Trend Analysis - Anova One-way with intervals that are not equal
Locked
Trend Analysis - Anova One-way with intervals that are not equal
 
|
Hi everybody, I would really appreciate your very expert help with
this problem. I cannot find the solution. I am doing a Trend Analysis. The only way I know how
to do it is with ANOVA One-way. I have a ratio dependent variable (Y axis). As factor I have measured the exact point where in the
X axis I get the values for that dependent variable: the intervals among the
values of the factor variable are not equal. I guess that the key point to express the problem is
that I wish to do a Trend Analysis and the intervals in the factor variable are
not equal. Somebody has told me that SPSS (or the statistic
itself, I don’t know) when calculating ANOVA One-way/Polynomic considers
that the intervals among the values of the factor are constant. I would like to
consider these intervals as unequal, I would like to give to SPSS the values
that I have found… and I would love SPSS to take into consideration these
values for the Factor... J Possibly the questions are: 1.- Can SPSS calculate Trend Analysis with ANOVA
One-way having a Factor with different intervals? 2.- If SPSS cannot do it with the menu, can SPSS do it
with syntax? How? 3.- Is it there some other way to calculate
Trend-Analysis? I ask it in case this would be another way to solve the problem. I would like to thank any in advance for all your help,
time and suggestions. Kind regards, Giorgio |
Re: Trend Analysis - Anova One-way with intervals that are not equal
|
|
Giorgio, It sound like you are looking at a growth curve analysis . I would recommend that you look at the work by Judith Singer esp her "applied Longitudinal Dada Analysis " text with John Willett. SPSS can handle these models in the MIXED model and the web page http://www.ats.ucla.edu/stat/spss/examples/alda/ provides SPSS programming to accomplish the example set forth by Singer and Willett. Also you may want to look at the Duncan et al text on Latent Growth Curve Modeling which has a chapter that relates the ANOVA to the study of growth curves. Bill William N. Dudley, PhD Associate Dean for Research The School of Health and Human Performance Office of Research The University of North Carolina at Greensboro 126 HHP Building, PO Box 26170 Greensboro, NC 27402-6170 VOICE 336.2562475 FAX 336.334.3238
Hi everybody, I would really appreciate your very expert help with this problem. I cannot find the solution. I am doing a Trend Analysis. The only way I know how to do it is with ANOVA One-way. I have a ratio dependent variable (Y axis). As factor I have measured the exact point where in the X axis I get the values for that dependent variable: the intervals among the values of the factor variable are not equal. I guess that the key point to express the problem is that I wish to do a Trend Analysis and the intervals in the factor variable are not equal. Somebody has told me that SPSS (or the statistic itself, I don’t know) when calculating ANOVA One-way/Polynomic considers that the intervals among the values of the factor are constant. I would like to consider these intervals as unequal, I would like to give to SPSS the values that I have found… and I would love SPSS to take into consideration these values for the Factor... J Possibly the questions are: 1.- Can SPSS calculate Trend Analysis with ANOVA One-way having a Factor with different intervals? 2.- If SPSS cannot do it with the menu, can SPSS do it with syntax? How? 3.- Is it there some other way to calculate Trend-Analysis? I ask it in case this would be another way to solve the problem. I would like to thank any in advance for all your help, time and suggestions. Kind regards, Giorgio |
|
|
Hi all- I’m having a small problem with ntiles
syntax (a subcommand of rank). ***this works, but does not let me specify a
name for the new variable created RANK VARIABLES=TotalOrdersSum$08 (A)
BY FIRSTOrder_Year_coded /TIES=Low /ntiles
(5) /PRINT=No . ****neither
of the following 2 work, but seem to fit with the documentation (included at
the end) RANK VARIABLES=TotalOrdersSum$08 (A)
BY FIRSTOrder_Year_coded /TIES=Low /ntiles (5)
/PRINT=No /INTO OverallDataNTile . RANK VARIABLES=TotalOrdersSum$08 (A)
BY FIRSTOrder_Year_coded /TIES=Low /ntiles (5) /
INTO OverallDataNTile /PRINT=No . I’m trying to name the new variable OverallDataNTile But am not clear where in the INTO command
goes. Thanks, Brian ****documentation RANK [VARIABLES=] varlist [({A**})]
{D } [BY varlist] [/TIES={MEAN** }]
{LOW }
{HIGH }
{CONDENSE} [/FRACTION={BLOM**}]
{TUKEY }
{VW }
{RANKIT} [/PRINT={YES**}]
{NO } [/MISSING={EXCLUDE**}]
{INCLUDE } The following can each be specified once: [/RANK**] [/NTILES(k)] [/NORMAL] [/PERCENT] [/RFRACTION] [/PROPORTION] [/N] [/SAVAGE] INTO can be used after any of the above rank function subcommands: [INTO rankvarlist] **Default
if the subcommand is omitted. |
|
|
INTO belongs on the subcommand that specifies the RANK function, in this case NTILES:
RANK VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded /TIES=Low /ntiles (5) INTO OverallDataNTile /PRINT=No.
From: SPSSX(r) Discussion on behalf of Brian Moore Sent: Mon 4/27/2009 3:12 PM To: [hidden email] Subject: Ntiles Syntax Hi all- I’m having a small problem with ntiles syntax (a subcommand of rank).
***this works, but does not let me specify a name for the new variable created
RANK VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded /TIES=Low /ntiles (5) /PRINT=No .
****neither of the following 2 work, but seem to fit with the documentation (included at the end)
RANK VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded /TIES=Low /ntiles (5) /PRINT=No /INTO OverallDataNTile .
RANK VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded /TIES=Low /ntiles (5) / INTO OverallDataNTile /PRINT=No .
I’m trying to name the new variable OverallDataNTile
But am not clear where in the INTO command goes.
Thanks, Brian
****documentation
RANK [VARIABLES=] varlist [({A**})] {D } [BY varlist] [/TIES={MEAN** }] {LOW } {HIGH } {CONDENSE} [/FRACTION={BLOM**}] {TUKEY } {VW } {RANKIT} [/PRINT={YES**}] {NO } [/MISSING={EXCLUDE**}] {INCLUDE } The following can each be specified once: [/RANK**] [/NTILES(k)] [/NORMAL] [/PERCENT] [/RFRACTION] [/PROPORTION] [/N] [/SAVAGE] INTO can be used after any of the above rank function subcommands: [INTO rankvarlist] **Default if the subcommand is omitted. |
Macros for Deming regression uploaded to web site
|
|
In reply to this post by William Dudley WNDUDLEY
Hi everybody!
Thanks to all who helped in developing my Deming regression macros. As with every "not too standard" statistical method, reading the recommended references is mandatory (to understand the difference between weighted and unweighted Deming regression, what Lambda is...). They can be downloaded from: http://gjyp.nl/marta/Deming%20Regression%20MACROS.SPS The web page does not show the link yet, until the page is modified (I don't perform myself the modifications). Any feedback or suggestions to improve the code is welcome. Regards, Marta García-Granero ===================== 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 Oliver, Richard
I was surprised this didn’t work as well. The error
message is below & I’m using version 15. Warnings
From: SPSSX(r) Discussion
[mailto:[hidden email]] On Behalf Of Oliver, Richard INTO belongs on the subcommand that specifies the RANK function,
in this case NTILES: RANK
VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded
/TIES=Low /ntiles (5) INTO OverallDataNTile /PRINT=No. From: SPSSX(r) Discussion on behalf of Brian
Moore Hi
all- I’m
having a small problem with ntiles syntax (a subcommand of rank). ***this
works, but does not let me specify a name for the new variable created RANK
VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded
/TIES=Low /ntiles (5) /PRINT=No . ****neither
of the following 2 work, but seem to fit with the documentation (included at
the end) RANK
VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded
/TIES=Low /ntiles (5) /PRINT=No /INTO OverallDataNTile . RANK
VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded
/TIES=Low /ntiles (5) / INTO OverallDataNTile
/PRINT=No . I’m
trying to name the new variable OverallDataNTile But
am not clear where in the INTO command goes. Thanks, Brian ****documentation RANK
[VARIABLES=] varlist [({A**})]
{D }
[BY varlist]
[/TIES={MEAN** }]
{LOW }
{HIGH }
{CONDENSE}
[/FRACTION={BLOM**}]
{TUKEY }
{VW }
{RANKIT}
[/PRINT={YES**}]
{NO }
[/MISSING={EXCLUDE**}]
{INCLUDE } The
following can each be specified once:
[/RANK**] [/NTILES(k)] [/NORMAL]
[/PERCENT] [/RFRACTION]
[/PROPORTION] [/N] [/SAVAGE] INTO
can be used after any of the above rank
function subcommands:
[INTO rankvarlist] **Default
if the subcommand is omitted. |
Re: Trend Analysis - Anova One-way with intervals that are not equal
|
|
In reply to this post by giorgio de marchis
I don’t think this is a growth curve. I think you have different
values of X that are not equidistant and their associated values on Y. Why not
do an ANOVA first, assuming X is categorical and then do a regressions assuming
the relation between x and Y is linear. The difference in these models
quantified by the partial F test will tell you if you have deviations from
linearity. If not, then interpret the slope and intercept form the regression. Dr. Paul R. Swank, Professor and Director of Research Children's Learning Institute University of Texas Health Science Center-Houston From: SPSSX(r) Discussion
[mailto:[hidden email]] On Behalf Of giorgio de marchis Hi everybody, I would really
appreciate your very expert help with this problem. I cannot find the solution. I am doing a Trend
Analysis. The only way I know how to do it is with ANOVA One-way. I have a ratio
dependent variable (Y axis). As factor I have
measured the exact point where in the X axis I get the values for that
dependent variable: the intervals among the values of the factor variable are
not equal. I guess that the
key point to express the problem is that I wish to do a Trend Analysis and the
intervals in the factor variable are not equal. Somebody has told
me that SPSS (or the statistic itself, I don’t know) when calculating ANOVA
One-way/Polynomic considers that the intervals among the values of the factor
are constant. I would like to consider these intervals as unequal, I would like
to give to SPSS the values that I have found… and I would love SPSS to take
into consideration these values for the Factor... J Possibly the
questions are: 1.- Can SPSS
calculate Trend Analysis with ANOVA One-way having a Factor with different
intervals? 2.- If SPSS cannot
do it with the menu, can SPSS do it with syntax? How? 3.- Is it there
some other way to calculate Trend-Analysis? I ask it in case this would be
another way to solve the problem. I would like to
thank any in advance for all your help, time and suggestions. Kind regards, Giorgio |
|
|
In reply to this post by Brian Moore-7
Brian Moore wrote:
> > I was surprised this didn’t work as well. The error message is below & > I’m using version 15. > > * Warnings* > > * * > > One of the following subcommands was expected but not found: MISSING, > TIES, FRACTION, PRINT, RANK, RFRACTION, NORMAL, PERCENT, NTILES, > PROPORTION, N, SAVAGE. > > This command is not executed. > RANK VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded /ntiles (5) INTO OverallDataNTile /PRINT=No /TIES=Low . HTH, Marta -- 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 |
|
|
Other than the VARIABLES subcommand coming first, subcommand order shouldn't make any difference. The documentation is clear on that point, and the example I provided works in release 17. Here it is with some test data:
data list free /TotalOrdersSum$08 FIRSTOrder_Year_coded. begin data 1 1 2 1 3 1 4 2 5 2 6 2 end data. RANK VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded /TIES=Low /ntiles (5) INTO OverallDataNTile /PRINT=No. -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Marta García-Granero Sent: Wednesday, April 29, 2009 9:13 AM To: [hidden email] Subject: Re: Ntiles Syntax Brian Moore wrote: > > I was surprised this didn't work as well. The error message is below & > I'm using version 15. > > * Warnings* > > * * > > One of the following subcommands was expected but not found: MISSING, > TIES, FRACTION, PRINT, RANK, RFRACTION, NORMAL, PERCENT, NTILES, > PROPORTION, N, SAVAGE. > > This command is not executed. > RANK VARIABLES=TotalOrdersSum$08 (A) BY FIRSTOrder_Year_coded /ntiles (5) INTO OverallDataNTile /PRINT=No /TIES=Low . HTH, Marta -- 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 ===================== 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 giorgio de marchis
You don't provide enough information to give you a very specific answer but here are some comments. (1) If you have many levels of X a regression is more appropriate than a trend analysis. (2) No, a trend analysis does not assume equal intervals but the orthogonal polynomial coefficients that Spss uses, as well as those published in the tables in the back of stats books, do assume equal intervals. If you have unequal intervals you need to find a statistician who can assist you in calculating a set of appropriate coefficients. When you have these you can use them in spss by specifying the desired contrasts. (3) If you have a between subjects design then a growth curve analysis is irrelevant to your problem.
<quote author="giorgio de marchis"> Hi everybody, I would really appreciate your very expert help with this problem. I cannot find the solution. I am doing a Trend Analysis. The only way I know how to do it is with ANOVA One-way. I have a ratio dependent variable (Y axis). As factor I have measured the exact point where in the X axis I get the values for that dependent variable: the intervals among the values of the factor variable are not equal. I guess that the key point to express the problem is that I wish to do a Trend Analysis and the intervals in the factor variable are not equal. Somebody has told me that SPSS (or the statistic itself, I don't know) when calculating ANOVA One-way/Polynomic considers that the intervals among the values of the factor are constant. I would like to consider these intervals as unequal, I would like to give to SPSS the values that I have found. and I would love SPSS to take into consideration these values for the Factor... :-) Possibly the questions are: 1.- Can SPSS calculate Trend Analysis with ANOVA One-way having a Factor with different intervals? 2.- If SPSS cannot do it with the menu, can SPSS do it with syntax? How? 3.- Is it there some other way to calculate Trend-Analysis? I ask it in case this would be another way to solve the problem. I would like to thank any in advance for all your help, time and suggestions. Kind regards, Giorgio |
|
|
In reply to this post by giorgio de marchis
p.s. - If you want to compute the coefficients yourself consult Roger Kirk's book on experimental design (now in its 3rd edition I believe). He has an appendix that will show you how to do it.
|
Re: Trend Analysis - Anova One-way with intervals that are not equal
|
|
In reply to this post by giorgio de marchis
In SPSS Oneway the metric for the polynomial is the group code. See the
Statistical Algorithms document.
Marija Norusis
In a message dated 4/30/2009 7:52:53 P.M. Central Daylight Time,
[hidden email] writes:
You don't provide enough information to give you a very specific answer but |
| Free forum by Nabble | Edit this page |