Making SPSS better
Locked
123
123
 
|
Dear SPSS
friends,
today I'm interested in your opinion on some of
my thoughts about where to focus when making SPSS better. First thought is that
SPSS should concentrate on making those things better that a lot of people use
and which are far from perfect. Here I have two proposals: 1. Improve the syntax editor. Despite
syntax highlighting which was the last major development I can remember I
cannot help to critisize that we still have to keep struggling with that
antiquated editor. I already write most of my code in Notepad++ which has more
functionality and easens my workflow. But there are some copy paste issues
within the SPSS editor and after a couple of times it won't paste any longer. Also
there is a too many lines issue. So how about giving us a modern syntax editor
that allows us functionality similar to Notepad++ finally? 2. Despite its richness when compared
to former TABLES command the CTABLES module is ageing. The resulting tables
look close to out of fashion and boring. As an example: how amazing would it be
if we could make tables with brand logos in the header and get them exported to Excel? Also I find the
functionality still not flexible enough to fulfil all user requirements, like not
showing redundant rows and also keep them away when exporting to Excel. I have a
second idea which admittedly deviates from my principle of making those things
better that most of us use. It's the macro facility. I know there is Python and
other programming languages and I know they are more powerful than macros. But:
most of my colleagues feel a reservation for this "programming stuff"
which makes standardisation in teams difficult. I learned that there is much
more tolerance for macro syntax which feels like "more manageable".
Unfortunately there is zero development in this area which I find suboptimal. How
do other users think about this? Thanks for
any opinion!
Mario |
Administrator
|
"I learned that there is much more tolerance for macro syntax which feels like "more manageable". Unfortunately there is zero development in this area which I find suboptimal."
Aside from a fix (Thanks Tex) in an odd extra space being put into titles etc (roughly 1995 or 6) there has been NO development in the macro facility since about 1990 ;-) I think people are scared of Bill's code ;-) I don't think DEFINE !ENDDEFINE will ever go away since it is deeply entrenched in the Central system but I don't believe it will ever evolve beyond what it is or isn't. Nonetheless you can still do quite a bit with that Neolithic bear skin and elk skull firestarter.
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
|
I'm all for macros and MATRIX (like David Marso, on'y w/o foul
speach).
I was writing repeatedly to SPSS team, Jon Peck personally (I thank him) etc. I've always said that since they introduced Python and other programmability - get calm and turn now for macro and matrix facilities, to enhance them. Both these have some basic ideas behind, which will never be compensated with Python, an "external" scripting language. 06.07.2016 14:48, David Marso пишет:
"I learned that there is much more tolerance for macro syntax which feels like "more manageable". Unfortunately there is zero development in this area which I find suboptimal." Aside from a fix (Thanks Tex) in an odd extra space being put into titles etc (roughly 1995 or 6) there has been NO development in the macro facility since about 1990 ;-) I think people are scared of Bill's code ;-) I don't think DEFINE !ENDDEFINE will ever go away since it is deeply entrenched in the Central system but I don't believe it will ever evolve beyond what it is or isn't. Nonetheless you can still do quite a bit with that Neolithic bear skin and elk skull firestarter. ----- Please reply to the list and not to my personal email. Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Making-SPSS-better-tp5732642p5732645.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 |
Administrator
|
In reply to this post by spss.giesel@yahoo.de
Hi Mario (and others). Good idea for a thread. Here are some of my thoughts on how to improve SPSS.
1. SPSS is quite slow (compared to its main competitors) at adding new statistical methods. It needs to get quicker at this. And IMO, the new methods should be implemented using SPSS native code, not R-programmability, etc. (When I see the increasing use of R-programmability, it makes me wonder if I should just use R, and cut SPSS out of the equation altogether.) 2. Newer procedures should bring along all of the best features of older procedures they could potentially replace. For example, UNIANOVA, GENLIN, MIXED etc. should allow two or more nested models with a single command (as one can get with multiple /ENTER sub-commands for REGRESSION). And for methods that use MLE, there should be an easy way to get a likelihood ratio test comparing one nested model to the next. All of these modeling procedures should also include the multicollinearity diagnostics (e.g., VIF & Tolerance) from REGRESSION. Generalized VIF (which accomodate categorical variables correctly) would also be a nice addition. 3. Modularity & pricing. I mention these together, because I see them as related. Compared to something like Stata, SPSS is quite expensive, especially if one does not have access to academic pricing. And there is no real comparison when it comes to features. Stata is non-modular, and when you buy it, you get it all (including things like IRT and SEM). (A complete list of features for the current version of Stata can be seen here: http://www.stata.com/features/.) 4. More seamless integration of user-contributed packages would be good. Again, I point to Stata as an example of how this has been done. After a one-time installation of a user-contributed package, it can be used in exactly the same manner as an official Stata command (including access to the help file). 5. The way in which patches & updates are applied could be improved. At the risk of sounding like a Stata promoter, in that program, one simply types -update all- in the command window to check for and install any updates that are available. No need to navigate a labyrinthine website, find the right patch for your version, log in, download the file and manually run it to install the update. Just a simple -update all-. I think something similar to that in SPSS would go over well with users. ;-) 6. As others have suggested macro & matrix need to be updated. For example, matrix is lacking some basic distribution functions, and as a result clumsy kludges are sometimes required. Cheers, Bruce p.s. - The model viewer should be killed off, immediately! ;-)
--
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/). |
|
Too much fancy statistics. To get back to original SPSS survey analysis basics, my two penn'orth is a new command to produce %% tables with epsilon (percentage point difference):
ELABORATE depvar by indvar by testvar /sta eps. . . with whatever other statistics Kalton recommended. John F Hall (Mr) [Retired academic survey researcher] Email: [hidden email] Website: www.surveyresearch.weebly.com SPSS start page: www.surveyresearch.weebly.com/1-survey-analysis-workshop -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver Sent: 06 July 2016 20:01 To: [hidden email] Subject: Re: Making SPSS better Hi Mario (and others). Good idea for a thread. Here are some of my thoughts on how to improve SPSS. 1. SPSS is quite slow (compared to its main competitors) at adding new statistical methods. It needs to get quicker at this. And *IMO*, the new methods should be implemented using SPSS native code, not R-programmability, etc. (When I see the increasing use of R-programmability, it makes me wonder if I should just use R, and cut SPSS out of the equation altogether.) 2. Newer procedures should bring along all of the best features of older procedures they could potentially replace. For example, UNIANOVA, GENLIN, MIXED etc. should allow two or more nested models with a single command (as one can get with multiple /ENTER sub-commands for REGRESSION). And for methods that use MLE, there should be an easy way to get a likelihood ratio test comparing one nested model to the next. All of these modeling procedures should also include the multicollinearity diagnostics (e.g., VIF & Tolerance) from REGRESSION. Generalized VIF (which accomodate categorical variables correctly) would also be a nice addition. 3. Modularity & pricing. I mention these together, because I see them as related. Compared to something like Stata, SPSS is quite expensive, especially if one does not have access to academic pricing. And there is no real comparison when it comes to features. Stata is non-modular, and when you buy it, you get it all (including things like IRT and SEM). (A complete list of features for the current version of Stata can be seen here: http://www.stata.com/features/.) 4. More seamless integration of user-contributed packages would be good. Again, I point to Stata as an example of how this has been done. After a one-time installation of a user-contributed package, it can be used in exactly the same manner as an official Stata command (including access to the help file). 5. The way in which patches & updates are applied could be improved. At the risk of sounding like a Stata promoter, in that program, one simply types -update all- in the command window to check for and install any updates that are available. No need to navigate a labyrinthine website, find the right patch for your version, log in, download the file and manually run it to install the update. Just a simple -update all-. I think something similar to that in SPSS would go over well with users. ;-) 6. As others have suggested macro & matrix need to be updated. For example, matrix is lacking some basic distribution functions, and as a result clumsy kludges are sometimes required. Cheers, Bruce p.s. - The model viewer should be killed off, immediately! ;-) var Mario Giesel-2 wrote > Dear SPSSfriends, today I'm interested in your opinion on some ofmy > thoughts about where to focus when making SPSS better. First thought > is thatSPSS should concentrate on making those things better that a > lot of people useand which are far from perfect. Here I have two proposals: > 1. Improve the syntax editor. Despitesyntax highlighting which > was the last major development I can remember Icannot help to > critisize that we still have to keep struggling with thatantiquated > editor. I already write most of my code in Notepad++ which has > morefunctionality and easens my workflow. But there are some copy > paste issueswithin the SPSS editor and after a couple of times it > won't paste any longer. Alsothere is a too many lines issue. So how > about giving us a modern syntax editorthat allows us functionality similar to Notepad++ finally? > 2. Despite its richness when comparedto former TABLES command > the CTABLES module is ageing. The resulting tableslook close to out of > fashion and boring. As an example: how amazing would it beif we could > make tables with brand logos in the header and get them exported to > Excel? Also I find thefunctionality still not flexible enough to > fulfil all user requirements, like notshowing redundant rows and also > keep them away when exporting to Excel. > > I have asecond idea which admittedly deviates from my principle of > making those thingsbetter that most of us use. It's the macro > facility. I know there is Python andother programming languages and I > know they are more powerful than macros. But:most of my colleagues > feel a reservation for this "programming stuff"which makes > standardisation in teams difficult. I learned that there is muchmore > tolerance for macro syntax which feels like "more > manageable".Unfortunately there is zero development in this area which I find suboptimal. Howdo other users think about this? > Thanks forany opinion! Mario > > > ===================== > To manage your subscription to SPSSX-L, send a message to > LISTSERV@.UGA > (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/Making-SPSS-better-tp5732642p5732648.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 |
|
Administrator
|
In reply to this post by Kirill Orlov
What foul speech?
My post was quite f'ing sanitary ;-) ---
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
|
|
In reply to this post by John F Hall
Agree with so much of this - in particular
> 2. Newer procedures should bring along all of the best features of older procedures they could potentially replace. For example, UNIANOVA, GENLIN, MIXED etc. should allow two or more nested models with a single command (as one can get with multiple /ENTER sub-commands for REGRESSION). And for methods that use MLE, there should be an easy way to get a likelihood ratio test comparing one nested model to the next. All of these modeling procedures should also include the multicollinearity diagnostics (e.g., VIF & Tolerance) from REGRESSION. Generalized VIF (which accomodate categorical variables correctly) would also be a nice addition. > Absolutely agreed - and also back-fill in this way i.e. add the automatic calculation of interactions and contrasts that exists for logistic regression into linear regression. > 5. The way in which patches & updates are applied could be improved. At the risk of sounding like a Stata promoter, in that program, one simply types -update all- in the command window to check for and install any updates that are available. No need to navigate a labyrinthine website, find the right patch for your version, log in, download the file and manually run it to install the update. Just a simple -update all-. I think something similar to that in SPSS would go over well with users. ;-) Ditto - and sort the website out to make finding and downloading patches and plug-ins a simple process > > p.s. - The model viewer should be killed off, immediately! ;-) Yes!!!! > var > > Mario Giesel-2 wrote >> Dear SPSSfriends, today I'm interested in your opinion on some ofmy >> thoughts about where to focus when making SPSS better. First thought >> is thatSPSS should concentrate on making those things better that a >> lot of people useand which are far from perfect. Here I have two proposals: >> 1. Improve the syntax editor. Despitesyntax highlighting which >> was the last major development I can remember Icannot help to >> critisize that we still have to keep struggling with thatantiquated >> editor. I already write most of my code in Notepad++ which has >> morefunctionality and easens my workflow. But there are some copy >> paste issueswithin the SPSS editor and after a couple of times it >> won't paste any longer. Alsothere is a too many lines issue. So how >> about giving us a modern syntax editorthat allows us functionality similar to Notepad++ finally? >> 2. Despite its richness when comparedto former TABLES command >> the CTABLES module is ageing. The resulting tableslook close to out of >> fashion and boring. As an example: how amazing would it beif we could >> make tables with brand logos in the header and get them exported to >> Excel? Also I find thefunctionality still not flexible enough to >> fulfil all user requirements, like notshowing redundant rows and also >> keep them away when exporting to Excel. >> >> I have asecond idea which admittedly deviates from my principle of >> making those thingsbetter that most of us use. It's the macro >> facility. I know there is Python andother programming languages and I >> know they are more powerful than macros. But:most of my colleagues >> feel a reservation for this "programming stuff"which makes >> standardisation in teams difficult. I learned that there is muchmore >> tolerance for macro syntax which feels like "more >> manageable".Unfortunately there is zero development in this area which I find suboptimal. Howdo other users think about this? >> Thanks forany opinion! Mario >> >> >> ===================== >> To manage your subscription to SPSSX-L, send a message to >> LISTSERV@.UGA >> (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/Making-SPSS-better-tp5732642p5732648.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 -- -- Dr Chris Stride, C. Stat, Statistician, Institute of Work Psychology, University of Sheffield Telephone: 0114 2223262 Fax: 0114 2727206 "Figure It Out" Statistical Consultancy and Training Service for Social Scientists Visit www.figureitout.org.uk for details of my consultancy services, and forthcoming training courses, which are also available on an in-house basis: - Data management using SPSS syntax - Advanced SPSS syntax and SPSS macros - Testing for Mediation and Moderation using SPSS - Multi-level Modelling using SPSS - Introduction to Structural Equation Modelling using Mplus - Testing for Mediation and Moderation using Mplus - Multi-level Modelling using Mplus - Latent Growth Curve Modelling using Mplus ===================== 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 John F Hall
Bruce, I like your suggestions.
My penny. I'd like to see nominal regression included in GenLin and GenLinMixed. I'd like to see a parallel lines test (ordinal regression as in Plum) in GenLin and GenLinMixed. I'd like to see, let's call it a standardization, across related routines in terms of a common vocabulary, syntax, keyword assumptions, and symbol (commas, equal signs) processing. This probably seems cryptic. Consider regression and logistic regression. I'll get in trouble for this but the only difference between them is the distribution, which implies some other differences but not necessarily keyword differences, I'll bet. Another is oneway, unianova, and glm. A particularly irksome example is genlinmixed. I'd offer mplus as an extremely compact example. Gene Maguin -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of John F Hall Sent: Wednesday, July 06, 2016 3:17 PM To: [hidden email] Subject: Re: Making SPSS better Too much fancy statistics. To get back to original SPSS survey analysis basics, my two penn'orth is a new command to produce %% tables with epsilon (percentage point difference): ELABORATE depvar by indvar by testvar /sta eps. . . with whatever other statistics Kalton recommended. John F Hall (Mr) [Retired academic survey researcher] Email: [hidden email] Website: www.surveyresearch.weebly.com SPSS start page: www.surveyresearch.weebly.com/1-survey-analysis-workshop -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver Sent: 06 July 2016 20:01 To: [hidden email] Subject: Re: Making SPSS better Hi Mario (and others). Good idea for a thread. Here are some of my thoughts on how to improve SPSS. 1. SPSS is quite slow (compared to its main competitors) at adding new statistical methods. It needs to get quicker at this. And *IMO*, the new methods should be implemented using SPSS native code, not R-programmability, etc. (When I see the increasing use of R-programmability, it makes me wonder if I should just use R, and cut SPSS out of the equation altogether.) 2. Newer procedures should bring along all of the best features of older procedures they could potentially replace. For example, UNIANOVA, GENLIN, MIXED etc. should allow two or more nested models with a single command (as one can get with multiple /ENTER sub-commands for REGRESSION). And for methods that use MLE, there should be an easy way to get a likelihood ratio test comparing one nested model to the next. All of these modeling procedures should also include the multicollinearity diagnostics (e.g., VIF & Tolerance) from REGRESSION. Generalized VIF (which accomodate categorical variables correctly) would also be a nice addition. 3. Modularity & pricing. I mention these together, because I see them as related. Compared to something like Stata, SPSS is quite expensive, especially if one does not have access to academic pricing. And there is no real comparison when it comes to features. Stata is non-modular, and when you buy it, you get it all (including things like IRT and SEM). (A complete list of features for the current version of Stata can be seen here: http://www.stata.com/features/.) 4. More seamless integration of user-contributed packages would be good. Again, I point to Stata as an example of how this has been done. After a one-time installation of a user-contributed package, it can be used in exactly the same manner as an official Stata command (including access to the help file). 5. The way in which patches & updates are applied could be improved. At the risk of sounding like a Stata promoter, in that program, one simply types -update all- in the command window to check for and install any updates that are available. No need to navigate a labyrinthine website, find the right patch for your version, log in, download the file and manually run it to install the update. Just a simple -update all-. I think something similar to that in SPSS would go over well with users. ;-) 6. As others have suggested macro & matrix need to be updated. For example, matrix is lacking some basic distribution functions, and as a result clumsy kludges are sometimes required. Cheers, Bruce p.s. - The model viewer should be killed off, immediately! ;-) var Mario Giesel-2 wrote > Dear SPSSfriends, today I'm interested in your opinion on some ofmy > thoughts about where to focus when making SPSS better. First thought > is thatSPSS should concentrate on making those things better that a > lot of people useand which are far from perfect. Here I have two proposals: > 1. Improve the syntax editor. Despitesyntax highlighting which > was the last major development I can remember Icannot help to > critisize that we still have to keep struggling with thatantiquated > editor. I already write most of my code in Notepad++ which has > morefunctionality and easens my workflow. But there are some copy > paste issueswithin the SPSS editor and after a couple of times it > won't paste any longer. Alsothere is a too many lines issue. So how > about giving us a modern syntax editorthat allows us functionality similar to Notepad++ finally? > 2. Despite its richness when comparedto former TABLES command > the CTABLES module is ageing. The resulting tableslook close to out of > fashion and boring. As an example: how amazing would it beif we could > make tables with brand logos in the header and get them exported to > Excel? Also I find thefunctionality still not flexible enough to > fulfil all user requirements, like notshowing redundant rows and also > keep them away when exporting to Excel. > > I have asecond idea which admittedly deviates from my principle of > making those thingsbetter that most of us use. It's the macro > facility. I know there is Python andother programming languages and I > know they are more powerful than macros. But:most of my colleagues > feel a reservation for this "programming stuff"which makes > standardisation in teams difficult. I learned that there is muchmore > tolerance for macro syntax which feels like "more > manageable".Unfortunately there is zero development in this area which I find suboptimal. Howdo other users think about this? > Thanks forany opinion! Mario > > > ===================== > To manage your subscription to SPSSX-L, send a message to > LISTSERV@.UGA > (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/Making-SPSS-better-tp5732642p5732648.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 |
|
|
In reply to this post by Kirill Orlov
I have advocated for a lot of improvements in MATRIX or a replacement, but it's a big job and has never made it to the top of the wish list. The Python numPy/sciPy libraries or the R matrix capabilities, both of which work in Statistics, provide the missing functionality but, of course, require knowing one of these languages. NumPy is particularly elegant. I would still love to see some MATRIX enhancements, though.
As for MACRO, everyone knows my dislike of it. The trouble with adding to it IMO is that adding some of the missing features that were planned would still leave it as a preprocessor, not a scripting language, so it would never really be satisfactory. The macro has to be fully expanded before any of it is executed, like the C and other preprocessors on which it was modeled. This means that it can't adapt to results or use the metadata. It's not a "Bill" issue. You will be amused to learn, though, that the programmability hooks that enable Python, R, and Java integration rely on the macro infrastructure in the system (but not the macro architecture itself). On Wednesday, July 6, 2016, Kirill Orlov <[hidden email]> wrote:
-- ===================== 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 |
|
|
Hi, Doug, the missing syntax functionality I'm referring to is mainly included in Notepad++ menue Edit. Here you can - via Line operations - Duplicate current line - Move up/down current line - Remove empty lines - Join lines - Sort lines according to various criteria Very important is via Edit - Column Mode - Column mode editing Also useful is via Edit - Blank operations - Trimming leading or trailing spaces Especially useful: - Making newline characters available in search and replace like replacing a space by \n - allow access to regular expressions functionality for an advanced search & replace technology Hope that helps! Thanks, Mario Jon Peck <[hidden email]> schrieb am 22:03 Mittwoch, 6.Juli 2016: I have advocated for a lot of improvements in MATRIX or a replacement, but it's a big job and has never made it to the top of the wish list. The Python numPy/sciPy libraries or the R matrix capabilities, both of which work in Statistics, provide the missing functionality but, of course, require knowing one of these languages. NumPy is particularly elegant. I would still love to see some MATRIX enhancements, though. As for MACRO, everyone knows my dislike of it. The trouble with adding to it IMO is that adding some of the missing features that were planned would still leave it as a preprocessor, not a scripting language, so it would never really be satisfactory. The macro has to be fully expanded before any of it is executed, like the C and other preprocessors on which it was modeled. This means that it can't adapt to results or use the metadata. It's not a "Bill" issue. You will be amused to learn, though, that the programmability hooks that enable Python, R, and Java integration rely on the macro infrastructure in the system (but not the macro architecture itself). On Wednesday, July 6, 2016, Kirill Orlov <[hidden email]> wrote:
-- ===================== 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 |
|
|
" p.s., What do you mean by "redundant rows" in CTABLES? " Hi, Jon, thanks for staying available even during retirement :-) They did not offer us SPSS24 so far - I don't know why. So I don't know about the improvements that took place in that version. But I recently had a post about making compact tables with MR dichotomous variables (0/1) that have different bases. It's not possible via MR Sets to get correct percentages. So I have to build a "normal" custom table by adding the responses by + Then I have to delete the redundant rows showing 0 via OUTPUT MODIFY I export to xlsx via Output EXPORT and the 0 rows are not deleted but only hidden (row height is 0). So I cannot copy paste the results from xlsx to powerpoint but have to first delete rows manually. It feels like too much effort to remove lines I don't need. And definitely not helpful if you want to automate things. Hope that helps, Mario Mario Giesel <[hidden email]> schrieb am 6:04 Donnerstag, 7.Juli 2016: Hi, Doug, the missing syntax functionality I'm referring to is mainly included in Notepad++ menue Edit. Here you can - via Line operations - Duplicate current line - Move up/down current line - Remove empty lines - Join lines - Sort lines according to various criteria Very important is via Edit - Column Mode - Column mode editing Also useful is via Edit - Blank operations - Trimming leading or trailing spaces Especially useful: - Making newline characters available in search and replace like replacing a space by \n - allow access to regular expressions functionality for an advanced search & replace technology Hope that helps! Thanks, Mario Jon Peck <[hidden email]> schrieb am 22:03 Mittwoch, 6.Juli 2016: I have advocated for a lot of improvements in MATRIX or a replacement, but it's a big job and has never made it to the top of the wish list. The Python numPy/sciPy libraries or the R matrix capabilities, both of which work in Statistics, provide the missing functionality but, of course, require knowing one of these languages. NumPy is particularly elegant. I would still love to see some MATRIX enhancements, though. As for MACRO, everyone knows my dislike of it. The trouble with adding to it IMO is that adding some of the missing features that were planned would still leave it as a preprocessor, not a scripting language, so it would never really be satisfactory. The macro has to be fully expanded before any of it is executed, like the C and other preprocessors on which it was modeled. This means that it can't adapt to results or use the metadata. It's not a "Bill" issue. You will be amused to learn, though, that the programmability hooks that enable Python, R, and Java integration rely on the macro infrastructure in the system (but not the macro architecture itself). On Wednesday, July 6, 2016, Kirill Orlov <[hidden email]> wrote:
-- ===================== 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 spss.giesel@yahoo.de
There are also some simple procedures which I don't understand why they haven't been included, and that's inference for proportions. There are so many users who deal with proportions all the time, and even if the chi-square test solves the problem for testing two proportions, there are still no confidence intervals. And also none of the procedures available which are less sensitive to small samples.
It happens every now and then that I suggest my clients to go to http://openepi.com and enter the proportions they get from SPSS in order to get the intervals. Not a very high-tech issue, but regarding what ordinary users work with, this is in my view clearly missing. ===================== 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
Robert Lundqvist
|
|
|
In reply to this post by Bruce Weaver
Jon Peck asked me to explain what I mean. To "elaborate" further (and at the risk of appearing to teach grandmothers to suck eggs) . . Elaboration is a technique for progressively breaking down contingency tables to see what happens to differences on a dependent variable between categories of an independent variable when controlling for categories of a test variable. I'm sure there are plenty of fancy statistical modelling techniques to do this, but beginners need to start with percentages (and charts?) which are easier to understand. At the moment this has to be done with CROSSTABS and epsilon calculated manually (tedious, but simple for students) or with complex CTABLES specifications (far beyond the technical capacity of most students and, I suspect, many researchers). It can be used with percentages or means, but percentages are easier to understand and interpret. On my site there are fully worked examples exploring the difference between the incomes of men and women from paid work. They start at the end of section 2.3 Data Transformations: http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/2.3.1.1__data_transformations.pdf 2.3.1.3 Conditional frequencies exercise Examples are in preparation using data from more recent waves up to 2014 3.2 Three variables http://surveyresearch.weebly.com/32-three-variables.html Introducing a third variable. Controlling for test variables. Elaboration. 3.2.4 Income differences - Elaboration http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/3.2.4b__income_differences__elaboration.pdf This is a detailed work-though which repays thorough study: it is less concerned with statistical techniques and formulae, but concentrates on the process of exploring the substantive social research question, “Is there a difference in income from paid work between men and women, regardless of the effect of other variables?” It produces a series of three-way contingency tables to see what happens to differences in earnings between men and women when controlling for selected test variables. Uses SPSS CROSSTABS depvar BY indvar BY testvar for some tables but also demonstrates how CTABLES produces neater and cleaner output. Epsilon (percentage point difference) is calculated at each level. I am indebted to Jon Peck and Jignesh Sutar for advice and assistance in getting CTABLES to produce the required out in the format I needed. 4.2.1 Income differences – Statistical significance http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/4.2.1_income_differences__statistical_significance.pdf Demonstration, using a two-way contingency table from CROSSTABS, to test the null hypothesis that there is no difference between the earnings (from paid work) of men and women. Step-by-step procedure to produce expected cell values (E) compare them to observed values (O) and gradually build up the equation for chi-square. This is the first time in 650 pages that a statistical formula appears. It’s a lot easier to understand than most textbooks. References: Moser C and Kalton G Survey Methods in Social Investigation (2nd edition, London, Heinemann, 1971) but there are several later editions. Reviewed by Guy Cumberbatch Ergonomics, Volume 16, Issue 1, 1973 http://www.tandfonline.com/doi/abs/10.1080/00140137308928406 Rosenberg M The Logic of Survey Analysis (New York, Basic Books, 1968) Copies available from $0.01 on https://www.amazon.com/Logic-Survey-Analysis-Morris-Rosenberg/dp/0465042058 Reviewed by RE Herriott Am Educ Res J November 1969 vol. 6 no. 4 685-687. http://aer.sagepub.com/content/6/4/685.extract See also Ch 16 in Babbie E The Practice of Social Research (9th edition, Belmont, CA : Wadsworth Thomson Learning, 2001) http://www.worldcat.org/title/practice-of-social-research/oclc/44076041 There's a paradigm of the epsilon statistic on http://www.ebabbie.net/resource/practice/15/paradigm.html and a short definition on p172 (with an example using SPSS) on p297 of Babbie E Adventures in Social Research: Data Analysis Using IBM SPSS Statistics (7th Edition, Pine Forge Press, 2011) Limited preview on https://books.google.fr/books/about/Adventures_in_Social_Research.html?id=7vwMQCxrr2AC&redir_esc=y -----Original Message----- Too much fancy statistics. To get back to original SPSS survey analysis basics, my two penn'orth is a new command to produce %% tables with epsilon (percentage point difference): ELABORATE depvar by indvar by testvar /sta eps. . . with whatever other statistics Kalton recommended. John F Hall (Mr) [Retired academic survey researcher] Email: [hidden email] Website: www.surveyresearch.weebly.com SPSS start page: www.surveyresearch.weebly.com/1-survey-analysis-workshop -----Original Message----- From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Bruce Weaver Sent: 06 July 2016 20:01 To: [hidden email] Subject: Re: Making SPSS better Hi Mario (and others). Good idea for a thread. Here are some of my thoughts on how to improve SPSS. 1. SPSS is quite slow (compared to its main competitors) at adding new statistical methods. It needs to get quicker at this. And *IMO*, the new methods should be implemented using SPSS native code, not R-programmability, etc. (When I see the increasing use of R-programmability, it makes me wonder if I should just use R, and cut SPSS out of the equation altogether.) 2. Newer procedures should bring along all of the best features of older procedures they could potentially replace. For example, UNIANOVA, GENLIN, MIXED etc. should allow two or more nested models with a single command (as one can get with multiple /ENTER sub-commands for REGRESSION). And for methods that use MLE, there should be an easy way to get a likelihood ratio test comparing one nested model to the next. All of these modeling procedures should also include the multicollinearity diagnostics (e.g., VIF & Tolerance) from REGRESSION. Generalized VIF (which accomodate categorical variables correctly) would also be a nice addition. 3. Modularity & pricing. I mention these together, because I see them as related. Compared to something like Stata, SPSS is quite expensive, especially if one does not have access to academic pricing. And there is no real comparison when it comes to features. Stata is non-modular, and when you buy it, you get it all (including things like IRT and SEM). (A complete list of features for the current version of Stata can be seen here: http://www.stata.com/features/.) 4. More seamless integration of user-contributed packages would be good. Again, I point to Stata as an example of how this has been done. After a one-time installation of a user-contributed package, it can be used in exactly the same manner as an official Stata command (including access to the help file). 5. The way in which patches & updates are applied could be improved. At the risk of sounding like a Stata promoter, in that program, one simply types -update all- in the command window to check for and install any updates that are available. No need to navigate a labyrinthine website, find the right patch for your version, log in, download the file and manually run it to install the update. Just a simple -update all-. I think something similar to that in SPSS would go over well with users. ;-) 6. As others have suggested macro & matrix need to be updated. For example, matrix is lacking some basic distribution functions, and as a result clumsy kludges are sometimes required. Cheers, Bruce p.s. - The model viewer should be killed off, immediately! ;-) var Mario Giesel-2 wrote > Dear SPSSfriends, today I'm interested in your opinion on some ofmy > thoughts about where to focus when making SPSS better. First thought > is thatSPSS should concentrate on making those things better that a > lot of people useand which are far from perfect. Here I have two proposals: > 1. Improve the syntax editor. Despitesyntax highlighting which > was the last major development I can remember Icannot help to > critisize that we still have to keep struggling with thatantiquated > editor. I already write most of my code in Notepad++ which has > morefunctionality and easens my workflow. But there are some copy > paste issueswithin the SPSS editor and after a couple of times it > won't paste any longer. Alsothere is a too many lines issue. So how > about giving us a modern syntax editorthat allows us functionality similar to Notepad++ finally? > 2. Despite its richness when comparedto former TABLES command > the CTABLES module is ageing. The resulting tableslook close to out of > fashion and boring. As an example: how amazing would it beif we could > make tables with brand logos in the header and get them exported to > Excel? Also I find thefunctionality still not flexible enough to > fulfil all user requirements, like notshowing redundant rows and also > keep them away when exporting to Excel. > > I have asecond idea which admittedly deviates from my principle of > making those thingsbetter that most of us use. It's the macro > facility. I know there is Python andother programming languages and I > know they are more powerful than macros. But:most of my colleagues > feel a reservation for this "programming stuff"which makes > standardisation in teams difficult. I learned that there is muchmore > tolerance for macro syntax which feels like "more > manageable".Unfortunately there is zero development in this area which I find suboptimal. Howdo other users think about this? > Thanks forany opinion! Mario > > > ===================== > To manage your subscription to SPSSX-L, send a message to > (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 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/Making-SPSS-better-tp5732642p5732648.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 |
|
Administrator
|
I suspect most people with minimal common sense and training would utilize LOG LINEAR models to parse this sort of business. Not so fancy. Civilized people such as myself have been using them for over 30 years. I used them to great effect for my thesis back in 1986.
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
|
|
In reply to this post by Jon Peck
The most sad drawback of SPSS macros is not that they have no access
to metadata (which is also sad) but that it has no access to data
values: you cannot pass (in a handy way) a data value back to the
macro as to become its data (i.e. macro variable). This is
because a macro must fully and "once for all" expand before any
syntax that it "unwraps" can start running. I believe that clever
computer scientists must have already found solutions to overcome
such a limitation. I believe that a macro engine could be designed
to expand and re-expand in portions. Those specific sections that
are declared (by a user writing the macro) to be re-expandible on
macro loops should be able to pick a data value passed to it from a
previous loop and re-expand differently on the next loop, depending
on the value.
06.07.2016 22:58, Jon Peck пишет:
===================== 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 Robert L
Good point, Robert. SPSS is indeed lacking many very basic epidemiological tools. By way of comparison, here are the epidemiological tools available in the latest version of Stata:
http://www.stata.com/features/epidemiology/ Re confidence intervals for proportions, see the list of available types here: http://www.stata.com/features/basic-statistics/ Speaking of confidence intervals, CORRELATIONS should have an option to display confidence intervals. Way back in the 1990s, the APA Task Force on Statistical Inference urged authors to report CIs for measures of effect size, and gave correlations as a specific example. These many years later, despite being the most popular stats package among psychologists, all SPSS has to offer, it seems, is bootstrap CIs (only available to those who have the bootstrap module) and this Technote: http://www-01.ibm.com/support/docview.wss?uid=swg21478368 The lack of CIs for correlations prompted me to write a macro, which is fine as far as it goes. But it lacks the flexibility that one would have if CIs were built in to the CORRELATIONS procedure. I keep hoping that SPSS will (eventually) add CIs, making my macro obsolete. http://www.tqmp.org/Content/vol10-1/p029/p029.pdf https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/my-spss-page/rhoci By the way, if anyone is using the "trick" shown in the YouTube video linked below, please note that this trick gives incorrect results. One way to verify this quickly is to note that it always yields a symmetrical CI, regardless of the value of r. But when using the usual (correct) method of computing the CI for rho, the CI is symmetrical only when r = 0; otherwise, it is asymmetrical, with the wider side toward 0 (see attached image if reading this via Nabble). https://www.youtube.com/watch?v=-dSoWqDyT4E CI_for_rho_as_function_of_r.png 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/). |
|
Administrator
|
In reply to this post by Maguin, Eugene
Good suggestions, Gene. You've reminded me that SPSS still does not have conditional logistic regression. Yes, I know there is a work-around using COXREG. But it seems crazy to me that there is no actual conditional logistic regression option, given how widely it is used in medical & health-related research.
--
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/). |
|
|
There is also a way to test:
H0: RR=1 where RR = ratio of probabilities ...instead of: H0: OR=1 where OR = odds ratio ...via any procedure which can fit a logistic regression in SPSS by applying a "trick" as well. Some other alternatives to test RR include log-binomial and Poisson regression., each of which has its limitations. Side note: I generally prefer OR due to its properties but I know researchers in certain fields want to test probabilities of events directly, as suggested in a previous post. Ryan Sent from my iPhone > On Jul 7, 2016, at 8:21 AM, Bruce Weaver <[hidden email]> wrote: > > Good suggestions, Gene. You've reminded me that SPSS still does not have > conditional logistic regression. Yes, I know there is a work-around using > COXREG. But it seems crazy to me that there is no actual conditional > logistic regression option, given how widely it is used in medical & > health-related research. > > > > Maguin, Eugene wrote >> Bruce, I like your suggestions. >> >> My penny. >> I'd like to see nominal regression included in GenLin and GenLinMixed. >> I'd like to see a parallel lines test (ordinal regression as in Plum) in >> GenLin and GenLinMixed. >> I'd like to see, let's call it a standardization, across related routines >> in terms of a common vocabulary, syntax, keyword assumptions, and symbol >> (commas, equal signs) processing. This probably seems cryptic. Consider >> regression and logistic regression. I'll get in trouble for this but the >> only difference between them is the distribution, which implies some other >> differences but not necessarily keyword differences, I'll bet. Another is >> oneway, unianova, and glm. A particularly irksome example is genlinmixed. >> I'd offer mplus as an extremely compact example. >> >> Gene Maguin >> >> --- snip --- > > > > > > ----- > -- > 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/Making-SPSS-better-tp5732642p5732662.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 David Marso
Yes, but did anyone read it? If so, how many of them were undergraduates in Sociology, Political Science, Social Administration etc. and how many of them understood it? As my old boss, the late Dr Mark Abrams (1906 - 1994) http://surveyresearch.weebly.com/mark-abrams.html (one of the founding fathers of market and survey research) once said, "If it's worth saying, you can say it with percentages!" Today he might have conceded the same for charts in SPSS.
John F Hall (Mr) [Retired academic survey researcher] Email: [hidden email] Website: www.surveyresearch.weebly.com SPSS start page: www.surveyresearch.weebly.com/1-survey-analysis-workshop -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of David Marso Sent: 07 July 2016 12:54 To: [hidden email] Subject: Re: Making SPSS better I suspect most people with minimal common sense and training would utilize LOG LINEAR models to parse this sort of business. Not so fancy. Civilized people such as myself have been using them for over 30 years. I used them to great effect for my thesis back in 1986. John F Hall wrote > Jon Peck asked me to explain what I mean. To "elaborate" further (and > at the risk of appearing to teach grandmothers to suck eggs) . . > > Elaboration is a technique for progressively breaking down contingency > tables to see what happens to differences on a dependent variable > between categories of an independent variable when controlling for > categories of a test variable. I'm sure there are plenty of fancy > statistical modelling techniques to do this, but beginners need to > start with percentages (and > charts?) which are easier to understand. At the moment this has to be > done with CROSSTABS and epsilon calculated manually (tedious, but > simple for students) or with complex CTABLES specifications (far > beyond the technical capacity of most students and, I suspect, many > researchers). It can be used with percentages or means, but > percentages are easier to understand and interpret. > > On my site there are fully worked examples exploring the difference > between the incomes of men and women from paid work. They start at > the end of section 2.3 Data Transformations: > > <http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/2.3.1.1__ > data_transformations.pdf> > http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/2.3.1.1__data > _transformations.pdf > > 2.3.1.3 Conditional frequencies exercise Research question: What is > the distribution of respondents' personal gross income from paid work? > What shape does the distribution have? What is the distribution for > women only? What is the distribution for men only? > Are there any differences? > > 2.3.1.4 Specimen answers for exercise 2.3.1.3 (Conditional > frequencies) This is the final draft of a very extended and detailed > exercise: it's quite long (26 pages) but manageable. It uses data > from the 1986 British Social Attitudes survey as an exemplar in > preparation for moving from analysing one variable to analysing two or more variables. > > 2.3.1.5 Conditional frequencies homework [BSA89] Repeats exercise > 2.3.1.3 on the same variables from the 1989 British Social Attitudes > survey. > > [Specimen answers for 2.3.1.5] > 2.3.1.6.1 Specimen answer for conditional frequencies homework [Tasks > 1 and 2] > 2.3.1.6.2 Specimen answer for conditional frequencies homework [Tasks > 3 and 4] > > > Examples are in preparation using data from more recent waves up to > 2014 > > 3.2 Three variables > http://surveyresearch.weebly.com/32-three-variables.html > Introducing a third variable. Controlling for test variables. Elaboration. > Logical model is X → Y . T (the effect of X on Y controlling for T) > where > > Y = Dependent variable > X = Independent variable > T = Test variable > > 3.2.4 Income differences - Elaboration > http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/3.2.4b__incom > e_differences__elaboration.pdf This is a detailed work-though which > repays thorough study: it is less concerned with statistical > techniques and formulae, but concentrates on the process of exploring > the substantive social research question, “Is there a difference in > income from paid work between men and women, regardless of the effect > of other variables?” > > It produces a series of three-way contingency tables to see what > happens to differences in earnings between men and women when controlling for > selected test variables. Uses SPSS CROSSTABS depvar BY indvar BY > testvar for some tables but also demonstrates how CTABLES produces > neater and cleaner output. Epsilon (percentage point difference) is > calculated at each level. > > I am indebted to Jon Peck and Jignesh Sutar for advice and assistance > in getting CTABLES to produce the required out in the format I needed. > > 4.2.1 Income differences – Statistical significance > http://surveyresearch.weebly.com/uploads/2/9/9/8/2998485/4.2.1_income_ > differences__statistical_significance.pdf > (Not specifically elaboration, but includes epsilon calculations) > Demonstration, using a two-way contingency table from CROSSTABS, to > test the null hypothesis that there is no difference between the > earnings (from paid work) of men and women. Step-by-step procedure to > produce expected cell values (E) compare them to observed values (O) > and gradually build up the equation for chi-square. This is the first > time in 650 pages that a statistical formula appears. It’s a lot > easier to understand than most textbooks. > > References: > > Moser C and Kalton G > Survey Methods in Social Investigation (2nd edition, London, > Heinemann, 1971) but there are several later editions. > Reviewed by Guy Cumberbatch > Ergonomics, Volume 16, Issue 1, 1973 > <http://www.tandfonline.com/doi/abs/10.1080/00140137308928406> > http://www.tandfonline.com/doi/abs/10.1080/00140137308928406 > > Rosenberg M > The Logic of Survey Analysis > (New York, Basic Books, 1968) > Copies available from $0.01 on > > <https://www.amazon.com/Logic-Survey-Analysis-Morris-Rosenberg/dp/0 > 465042058> > https://www.amazon.com/Logic-Survey-Analysis-Morris-Rosenberg/dp/04650 > 42058 > Reviewed by RE Herriott > Am Educ Res J November 1969 vol. 6 no. 4 685-687. > <http://aer.sagepub.com/content/6/4/685.extract> > http://aer.sagepub.com/content/6/4/685.extract > > See also Ch 16 in Babbie E > The Practice of Social Research > (9th edition, Belmont, CA : Wadsworth Thomson Learning, 2001) > > <http://www.worldcat.org/title/practice-of-social-research/oclc/440 > 76041> > http://www.worldcat.org/title/practice-of-social-research/oclc/4407604 > 1 > > There's a paradigm of the epsilon statistic on > http://www.ebabbie.net/resource/practice/15/paradigm.html and a short > definition on p172 (with an example using SPSS) on p297 of Babbie E > Adventures in Social Research: Data Analysis Using IBM SPSS Statistics > (7th Edition, Pine Forge Press, 2011) Limited preview on > https://books.google.fr/books/about/Adventures_in_Social_Research.html > ?id=7vwMQCxrr2AC > <https://books.google.fr/books/about/Adventures_in_Social_Research. > html?id=7vwMQCxrr2AC&redir_esc=y> > &redir_esc=y > > -----Original Message----- > From: John F Hall [mailto: > johnfhall@ > ] > Sent: 06 July 2016 21:17 > To: ' > SPSSX-L@.UGA > ' < > SPSSX-L@.UGA > > > Subject: RE: Making SPSS better > > Too much fancy statistics. To get back to original SPSS survey > analysis basics, my two penn'orth is a new command to produce %% > tables with epsilon (percentage point difference): > > ELABORATE depvar by indvar by testvar /sta eps. > > . . with whatever other statistics Kalton recommended. > > John F Hall (Mr) > [Retired academic survey researcher] > > Email: <mailto: > johnfhall@ > > > johnfhall@ > > Website: <http://www.surveyresearch.weebly.com> > www.surveyresearch.weebly.com SPSS start page: > <http://www.surveyresearch.weebly.com/1-survey-analysis-workshop> > ; www.surveyresearch.weebly.com/1-survey-analysis-workshop > > > > > -----Original Message----- > From: SPSSX(r) Discussion [ <mailto: > SPSSX-L@.UGA > > mailto: > SPSSX-L@.UGA > ] On Behalf Of Bruce Weaver > Sent: 06 July 2016 20:01 > To: <mailto: > SPSSX-L@.UGA > > > SPSSX-L@.UGA > Subject: Re: Making SPSS better > > Hi Mario (and others). Good idea for a thread. Here are some of my > thoughts on how to improve SPSS. > > 1. SPSS is quite slow (compared to its main competitors) at adding new > statistical methods. It needs to get quicker at this. And *IMO*, the > new methods should be implemented using SPSS native code, not > R-programmability, etc. (When I see the increasing use of > R-programmability, it makes me wonder if I should just use R, and cut > SPSS out of the equation altogether.) > > 2. Newer procedures should bring along all of the best features of > older procedures they could potentially replace. For example, > UNIANOVA, GENLIN, MIXED etc. should allow two or more nested models > with a single command (as one can get with multiple /ENTER > sub-commands for REGRESSION). And for methods that use MLE, there > should be an easy way to get a likelihood ratio test comparing one > nested model to the next. All of these modeling procedures should > also include the multicollinearity diagnostics (e.g., VIF & Tolerance) > from REGRESSION. Generalized VIF (which accomodate categorical variables correctly) would also be a nice addition. > > 3. Modularity & pricing. I mention these together, because I see them > as related. Compared to something like Stata, SPSS is quite > expensive, especially if one does not have access to academic pricing. > And there is no real comparison when it comes to features. Stata is > non-modular, and when you buy it, you get it all (including things > like IRT and SEM). (A complete list of features for the current > version of Stata can be seen > here: > <http://www.stata.com/features/> > http://www.stata.com/features/.) > > 4. More seamless integration of user-contributed packages would be good. > Again, I point to Stata as an example of how this has been done. > After a one-time installation of a user-contributed package, it can be > used in exactly the same manner as an official Stata command > (including access to the help file). > > 5. The way in which patches & updates are applied could be improved. > At the risk of sounding like a Stata promoter, in that program, one > simply types -update all- in the command window to check for and > install any updates that are available. No need to navigate a > labyrinthine website, find the right patch for your version, log in, > download the file and manually run it to install the update. Just a > simple -update all-. I think something similar to that in SPSS would go over well with users. > ;-) > > 6. As others have suggested macro & matrix need to be updated. For > example, matrix is lacking some basic distribution functions, and as a > result clumsy kludges are sometimes required. > > Cheers, > Bruce > > p.s. - The model viewer should be killed off, immediately! ;-) > > var > > Mario Giesel-2 wrote >> Dear SPSSfriends, today I'm interested in your opinion on some ofmy >> thoughts about where to focus when making SPSS better. First thought >> is thatSPSS should concentrate on making those things better that a >> lot of people useand which are far from perfect. Here I have two >> proposals: >> 1. Improve the syntax editor. Despitesyntax highlighting which >> was the last major development I can remember Icannot help to >> critisize that we still have to keep struggling with thatantiquated >> editor. I already write most of my code in Notepad++ which has >> morefunctionality and easens my workflow. But there are some copy >> paste issueswithin the SPSS editor and after a couple of times it >> won't paste any longer. Alsothere is a too many lines issue. So how >> about giving us a modern syntax editorthat allows us functionality >> similar to Notepad++ finally? >> 2. Despite its richness when comparedto former TABLES command >> the CTABLES module is ageing. The resulting tableslook close to out >> of fashion and boring. As an example: how amazing would it beif we >> could make tables with brand logos in the header and get them >> exported to Excel? Also I find thefunctionality still not flexible >> enough to fulfil all user requirements, like notshowing redundant >> rows and also keep them away when exporting to Excel. >> >> I have asecond idea which admittedly deviates from my principle of >> making those thingsbetter that most of us use. It's the macro >> facility. I know there is Python andother programming languages and I >> know they are more powerful than macros. But:most of my colleagues >> feel a reservation for this "programming stuff"which makes >> standardisation in teams difficult. I learned that there is muchmore >> tolerance for macro syntax which feels like "more >> manageable".Unfortunately there is zero development in this area >> which I find suboptimal. Howdo other users think about this? >> Thanks forany opinion! Mario >> >> >> ===================== >> To manage your subscription to SPSSX-L, send a message to > >> <mailto:LISTSERV@.UGA> LISTSERV@.UGA > >> (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 > <mailto: > bweaver@ > > > bweaver@ > <http://sites.google.com/a/lakeheadu.ca/bweaver/> > 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/Making-SPSS-better-t > p5732642p5732648.html> > http://spssx-discussion.1045642.n5.nabble.com/Making-SPSS-better-tp573 > 2642p5732648.html Sent from the SPSSX Discussion mailing list archive > at Nabble.com. > > ===================== > To manage your subscription to SPSSX-L, send a message to <mailto: > LISTSERV@.UGA > > > LISTSERV@.UGA > (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 > LISTSERV@.UGA > (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 ----- Please reply to the list and not to my personal email. Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Making-SPSS-better-tp5732642p5732659.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 |
|
Administrator
|
John, I understand that you are trying to make a point about the usefulness of basic descriptive statistics, cross-tabulations and graphs. No one disagrees with that, and frequent reminders of their usefulness are probably helpful. But the usefulness of those things does not negate the usefulness of more advanced methods. We need both.
Re David's thesis, notice that he indicated neither the discipline nor the level (undergrad, Masters, PhD) of the thesis. I hope that you are not seriously suggesting that every thesis, at every level and in every discipline should be written in such a way that it can be easily understood by undergraduates in Sociology, Political Science, Social Administration etc. That's just silly.
--
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/). |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |