Stanford Phd Comparison Table?

[Muenchen, Robert A (Bob)]

Hi All,

 

I was wondering if anyone has had the chance to study this comparison table to determine how accurate it is for SPSS.

http://stanfordphd.com/Statistical_Software.html

 

Thanks,

Bob

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

[Nils Braakmann-2]

I'd say it painted with a bit of a broad brush, but the gist of it seems accurate. I haven't used SPSS for a while now, mainly because it couldn't do essentially everything I needed, but the assessment seems accurate given what I remember (and it's definitely pretty spot-on for R and Stata).

Cheers,

Nils

On Thu, Aug 14, 2014 at 9:19 PM, Muenchen, Robert A (Bob) <[hidden email]> wrote:

Hi All,

 

I was wondering if anyone has had the chance to study this comparison table to determine how accurate it is for SPSS.

http://stanfordphd.com/Statistical_Software.html

 

Thanks,

Bob

===================== 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 Nils Braakmann

10 Alnwick Avenue

Whitley Bay

NE26 3PT

UK

Home [phone]

Office (Newcastle University) [phone]

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

[Alex Reutter]

Hi Bob,

Quickly looking at the table, I would note that:

* Ridge regression and lasso regression are available in CATREG (ridge regression has also long been available as a macro)
* Nonlinear regression is available in NLR and CNLR
* Bootstrap & jackknife are available in BOOTSTRAP
* There is an EM algorithm in MVA
* Without knowing more about what the website author wants for "Longitudinal Data", it's hard to know why they consider SPSS "Limited"
* Similar questions for why they consider SPSS "Limited" for univariate time series
* Multivariate time series can be fit with TSMODEL
* Without knowing more about what the website author wants for "Classification and Regression Trees", it's hard to know why they consider SPSS "Limited"
* I'm not entirely sure what stochastic volatility models the website author is fitting in SPSS

Cheers,
Alex



From:        "Muenchen, Robert A (Bob)" <[hidden email]>
To:        [hidden email]
Date:        08/14/2014 04:21 PM
Subject:        Stanford Phd Comparison Table?
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Hi All,
 
I was wondering if anyone has had the chance to study this comparison table to determine how accurate it is for SPSS.
http://stanfordphd.com/Statistical_Software.html
 
Thanks,
Bob

===================== To manage your subscription to SPSSX-L, send a message to LISTSERV@... (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

[Andy W]

Ditto, I think a more objective surveyor would place limited for some of the R routines, but to continue on...

For path analysis does AMOS count? (it is similar to just buying one of the regression additions for SPSS). Also Stanford has arguably several redundant rows in the table, simultaneous equations are the same thing as path analysis. SPSS can fit simultaneous equations in reduced form ;) (Stata also has path models now). AMOS also has Bayesian analysis, but I imagine you would consider these limited compared to the types of models you can program in BUGS.

SPSS has 2SLS for instrumental variables regression. This and the non-linear regression (see also CURVEFIT, which is a pretty simple interface and fits the most common types of non-linear regressions - which I'm positive Stata can fit as well) I'm confused why the author missed - they have been around along time.

SPSS can also estimate the fast fourier transform (and its inverse), see the CREATE command, so I imagine you could place "Limited" for the wavelet analysis row. SPSS also has an add-on for sample power, although I am not familiar with it.

Finally, Jon has written free add-ons to call many of the same R routines the author lists (Random Forests, SVM, plus others). I wrote a macro for restricted cubic splines.

Now, lets move on to things Stanford missed in the table. As a person who crunches numbers pretty much every day, I don't spend that much time estimating regression equations! I will tell you what I do spend most of my time doing:

 - Grabbing data from various formats (ODBC, csv, Excel)
 - Reshaping data into different forms (e.g. wide to long, long to wide)
 - Creating aggregate summaries of data
 - Doing string manipulations to text fields
 - Making graphs and tables

This is probably 99.9% of my job on a regular basis. I'm not quite sure what insepid situations statisticians get the data in already a convenient and suitable form to conduct analysis without spending the majority of their time simply munging data, I have never been so lucky working with police departments!

SPSS does these very well, and has done them very well since I started using it around 8 years ago. All of the statistical packages mentioned do these now, but 8 years ago this point is arguable...

[Kornbrot, Diana]

IT should also be noted that one can attempt to have best of both worlds by suing the R plug-in for SPSS
I found it straight forward to download and use  the R forest modules.  Does not have all the features of the native R forest procedures [did not give the cut-points numerically, only in the graphs], but it met my needs.
suspect could have done same for other procedures like ridge regression

Teaching R to psychology students is a daunting prospect, although Andy Field has nice text

best

Diana
On 15 Aug 2014, at 13:24, Andy W <[hidden email]> wrote:

> Ditto, I think a more objective surveyor would place limited for some of the
> R routines, but to continue on...
>
> For path analysis does AMOS count? (it is similar to just buying one of the
> regression additions for SPSS). Also Stanford has arguably several redundant
> rows in the table, simultaneous equations are the same thing as path
> analysis. SPSS can fit simultaneous equations in reduced form ;) (Stata also
> has path models now). AMOS also has Bayesian analysis, but I imagine you
> would consider these limited compared to the types of models you can program
> in BUGS.
>
> SPSS has 2SLS for instrumental variables regression. This and the non-linear
> regression (see also CURVEFIT, which is a pretty simple interface and fits
> the most common types of non-linear regressions - which I'm positive Stata
> can fit as well) I'm confused why the author missed - they have been around
> along time.
>
> SPSS can also estimate the fast fourier transform (and its inverse), see the
> CREATE command, so I imagine you could place "Limited" for the wavelet
> analysis row. SPSS also has an add-on for sample power, although I am not
> familiar with it.
>
> Finally, Jon has written free add-ons to call many of the same R routines
> the author lists (Random Forests, SVM, plus others). I wrote a macro for
> restricted cubic splines.
>
> Now, lets move on to things Stanford missed in the table. As a person who
> crunches numbers pretty much every day, I don't spend that much time
> estimating regression equations! I will tell you what I do spend most of my
> time doing:
>
> - Grabbing data from various formats (ODBC, csv, Excel)
> - Reshaping data into different forms (e.g. wide to long, long to wide)
> - Creating aggregate summaries of data
> - Doing string manipulations to text fields
> - Making graphs and tables
>
> This is probably 99.9% of my job on a regular basis. I'm not quite sure what
> insepid situations statisticians get the data in already a convenient and
> suitable form to conduct analysis without spending the majority of their
> time simply munging data, I have never been so lucky working with police
> departments!
>
> SPSS does these very well, and has done them very well since I started using
> it around 8 years ago. All of the statistical packages mentioned do these
> now, but 8 years ago this point is arguable...
>
>
>
> -----
> Andy W
> [hidden email]
> http://andrewpwheeler.wordpress.com/
> --
> View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Stanford-Phd-Comparison-Table-tp5726967p5726975.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

_______________
Professor Diana Kornbrot
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
[phone]
[phone]
[phone]
[hidden email]
http://dianakornbrot.wordpress.com/
http://go.herts.ac.uk/Diana_Kornbrot
skype: kornbrotme_______________________________

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

[Kornbrot, Diana]

ps. also fairly easily implemented routine to get best parameter values for equations of my choice
best

Diana

On 15 Aug 2014, at 13:24, Andy W <[hidden email]> wrote:

> Ditto, I think a more objective surveyor would place limited for some of the
> R routines, but to continue on...
>
> For path analysis does AMOS count? (it is similar to just buying one of the
> regression additions for SPSS). Also Stanford has arguably several redundant
> rows in the table, simultaneous equations are the same thing as path
> analysis. SPSS can fit simultaneous equations in reduced form ;) (Stata also
> has path models now). AMOS also has Bayesian analysis, but I imagine you
> would consider these limited compared to the types of models you can program
> in BUGS.
>
> SPSS has 2SLS for instrumental variables regression. This and the non-linear
> regression (see also CURVEFIT, which is a pretty simple interface and fits
> the most common types of non-linear regressions - which I'm positive Stata
> can fit as well) I'm confused why the author missed - they have been around
> along time.
>
> SPSS can also estimate the fast fourier transform (and its inverse), see the
> CREATE command, so I imagine you could place "Limited" for the wavelet
> analysis row. SPSS also has an add-on for sample power, although I am not
> familiar with it.
>
> Finally, Jon has written free add-ons to call many of the same R routines
> the author lists (Random Forests, SVM, plus others). I wrote a macro for
> restricted cubic splines.
>
> Now, lets move on to things Stanford missed in the table. As a person who
> crunches numbers pretty much every day, I don't spend that much time
> estimating regression equations! I will tell you what I do spend most of my
> time doing:
>
> - Grabbing data from various formats (ODBC, csv, Excel)
> - Reshaping data into different forms (e.g. wide to long, long to wide)
> - Creating aggregate summaries of data
> - Doing string manipulations to text fields
> - Making graphs and tables
>
> This is probably 99.9% of my job on a regular basis. I'm not quite sure what
> insepid situations statisticians get the data in already a convenient and
> suitable form to conduct analysis without spending the majority of their
> time simply munging data, I have never been so lucky working with police
> departments!
>
> SPSS does these very well, and has done them very well since I started using
> it around 8 years ago. All of the statistical packages mentioned do these
> now, but 8 years ago this point is arguable...
>
>
>
> -----
> Andy W
> [hidden email]
> http://andrewpwheeler.wordpress.com/
> --
> View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Stanford-Phd-Comparison-Table-tp5726967p5726975.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

_______________
Professor Diana Kornbrot
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
[phone]
[phone]
[phone]
[hidden email]
http://dianakornbrot.wordpress.com/
http://go.herts.ac.uk/Diana_Kornbrot
skype: kornbrotme_______________________________

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

[Art Kendall]

I also spend much time on
 "
- Grabbing data from various formats (ODBC, csv, Excel)
 - Reshaping data into different forms (e.g. wide to long, long to wide)
 - Doing string manipulations to text fields"

However, other data cleaning and prep tasks include
*making sure all of the data view in complete
*assuring the metadata is clear to the client and to potential jurors
*searching for data anomalies
*rearranging variable order
*drafting transformation syntax
*verifying scoring keys for summative scales
*merging/matching data
*finding duplicate cases
*redrafting syntax for readability and clarity of logic
*determining how to deal with missing data
 "- Making [exploratory] graphs and tables
 - Creating aggregate summaries of data"
* having someone cold read graphs and tables for readablity and communication
* understanding the substantive meaning and distribution of the data
-----
these things can easily take in excess of 90% for the labor time on a project.
Only after all that is done do I do
*statistical analysis
*presentation graphs and tables

SPSS is outstanding for the early steps and very good for the last two.

It is my experience that when someone says that something cannot be done in SPSS they are mistaken.

[Muenchen, Robert A (Bob)]

Thanks to those who commented. I thought Stanford was missing things but I don’t work with many of those allegedly missing tools.

 

Cheers,

Bob

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Alex Reutter
Sent: Thursday, August 14, 2014 7:50 PM
To: [hidden email]
Subject: Re: Stanford Phd Comparison Table?

 

Hi Bob,

Quickly looking at the table, I would note that:

* Ridge regression and lasso regression are available in CATREG (ridge regression has also long been available as a macro)
* Nonlinear regression is available in NLR and CNLR
* Bootstrap & jackknife are available in BOOTSTRAP
* There is an EM algorithm in MVA
* Without knowing more about what the website author wants for "Longitudinal Data", it's hard to know why they consider SPSS "Limited"
* Similar questions for why they consider SPSS "Limited" for univariate time series
* Multivariate time series can be fit with TSMODEL
* Without knowing more about what the website author wants for "Classification and Regression Trees", it's hard to know why they consider SPSS "Limited"
* I'm not entirely sure what stochastic volatility models the website author is fitting in SPSS

Cheers,
Alex



From:        "Muenchen, Robert A (Bob)" <[hidden email]>
To:        [hidden email]
Date:        08/14/2014 04:21 PM
Subject:        Stanford Phd Comparison Table?
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Hi All,
 
I was wondering if anyone has had the chance to study this comparison table to determine how accurate it is for SPSS.
http://stanfordphd.com/Statistical_Software.html
 
Thanks,
Bob

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