Panel analysis
Locked
 
|
Dear Listers
I have economic developement and institutional type variables which are fairly static over a number of countries and a number of years. Thus I am assuming that the fixed effect specification is not adequate. I wish to find out if there is test in SPSS that determines whether I should go for random effects or a simple pooled OLS regression. Apparently STATA has a sort of LM test in this regard.
Colin This email and all contents are subject to the following disclaimer: http://disclaimer.uj.ac.za |
Administrator
|
Have you looked at the documentation for MIXED? I'm not sure what an LM test is, but with MIXED, you can get a likelihood ratio test on the change in model fit (for nested models) by using the change in 2-LL. Change in -2LL = chi-square with df = difference in number of parameters. (Bear in mind that it is crucial that the two models use exactly the same cases from the data file.)
HTH.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
In reply to this post by Reddy, Colin
Colin,
Note that you should generally use ML estimation instead of REML estimation when conducting likelihood ratio tests. Ryan On Wed, Jul 20, 2011 at 8:38 AM, Reddy, Colin <[hidden email]> wrote: > Dear Listers > > > > I have economic developement and institutional type variables which are > fairly static over a number of countries and a number of years. Thus I am > assuming that the fixed effect specification is not adequate. I wish to find > out if there is test in SPSS that determines whether I should go for random > effects or a simple pooled OLS regression. Apparently STATA has a sort of LM > test in this regard. > > > > Colin > > ________________________________ > This email and all contents are subject to the following disclaimer: > > http://disclaimer.uj.ac.za > ===================== 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 |
|
Administrator
|
I've read (in Jos Twisk's introductory book on multilevel models, I think) that ML and REML tend to yield better estimates of the fixed and random effects respectively. The author reckoned that most folks are probably more interested in the fixed effects than the random effects, and therefore recommended ML estimation in most cases. But I don't remember coming across this point about likelihood ratio tests. Do you have a reference for that, Ryan? Makes one wonder why REML is the default for MIXED.
Thanks, Bruce
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
Bruce,
As far as I'm aware, it is standard practice to fit mixed models by employing restricted maximum likelihood estimation (hence REML is default in SPSS and SAS). That said, for likelihood ratio tests, there are certain circumstances under which ML is preferred, such as conducting likelihood ratio tests on fixed effects. Now, if you were testing for differences between random effects, then you could use REML. I should have really made this point during my last post. Sorry. Getting late here. I can write back with more details tomorrow and perhaps references. Certainly interested in hearing if others disagree. Ryan On Thu, Jul 21, 2011 at 9:33 PM, Bruce Weaver <[hidden email]> wrote: > I've read (in Jos Twisk's introductory book on multilevel models, I think) > that ML and REML tend to yield better estimates of the fixed and random > effects respectively. The author reckoned that most folks are probably more > interested in the fixed effects than the random effects, and therefore > recommended ML estimation in most cases. But I don't remember coming across > this point about likelihood ratio tests. Do you have a reference for that, > Ryan? Makes one wonder why REML is the default for MIXED. > > Thanks, > Bruce > > > > R B wrote: >> >> Colin, >> >> Note that you should generally use ML estimation instead of REML >> estimation when conducting likelihood ratio tests. >> >> Ryan >> >> On Wed, Jul 20, 2011 at 8:38 AM, Reddy, Colin <[hidden email]> >> wrote: >>> Dear Listers >>> >>> >>> >>> I have economic developement and institutional type variables which are >>> fairly static over a number of countries and a number of years. Thus I am >>> assuming that the fixed effect specification is not adequate. I wish to >>> find >>> out if there is test in SPSS that determines whether I should go for >>> random >>> effects or a simple pooled OLS regression. Apparently STATA has a sort of >>> LM >>> test in this regard. >>> >>> >>> >>> Colin >>> >>> ________________________________ >>> This email and all contents are subject to the following disclaimer: >>> >>> http://disclaimer.uj.ac.za >>> >> >> ===================== >> 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 > [hidden email] > http://sites.google.com/a/lakeheadu.ca/bweaver/ > > "When all else fails, RTFM." > > NOTE: My Hotmail account is not monitored regularly. > To send me an e-mail, please use the address shown above. > > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Panel-analysis-tp4615758p4621460.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 |
|
|
In reply to this post by Bruce Weaver
Bruce,
Singer & Willet have a short but nice discussion of ML vs REML in their book Applied Longitudinal Data Analysis p87 et seq. In their chapter on selecting covariance structures they say "Because [each model] has identical fixed effects we could have used either full or restricted methods to comopare models. We chose restricted methods becuase the obtained goodness-of-fit statistics then reflect only the fit of the model's stochatsic portion whioch is our docus here." HTH Regards Garruy Gelade -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver Sent: 22 July 2011 02:33 To: [hidden email] Subject: Re: Panel analysis I've read (in Jos Twisk's introductory book on multilevel models, I think) that ML and REML tend to yield better estimates of the fixed and random effects respectively. The author reckoned that most folks are probably more interested in the fixed effects than the random effects, and therefore recommended ML estimation in most cases. But I don't remember coming across this point about likelihood ratio tests. Do you have a reference for that, Ryan? Makes one wonder why REML is the default for MIXED. Thanks, Bruce R B wrote: > > Colin, > > Note that you should generally use ML estimation instead of REML > estimation when conducting likelihood ratio tests. > > Ryan > > On Wed, Jul 20, 2011 at 8:38 AM, Reddy, Colin <[hidden email]> > wrote: >> Dear Listers >> >> >> >> I have economic developement and institutional type variables which are >> fairly static over a number of countries and a number of years. Thus I am >> assuming that the fixed effect specification is not adequate. I wish to >> find >> out if there is test in SPSS that determines whether I should go for >> random >> effects or a simple pooled OLS regression. Apparently STATA has a sort of >> LM >> test in this regard. >> >> >> >> Colin >> >> ________________________________ >> This email and all contents are subject to the following disclaimer: >> >> http://disclaimer.uj.ac.za >> > > ===================== > 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 [hidden email] http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." NOTE: My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Panel-analysis-tp4615758p46214 60.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 |
Reliability coefficients for continuous scores ranging between 0 and 100
|
|
In reply to this post by Ryan
|
Re: Reliability coefficients for continuous scores ranging between 0 and 100
|
|
Eins Your four items are not independent because they sum to 100. You are possibly getting a negative average covariance because a high average score on any three items necessarily leads to a low score on the remaining item. Cronbach’s alpha is not appropriate for this kind of data. Garry From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Eins Bernardo
|
|
Administrator
|
In reply to this post by Garry Gelade
Thanks Garry. I have that book, and really like it. I guess that bit didn't sink in the first time I read it. I'll have to look at it again, and add a bit to my notes. Thanks to Ryan for his response too.
Cheers, Bruce
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
In reply to this post by Ryan
As I understand it, REML has less unbiased estimates of random effects than ML but does not take into account fixed effects.\\Paul
Dr. Paul R. Swank, Professor Children's Learning Institute University of Texas Health Science Center-Houston -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of R B Sent: Friday, July 22, 2011 12:49 AM To: [hidden email] Subject: Re: Panel analysis Bruce, As far as I'm aware, it is standard practice to fit mixed models by employing restricted maximum likelihood estimation (hence REML is default in SPSS and SAS). That said, for likelihood ratio tests, there are certain circumstances under which ML is preferred, such as conducting likelihood ratio tests on fixed effects. Now, if you were testing for differences between random effects, then you could use REML. I should have really made this point during my last post. Sorry. Getting late here. I can write back with more details tomorrow and perhaps references. Certainly interested in hearing if others disagree. Ryan On Thu, Jul 21, 2011 at 9:33 PM, Bruce Weaver <[hidden email]> wrote: > I've read (in Jos Twisk's introductory book on multilevel models, I think) > that ML and REML tend to yield better estimates of the fixed and random > effects respectively. The author reckoned that most folks are probably more > interested in the fixed effects than the random effects, and therefore > recommended ML estimation in most cases. But I don't remember coming across > this point about likelihood ratio tests. Do you have a reference for that, > Ryan? Makes one wonder why REML is the default for MIXED. > > Thanks, > Bruce > > > > R B wrote: >> >> Colin, >> >> Note that you should generally use ML estimation instead of REML >> estimation when conducting likelihood ratio tests. >> >> Ryan >> >> On Wed, Jul 20, 2011 at 8:38 AM, Reddy, Colin <[hidden email]> >> wrote: >>> Dear Listers >>> >>> >>> >>> I have economic developement and institutional type variables which are >>> fairly static over a number of countries and a number of years. Thus I am >>> assuming that the fixed effect specification is not adequate. I wish to >>> find >>> out if there is test in SPSS that determines whether I should go for >>> random >>> effects or a simple pooled OLS regression. Apparently STATA has a sort of >>> LM >>> test in this regard. >>> >>> >>> >>> Colin >>> >>> ________________________________ >>> This email and all contents are subject to the following disclaimer: >>> >>> http://disclaimer.uj.ac.za >>> >> >> ===================== >> 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 > [hidden email] > http://sites.google.com/a/lakeheadu.ca/bweaver/ > > "When all else fails, RTFM." > > NOTE: My Hotmail account is not monitored regularly. > To send me an e-mail, please use the address shown above. > > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Panel-analysis-tp4615758p4621460.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 |
Re: Reliability coefficients for continuous scores ranging between 0 and 100
|
|
In reply to this post by Garry Gelade
- and that is a secondary reason why it is generally a bad
idea to elicit data scores as forced rankings, or other methods that add up to a fixed sum. (The main reason is that you don't ask for any anchor of absolute good or bad, which you usually want to know.) -- Rich Ulrich Date: Fri, 22 Jul 2011 11:09:15 +0100 From: [hidden email] Subject: Re: Reliability coefficients for continuous scores ranging between 0 and 100 To: [hidden email] Eins
Your four items are not independent because they sum to 100. You are possibly getting a negative average covariance because a high average score on any three items necessarily leads to a low score on the remaining item. Cronbach’s alpha is not appropriate for this kind of data.
Garry From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Eins Bernardo
|
|
|
In reply to this post by Bruce Weaver
Bruce,
Below are a couple excerpts from the Mixed Models Theory section in the SAS 9.2 User's Guide. Regarding random effects comparison of nested models: "A better alternative is the likelihood ratio statistic. This statistic compares two covariance models, one a special case of the other. To compute it, you must run PROC MIXED twice, once for each of the two models, and then subtract the corresponding values of times the log likelihoods. You can use either ML or REML to construct this statistic, which tests whether the full model is necessary beyond the reduced model." Regarding fixed effects comparison of nested models: "An alternative is the statistic associated with the likelihood ratio test. This statistic compares two fixed-effects models, one a special case of the other. It is computed just as when comparing different covariance models, although you should use ML and not REML here because the penalty term associated with restricted likelihoods depends upon the fixed-effects specification." HTH, Ryan On Fri, Jul 22, 2011 at 7:19 AM, Bruce Weaver <[hidden email]> wrote: > Thanks Garry. I have that book, and really like it. I guess that bit didn't > sink in the first time I read it. I'll have to look at it again, and add a > bit to my notes. Thanks to Ryan for his response too. > > Cheers, > Bruce > > > > Garry Gelade wrote: >> >> Bruce, >> >> Singer & Willet have a short but nice discussion of ML vs REML in their >> book >> Applied Longitudinal Data Analysis p87 et seq. >> >> In their chapter on selecting covariance structures they say "Because >> [each >> model] has identical fixed effects we could have used either full or >> restricted methods to comopare models. We chose restricted methods becuase >> the obtained goodness-of-fit statistics then reflect only the fit of the >> model's stochatsic portion whioch is our docus here." >> >> HTH >> >> Regards >> Garruy Gelade >> >> -----Original Message----- >> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of >> Bruce Weaver >> Sent: 22 July 2011 02:33 >> To: [hidden email] >> Subject: Re: Panel analysis >> >> I've read (in Jos Twisk's introductory book on multilevel models, I think) >> that ML and REML tend to yield better estimates of the fixed and random >> effects respectively. The author reckoned that most folks are probably >> more >> interested in the fixed effects than the random effects, and therefore >> recommended ML estimation in most cases. But I don't remember coming >> across >> this point about likelihood ratio tests. Do you have a reference for >> that, >> Ryan? Makes one wonder why REML is the default for MIXED. >> >> Thanks, >> Bruce >> >> >> >> R B wrote: >>> >>> Colin, >>> >>> Note that you should generally use ML estimation instead of REML >>> estimation when conducting likelihood ratio tests. >>> >>> Ryan >>> >>> On Wed, Jul 20, 2011 at 8:38 AM, Reddy, Colin <[hidden email]> >>> wrote: >>>> Dear Listers >>>> >>>> >>>> >>>> I have economic developement and institutional type variables which are >>>> fairly static over a number of countries and a number of years. Thus I >>>> am >>>> assuming that the fixed effect specification is not adequate. I wish to >>>> find >>>> out if there is test in SPSS that determines whether I should go for >>>> random >>>> effects or a simple pooled OLS regression. Apparently STATA has a sort >>>> of >>>> LM >>>> test in this regard. >>>> >>>> >>>> >>>> Colin >>>> >>>> ________________________________ >>>> This email and all contents are subject to the following disclaimer: >>>> >>>> http://disclaimer.uj.ac.za >>>> >>> >>> ===================== >>> 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 >> [hidden email] >> http://sites.google.com/a/lakeheadu.ca/bweaver/ >> >> "When all else fails, RTFM." >> >> NOTE: My Hotmail account is not monitored regularly. >> To send me an e-mail, please use the address shown above. >> >> -- >> View this message in context: >> http://spssx-discussion.1045642.n5.nabble.com/Panel-analysis-tp4615758p46214 >> 60.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 >> > > > ----- > -- > Bruce Weaver > [hidden email] > http://sites.google.com/a/lakeheadu.ca/bweaver/ > > "When all else fails, RTFM." > > NOTE: My Hotmail account is not monitored regularly. > To send me an e-mail, please use the address shown above. > > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Panel-analysis-tp4615758p4622697.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 |
Re: Reliability coefficients for continuous scores ranging between 0 and 100
|
|
In reply to this post by Rich Ulrich
|
Re: Reliability coefficients for continuous scores ranging between 0 and 100
|
|
This is another illustration of the saying, that the time to plan
your statistics is BEFORE you collect the data. I can't say whether you have any form of reliability possible, if you just have four assessments that add up to 100. I will add some detail here - Cronbach's alpha gives "internal reliability" by estimating the reliability of the TOTAL score, using the cor relations among the items. Since your total is always 100, it is meaningless as a measure for individuals; so alpha has no meaning for it. If one item is usually very large (more than half?), you might look at the correlations among the others. Or their alpha. If it is good, then you have "something", but it doesn't sound like you necessarily have any "latent factor" -- in which case, alpha can't be useful at all. Are there any two measures that should correlate? Pairwise correlations, while controlling for the largest one out? (partial correlation - a*b partialing c,d.) Other forms of reliability exist, which use other pieces of information. If an item predicts something well, that is Predictive reliability. If an item correlates with similar measures, that is convergent reliability. - Test-retest is *always* nice, even when there may be time or treatment intervening. -- Rich Ulrich Date: Sat, 23 Jul 2011 12:40:53 +0800 From: [hidden email] Subject: Re: Reliability coefficients for continuous scores ranging between 0 and 100 To: [hidden email]
|
|
|
In reply to this post by Ryan
Can someone point me to a thread that may contain any strange quirks in SPSS 19.0. and potential fixes.
So far, i've found these: 1) Processing time - A simple frequency table with 1000 cases has a delay in processing. I quiver to think what is going to happen when i try to run a mixed model with 14,000 cases 2) Print preview - some charts don't show up in print preview or can't be printed in SPSS. i had to export them to Word. There may be others that i haven't discovered yet but since I am late to the party on upgrading I suspect that these issues have been discussed previously and i ignored them. My search of the archives has not found these issues. Thank you. Carol ===================== 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 |
|
Administrator
|
In reply to this post by Ryan
Thanks Ryan, and thank you again to Garry. I finally got around to revisiting Singer & Willett's discussion of this (starting on p. 87). Here is a nice excerpt from the final paragraph in that section.
"An important issue is that goodness-of-fit statistics computed using the two methods (introduced in section 4.6) refer to different portions of the model. Under FML [i.e., full maximum likelihood, or ML in SPSS], they describe the fit of the entire model; under RML [restricted ML, or REML in SPSS], they describe the fit of only the stochastic portion (the random effects). This means that the goodness-of-fit statistics from FML can be used to test hypotheses about any type of parameter, either a fixed effect or a variance component, but those from RML can be used only to test hypotheses about variance components (not the fixed effects). This distinction has profound implications for hypothesis testing as a component of model building and data analysis (as we will soon describe). When we compare models that differ only in their variance components, we can use either method. When we compare models that differ in both fixed effects and variance components, we must use full information methods." (p. 90) For completeness, I would add that when comparing models that differ only in fixed effects, we must use ML rather than REML.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
In reply to this post by parisec
Hi Carol,
I can't speak to the issues you've listed, but have you installed the updates and fixes to SPSS 19.0? See the link below. http://www-947.ibm.com/support/entry/portal/Recommended_fix/Software/Information_Management/SPSS_Statistics You should install SPSS Statistics 19.0 Fix Pack 1 and then install SPSS Statistics 19.0 Fix Pack 1 Interim Fix 3. The second of the files above was released 4/15/2011. After that, let the list know whether you're having the same issues. Hope this helps, Ariel On Mon, Jul 25, 2011 at 1:41 PM, Parise, Carol A. <[hidden email]> wrote: > > Can someone point me to a thread that may contain any strange quirks in SPSS 19.0. and potential fixes. > > So far, i've found these: > > 1) Processing time - A simple frequency table with 1000 cases has a delay in processing. I quiver to think what is going to happen when i try to run a mixed model with 14,000 cases > > 2) Print preview - some charts don't show up in print preview or can't be printed in SPSS. i had to export them to Word. > > > There may be others that i haven't discovered yet but since I am late to the party on upgrading I suspect that these issues have been discussed previously and i ignored them. My search of the archives has not found these issues. > > Thank you. > > Carol > > ===================== > 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 |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |