Reaction time outliers in e-prime

[liwenna]

Hi Ashraf,

I think that many users don't really use e-merge for their data-
analysis (I don't at least and I know noone who does). So I really
couldn't tell you how to do that although I guess it would be similar
to the way you would do it when for instance using excel or spss.

What I myself do: I merge the edat files and export them to a .txt
file which I then import in SPSS. The next step is to know how you
want to define outliers: do you want to define outliers per subject,
per condition, per group etc etc. And do you want to define an outlier
as more than 2 sd's from the mean or 3 sd's from the mean? In many
tasks there's also a 'raw' cutoff of scores used based on the idea
that for instance (and depending on your task) any reactiontime under
say 200 ms or over 2000 ms are probably due to the subject not paying
attention to the task enough (i.e. reasonably too fast or too slow).
Once you have your definition you calculate the means and sd's over
the collection of reactiontimes that you want to remove outliers from
(for instance mean per group or mean per subject or mean per subject
AND condition etc). and then calculate the lower and upper bound (by
adding and subtracting 2 or 3 or depending on your definition sd's
from the mean) and then filter the reactiontimes under and over these
bounds out for that collection of reactiontimes. Alternatively you
could use median scores instead of mean scores for which you would not
need to calculate outliers... but... this might be harder to get
published, yet for a first glance at your data is a quicker way.

I agree with David that this is definitaly beyond the scope of this
google group which has more to do with using e-prime the program as
with the analysis of data collected with e-prime. In order to come to
a definition of your outliers you ought to consult with existing
literature on your task, with your statistics books and your common-
sense (and your supervisors perhaps? If you have one that is).

Good luck on it though :)

liw

On 11 dec, 00:01, ashraf <ash2003r... at yahoo.com> wrote:
>  David McFarlane,
> thank you for your reply ,there is a filter in e-prime  could help me
> to handle outliers in reaction time before analysis,my question about
> that,also about analysis the accuracy in e-merge. please help
> me ,thank you very much
>
> On Dec 11, 12:43 am, David McFarlane <mcfar... at msu.edu> wrote:
>
> > ashraf,
>
> > Offhand, I don't know if this is a question about how to do something
> > in E-Prime, or a more general question about how to handle outliers,
> > which would be better suited for a class on data analysis.  Also, is
> > there any reason to have E-Prime handle outliers at run time rather
> > than to gather all the raw data first and then deal with outliers
> > during analysis?  And you would probably use some statistics analysis
> > package to handle the outliers, not E-Prime.  But first you have to
> > have a good grounding in statistical data analysis before you start
> > throwing out data, and I would think that that topic goes way beyond
> > the purpose this E-Prime Group.
>
> > Regards,
> > -- David McFarlane, Professional Faultfinder
> > "For a successful technology, reality must take precedence over
> > public relations, for nature cannot be fooled."  (Richard Feynman,
> > Nobel prize-winning physicist)
>
> > >hi group
>
> > >what the effective way to remove reaction time outliers by e-prime
> > >and ,should i analysis the removable values  when i analysis the
> > >accuracy
> > >ashraf

--

You received this message because you are subscribed to the Google Groups "E-Prime" group.
To post to this group, send email to e-prime at googlegroups.com.
To unsubscribe from this group, send email to e-prime+unsubscribe at googlegroups.com.
For more options, visit this group at http://groups.google.com/group/e-prime?hl=en.