how to compare 3 or more means?

[Matteo Riccò-2]

Hallo everybody!

A young surgeon, in order to present his PhD thesis asked me about a
procedure to compare a set of means (i.e. mean lenght of surgical
procedure). He's looking for the classical "p" in order to confirm
that the surgical procedure he developed is better than classical
procedure in international experiences. Then I ask you, if someone
could suggest a simple test able to compare such a data (available:
mean, number of cases for every casuistic, minimum, maximum, standard
deviation). Any suggestion?

Thanks a lot

MR

=====================
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

[Richard Ristow]

At 08:18 AM 9/21/2009, Matteo Riccò wrote:

>A young surgeon, asked me about a procedure to
>compare a set of means (i.e. mean length of
>surgical procedure). He's looking for the
>classical "p" in order to confirm that the
>surgical procedure he developed is better than
>classical procedure in international
>experiences. Then I ask you, if someone could
>suggest a simple test able to compare such a
>data (available: mean, number of cases for every
>casuistic, minimum, maximum, standard deviation). Any suggestion?

That looks like a classic one-way ANOVA, with
each 'casuistic' (procedure?) as a cell. I
believe that cell n, mean, and standard deviation
are enough to run such an ANOVA. Look at
procedure ONEWAY, MATRIX input; restructure your
data into the form of an input matrix for ONEWAY.

-Best of luck,
  Richard

=====================
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

[Marta Garcia-Granero]

Hi everybody:

I wrote code time ago to complement the analysis:

http://www.spsstools.net/Syntax/T-Test/ONEWAYwithSummaryData1.txt

HTH,
Marta GG

Richard Ristow wrote:

> At 08:18 AM 9/21/2009, Matteo Riccò wrote:
>
>> A young surgeon, asked me about a procedure to
>> compare a set of means (i.e. mean length of
>> surgical procedure). He's looking for the
>> classical "p" in order to confirm that the
>> surgical procedure he developed is better than
>> classical procedure in international
>> experiences. Then I ask you, if someone could
>> suggest a simple test able to compare such a
>> data (available: mean, number of cases for every
>> casuistic, minimum, maximum, standard deviation). Any suggestion?
>
> That looks like a classic one-way ANOVA, with
> each 'casuistic' (procedure?) as a cell. I
> believe that cell n, mean, and standard deviation
> are enough to run such an ANOVA. Look at
> procedure ONEWAY, MATRIX input; restructure your
> data into the form of an input matrix for ONEWAY.
>
> -Best of luck,
>  Richard
>
> =====================
> 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
>

--
For miscellaneous SPSS related statistical stuff, visit:
http://gjyp.nl/marta/

=====================
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

[Bruce Weaver]

Matteo Riccò-2 wrote

Hallo everybody!

A young surgeon, in order to present his PhD thesis asked me about a
procedure to compare a set of means (i.e. mean lenght of surgical
procedure). He's looking for the classical "p" in order to confirm
that the surgical procedure he developed is better than classical
procedure in international experiences. Then I ask you, if someone
could suggest a simple test able to compare such a data (available:
mean, number of cases for every casuistic, minimum, maximum, standard
deviation). Any suggestion?

Thanks a lot

MR

You've had one response suggesting one-way ANOVA.  But note that it would require independence of observations--i.e., it would only be appropriate if each surgeon performed only one procedure.  Is that the case for your data?  If not, did each surgeon perform only one of the three procedures, or did they perform all 3?  For these latter two scenarios, a multilevel model (MIXED procedure) would be better than one-way ANOVA, IMO.  For an accessible introduction to multilevel models, see Jos Twisk's book, Applied Multilevel Analysis (Cambridge Univ. Press).