95% conf. intervals for geometric means?
Locked
95% conf. intervals for geometric means?
 
|
How can I calculate these for groups of observations?
Thanks for any suggestions. regards, Ian. Ian D. Martin, Ph.D. Tsuji Laboratory University of Waterloo Dept. of Environment & Resource Studies ===================== 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 |
Re: 95% conf. intervals for geometric means?
|
|
Calculate the 95% confidence interval in the log units and then back transform to the original units.
Paul R. Swank, Ph.D Professor and Director of Research Children's Learning Institute University of Texas Health Science Center Houston, TX 77038 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ian Martin Sent: Friday, February 06, 2009 8:40 AM To: [hidden email] Subject: 95% conf. intervals for geometric means? How can I calculate these for groups of observations? Thanks for any suggestions. regards, Ian. Ian D. Martin, Ph.D. Tsuji Laboratory University of Waterloo Dept. of Environment & Resource Studies ===================== 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 |
Re: 95% conf. intervals for geometric means?
|
|
In reply to this post by Ian Martin-4
Ian Martin wrote:
> How can I calculate these for groups of observations? Hi Martin: Log-transform your data, use EXAMINE to get 95%CI for group means, and back-transform the limits. Since the anti-log of the mean of the log(data) is the geometric mean, the anti-log of the confidence limits will also be the 95%CI for the geometric mean. HTH, Marta García-Granero -- For miscellaneous 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 |
Re: 95% conf. intervals for geometric means?
|
|
In reply to this post by Ian Martin-4
To follow up on the previous two comments: Marta and Paul's description
can include the further detail: the 2 resulting CI limits will not be symmetrical in the original units. One can do the same thing for standard deviations as for CI's: take the mean of the transformed data and subtract (for -1 SD) and add (for + 1 SD) the SD's to the mean. Take these resultant values, symmetrically different from the transformed mean, and back transform them. These will be the non-symetrical SD's in the original units. Joe Burleson -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ian Martin Sent: Friday, February 06, 2009 9:40 AM To: [hidden email] Subject: 95% conf. intervals for geometric means? How can I calculate these for groups of observations? Thanks for any suggestions. regards, Ian. Ian D. Martin, Ph.D. Tsuji Laboratory University of Waterloo Dept. of Environment & Resource Studies ===================== 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 |
Re: 95% conf. intervals for geometric means?
|
|
In reply to this post by Marta Garcia-Granero
Thanks to Marta, Paul, and others. I was hoping that a
backtransformed CI was explicitly implemented somewhere, so that I'd not have to transform and back transform for a large number of variables. Geometric mean is available through MEANS, so maybe SPSS could consider adding the CI for this measure too? There wouldn't be much point in adding (backtransformed) standard deviation or SEM, as these would erroneously imply symmetry about the mean. regards, Ian Ian D. Martin, Ph.D. Tsuji Laboratory University of Waterloo Dept. of Environment & Resource Studies On 06 Feb, 2009, at 11:15 AM, Marta García-Granero wrote: > Ian Martin wrote: >> How can I calculate these for groups of observations? > Hi Martin: > > Log-transform your data, use EXAMINE to get 95%CI for group means, and > back-transform the limits. Since the anti-log of the mean of the > log(data) is the geometric mean, the anti-log of the confidence limits > will also be the 95%CI for the geometric mean. > > HTH, > Marta García-Granero > > -- > For miscellaneous 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 ===================== 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 |
Re: 95% conf. intervals for geometric means?
|
|
You could probably use OMS and arrays to reduce the drudgery.
Paul R. Swank, Ph.D Professor and Director of Research Children's Learning Institute University of Texas Health Science Center Houston, TX 77038 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ian Martin Sent: Friday, February 06, 2009 12:09 PM To: [hidden email] Subject: Re: 95% conf. intervals for geometric means? Thanks to Marta, Paul, and others. I was hoping that a backtransformed CI was explicitly implemented somewhere, so that I'd not have to transform and back transform for a large number of variables. Geometric mean is available through MEANS, so maybe SPSS could consider adding the CI for this measure too? There wouldn't be much point in adding (backtransformed) standard deviation or SEM, as these would erroneously imply symmetry about the mean. regards, Ian Ian D. Martin, Ph.D. Tsuji Laboratory University of Waterloo Dept. of Environment & Resource Studies On 06 Feb, 2009, at 11:15 AM, Marta García-Granero wrote: > Ian Martin wrote: >> How can I calculate these for groups of observations? > Hi Martin: > > Log-transform your data, use EXAMINE to get 95%CI for group means, and > back-transform the limits. Since the anti-log of the mean of the > log(data) is the geometric mean, the anti-log of the confidence limits > will also be the 95%CI for the geometric mean. > > HTH, > Marta García-Granero > > -- > For miscellaneous 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 ===================== 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 |
Re: 95% conf. intervals for geometric means?
|
|
This is an implementation of Paul's idea of using OMS for the task (the
dataset used is "USA 1991 General Survey"): * This step is important for the success of the code (data cleaning steps are language dependent) *. SET OLANG=ENGLISH. * Variable to be transformed (replace by your own) *. COMPUTE LogVar = LG10(age) . * OMS (do not touch any line of code!) *. DATASET DECLARE CIData. OMS /SELECT TABLES /IF COMMANDS = ["Explore"] SUBTYPES = ["Descriptives"] /DESTINATION FORMAT = SAV OUTFILE = CIData VIEWER = NO. * Replace grouping variable by your own (more than one can be used at the same time) *. EXAMINE VARIABLES=LogVar BY sex /PLOT NONE /STATISTICS DESCRIPTIVES /CINTERVAL 95 /MISSING PAIRWISE /NOTOTAL. OMSEND. * Data cleaning and back-transform *. DATASET ACTIVATE CIData. IF Var4='' Var4=Var3. SELECT IF Var4='Mean' OR Var4='Lower Bound' OR Var4='Upper Bound'. EXE. DELETE VARIABLES Command_ TO Var1 Std.Error Var3. IF Var4='Lower Bound' Var4='LowerBound'. IF Var4='Upper Bound' Var4='UpperBound'. * Restructure output dataset *. SORT CASES BY Var2 Var4. CASESTOVARS /ID = Var2 /INDEX = Var4 /GROUPBY = VARIABLE . * Backtransform *. COMPUTE Geomean=10**Mean. COMPUTE Lower=10**Lowerbound. COMPUTE Upper=10**UpPerbound. * Report *. LIST VARIABLES Var2 Geomean Lower Upper. HTH, Marta García-Granero. > -----Original Message----- > From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ian Martin > Sent: Friday, February 06, 2009 12:09 PM > To: [hidden email] > Subject: Re: 95% conf. intervals for geometric means? > > Thanks to Marta, Paul, and others. I was hoping that a > backtransformed CI was explicitly implemented somewhere, so that I'd > not have to transform and back transform for a large number of > variables. Geometric mean is available through MEANS, so maybe SPSS > could consider adding the CI for this measure too? > --- For miscellaneous 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 |
|
|
Design your own exclusive Pingbox today! It's easy to create your personal chat space on your blogs |
|
|
you must sort data by the variable used as
subject
|
|
|
In reply to this post by E. Bernardo
Design your own exclusive Pingbox today! It's easy to create your personal chat space on your blogs |
Re: 95% conf. intervals for geometric means?
|
|
In reply to this post by Marta Garcia-Granero
Marta,
Thanks very much. It is most kind of you to spend the time to develop the OMS syntax. I'm sure others will find it useful too. best regards, Ian Ian D. Martin, Ph.D. Aquatic Ecologist On 08 Feb, 2009, at 5:26 AM, Marta García-Granero wrote: > This is an implementation of Paul's idea of using OMS for the task > (the > dataset used is "USA 1991 General Survey"): > > * This step is important for the success of the code (data cleaning > steps are language dependent) *. > SET OLANG=ENGLISH. > * Variable to be transformed (replace by your own) *. > COMPUTE LogVar = LG10(age) . > * OMS (do not touch any line of code!) *. > DATASET DECLARE CIData. > OMS > /SELECT TABLES > /IF COMMANDS = ["Explore"] > SUBTYPES = ["Descriptives"] > /DESTINATION FORMAT = SAV > OUTFILE = CIData > VIEWER = NO. > * Replace grouping variable by your own (more than one can be used at > the same time) *. > EXAMINE > VARIABLES=LogVar BY sex > /PLOT NONE > /STATISTICS DESCRIPTIVES > /CINTERVAL 95 > /MISSING PAIRWISE > /NOTOTAL. > OMSEND. > * Data cleaning and back-transform *. > DATASET ACTIVATE CIData. > IF Var4='' Var4=Var3. > SELECT IF Var4='Mean' OR Var4='Lower Bound' OR Var4='Upper Bound'. > EXE. > DELETE VARIABLES Command_ TO Var1 Std.Error Var3. > IF Var4='Lower Bound' Var4='LowerBound'. > IF Var4='Upper Bound' Var4='UpperBound'. > * Restructure output dataset *. > SORT CASES BY Var2 Var4. > CASESTOVARS > /ID = Var2 > /INDEX = Var4 > /GROUPBY = VARIABLE . > * Backtransform *. > COMPUTE Geomean=10**Mean. > COMPUTE Lower=10**Lowerbound. > COMPUTE Upper=10**UpPerbound. > * Report *. > LIST VARIABLES Var2 Geomean Lower Upper. > > HTH, > Marta García-Granero. > > >> -----Original Message----- >> From: SPSSX(r) Discussion [mailto:[hidden email]] On >> Behalf Of Ian Martin >> Sent: Friday, February 06, 2009 12:09 PM >> To: [hidden email] >> Subject: Re: 95% conf. intervals for geometric means? >> >> Thanks to Marta, Paul, and others. I was hoping that a >> backtransformed CI was explicitly implemented somewhere, so that I'd >> not have to transform and back transform for a large number of >> variables. Geometric mean is available through MEANS, so maybe SPSS >> could consider adding the CI for this measure too? >> > > --- > > For miscellaneous 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 ===================== 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 |
|
|
In reply to this post by E. Bernardo
Bring your friends to the fun. Invite your friends from Hotmail, Gmail to Yahoo! Mail today! |
|
|
Hi Eins
The scores of the utility
file are the global utility of the cards used in experiment. They are computed by
adding the constant + the part-worth utilities of each level/attribute of each
card
The importance is the relative weight (%) of the amplitude of the utilities of each attribute.
|
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |