness it would really only make a difference at the extreme
ends of the distribution. There are likely then to be a very small
number of women who are extremely talkative, and who can outtalk any
man.

How small? Have I met all of them?

I'm quite keen on chatting, but my mother in law would talk anybody
under the table.

That would depend on things like whether the distribution is normal,
how large the variance is, and the like.

Effect size for differences between groups is usually calculated by
dividing the difference between the groups by the standard deviation
of the combined groups.

Yes, all those would be important, I agree.

Also on experimental design. When the parties gave their consent to
take part in the experiment, did they know the objective? If they did,
then it might be a measure of which sex is better able to control
itslevel of discussion.

As I've said, I think that it is likely, from previous studies that
we've looked at here, even, that the SD bet of he male students would
be greater than that of the female.- Hide quoted text -

- Show quoted text -

They may have been asked for permission after the data had been
collected?

Lance

.

Relevant Pages

  • Re: Women talk only 3.4% more than men
    ... statistically significant rather than that it was statistically ... really only matters when one gets to the extremes of the ... Presuming a normal distribution of ... Surely their temperament and happiness are the main variables? ...
    (uk.philosophy.humanism)
  • Re: Women talk only 3.4% more than men
    ... statistically significant rather than that it was statistically ... really only matters when one gets to the extremes of the ... Presuming a normal distribution of ... Surely their temperament and happiness are the main variables? ...
    (uk.philosophy.humanism)
  • Re: Women talk only 3.4% more than men
    ... statistically significant rather than that it was statistically ... really only matters when one gets to the extremes of the ... Presuming a normal distribution of ... talkativeness it would really only make a difference at the extreme ...
    (uk.philosophy.humanism)
  • Re: Women talk only 3.4% more than men
    ... statistically significant rather than that it was statistically ... really only matters when one gets to the extremes of the ... Presuming a normal distribution of ... talkativeness it would really only make a difference at the extreme ...
    (uk.philosophy.humanism)
Loading