I have to agree with Martin. IMHO outliers need to be defined in the context of the data. And even then, as more cases are gathered, the chances of "outliers" being legitimate increases. A better use of ones time might be to identify the source of the outlier - is it a data entry error, a respondent error, a misinterpretation of the question, etc.
One accepted method of identifying outliers is more than 3 standard deviations from the mean. Using this as your criteria you could then automate their removal in a Python script. Again, be careful about the context - would you want ages greater than, say, 90 removed? but if the question was ages of active active military personnel, then removing those over 90 would make sense.
"Clean" data is most desirable but scrubbing it to the point of adversely affecting good analysis defeats the purpose.
Best,
George
--- On Fri, 1/16/09, Martin Holt <
[hidden email]> wrote:
From: Martin Holt <
[hidden email]>
Subject: Re: Remove Outliers
To:
[hidden email]
Date: Friday, January 16, 2009, 11:18 AM
I know that there have been a number of threads in the past on the dangers
of removing outliers, and I don't propose to go through these again. I do
recommend that the original OP review these from the archives. To
"automatically" remove outliers rings loud warning bells in my ears.
Best Wishes,
Martin Holt
----- Original Message -----
From: "Sam Ballard" <
[hidden email]>
To: <
[hidden email]>
Sent: Friday, January 16, 2009 3:53 PM
Subject: Re: Remove Outliers
> You can use Explore to find extreme scores (Analyze, Descriptive Stats,
> Explore). This will give you a Stem-and-Leaf Plot and Boxplot to find
> outliers.
>
> Sam
>
>
>
> On Thu, 15 Jan 2009 14:27:20 -0800, jimjohn <
[hidden email]>
wrote:
>
>>can someone plz tell me if there are any functions in SPSS that will
>>automatically remove outliers. or if not is there some way i can detect
>>outliers in SPSS? thanks.
>>--
>>View this message in context:
http://www.nabble.com/Remove-Outliers-> tp21488242p21488242.html
>>Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>>
>>=====================
>>To manage your subscription to SPSSX-L, send a message to
>>
[hidden email] (not to SPSSX-L), with no body text except
the
>>command. To leave the list, send the command
>>SIGNOFF SPSSX-L
>>For a list of commands to manage subscriptions, send the command
>>INFO REFCARD
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
>
[hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>
>
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
====================To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
