SPSSX Discussion

History of measurment levels in SPSS

Classic

List

Threaded

20 messages Options

John F Hall

History of measurment levels in SPSS

I’m working, and writing a commentary for newbies, on the 1975 SSRC Survey Unit Multipurpose Survey (the first of its kind in the UK) recently uploaded at my request, but originally deposited at UKDS (Essex) by the late Prof. Cathie Marsh, my then colleague and trainee researcher. (See http://discover.ukdataservice.ac.uk/catalogue/?sn=680&type=Data%20catalogue#documentation).

Measurement levels originally specified, if any, have been lost during processing at UKDS (apparently normal as they produce Stata and *.tab versions as well, plus documentation). The ones in this file are those automatically assigned by SPSS as Nominal or Scale, depending on the range between the minimum and maximum encountered. Scrolling up and down the Level column there are no variables specified as Ordinal. They all need to be checked and corrected.

Can anyone remember if variable level specifications were available as long ago as that? I can only remember integer and alpha from INPUT FORMAT commands.

Variable names are all lower case: variable and value labels are a mixture of mostly UPPER and a few lower case. Can anyone remember when lower/mixed case was first allowed?

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

Art Kendall

Re: History of measurment levels in SPSS

The change may have occurred at different time on different platforms.
For example, the 1972 version of SPSS on DEC 10s, did not require 16 spaces to indicate a continuation line. An ASCII <tab> could be used. On these machines card images were treated exactly as if they were input from a teletype, i.e, a carriage return/linefeed pair indicated a new line. An extra pair indicated a blank line to end a command.
A few years later, on the DEC10 the 8 character name length was retained and % and # were allowed as initial characters. I am not sure about @ as an initial character.

Names like #LicBeds and %PopVote were legitimate names.
You could get a lot of info into 8 characters that way. # for count, % for percents, and maybe @ as "per" or rate like @Dozen @KiloWat.

A "c", "C", or "!" indicated a continuation line.

Level of measurement was not in the metadata until there were PCs. IIRC it was mid 90's that the data view was introduced. I am not sure if level of measurement was always in the data view.

Hope this helps.

Art Kendall
Social Research Consultants

On 4/13/2014 8:00 AM, John F Hall wrote:

I’m working, and writing a commentary for newbies, on the 1975 SSRC Survey Unit Multipurpose Survey (the first of its kind in the UK) recently uploaded at my request, but originally deposited at UKDS (Essex) by the late Prof. Cathie Marsh, my then colleague and trainee researcher. (See http://discover.ukdataservice.ac.uk/catalogue/?sn=680&type=Data%20catalogue#documentation).

Measurement levels originally specified, if any, have been lost during processing at UKDS (apparently normal as they produce Stata and *.tab versions as well, plus documentation). The ones in this file are those automatically assigned by SPSS as Nominal or Scale, depending on the range between the minimum and maximum encountered. Scrolling up and down the Level column there are no variables specified as Ordinal. They all need to be checked and corrected.

Can anyone remember if variable level specifications were available as long ago as that? I can only remember integer and alpha from INPUT FORMAT commands.

Variable names are all lower case: variable and value labels are a mixture of mostly UPPER and a few lower case. Can anyone remember when lower/mixed case was first allowed?

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

===================== 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
Social Research Consultants

Jon K Peck

Re: History of measurement levels in SPSS

In reply to this post by John F Hall

1975 is well before my time at SPSS (and before the X series of SPSS), but I am quite sure that the measurement level attribute was introduced much later than that. It might have come in with version 11.5 when CTABLES, which relies heavily on it, was introduced.

However, measurement level was never assigned based on the range of values in a variable. In older releases, for numeric variables, it was based on the number of distinct values in the data. Strings were always nominal. Newer releases use a number of criteria for guessing the level. See help for Edit > Options.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621

From: John F Hall <[hidden email]>
To: [hidden email],
Date: 04/13/2014 06:01 AM
Subject: [SPSSX-L] History of measurment levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Can anyone remember if variable level specifications were available as long ago as that? I can only remember integer and alpha from INPUT FORMAT commands.

Variable names are all lower case: variable and value labels are a mixture of mostly UPPER and a few lower case. Can anyone remember when lower/mixed case was first allowed?

John F Hall (Mr)

[Retired academic survey researcher]

Email: johnfhall@...

Website: www.surveyresearch.weebly.com

SPSS start page: www.surveyresearch.weebly.com/1-survey-analysis-workshop

John F Hall

Re: History of measurement levels in SPSS

You guys are wonderful. I learn something new every day. Hopefully I’ll master everything before “pop my clogs” time.

Somewhere in the UKDS site, there is an explanation of what they do with what they call “standard format data” whatever that is, but it certainly leaves a complete mess to clean up for us poor survey researchers.

From: Jon K Peck [mailto:[hidden email]]
Sent: 13 April 2014 14:48
To: John F Hall
Cc: [hidden email]
Subject: Re: [SPSSX-L] History of measurement levels in SPSS

Can anyone remember if variable level specifications were available as long ago as that? I can only remember integer and alpha from INPUT FORMAT commands.

Variable names are all lower case: variable and value labels are a mixture of mostly UPPER and a few lower case. Can anyone remember when lower/mixed case was first allowed?

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

John F Hall

Re: History of measurment levels in SPSS

In reply to this post by Art Kendall

Art

Thanks for this: I’m clearly not the only one who works on Sundays!

John

From: Art Kendall [mailto:[hidden email]]
Sent: 13 April 2014 14:33
To: John F Hall; [hidden email]
Cc: 'Bruce Weaver'; 'Aidan Kelly'
Subject: Re: History of measurment levels in SPSS

Art Kendall

Social Research Consultants

On 4/13/2014 8:00 AM, John F Hall wrote:

I’m working, and writing a commentary for newbies, on the 1975 SSRC Survey Unit Multipurpose Survey (the first of its kind in the UK) recently uploaded at my request, but originally deposited at UKDS (Essex) by the late Prof. Cathie Marsh, my then colleague and trainee researcher. (See http://discover.ukdataservice.ac.uk/catalogue/?sn=680&type=Data%20catalogue#documentation).

Measurement levels originally specified, if any, have been lost during processing at UKDS (apparently normal as they produce Stata and *.tab versions as well, plus documentation). The ones in this file are those automatically assigned by SPSS as Nominal or Scale, depending on the range between the minimum and maximum encountered. Scrolling up and down the Level column there are no variables specified as Ordinal. They all need to be checked and corrected.
Can anyone remember if variable level specifications were available as long ago as that? I can only remember integer and alpha from INPUT FORMAT commands.

Variable names are all lower case: variable and value labels are a mixture of mostly UPPER and a few lower case. Can anyone remember when lower/mixed case was first allowed?

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

Art Kendall

Re: History of measurment levels in SPSS

I check my email most days.
I guess one could call that work.

Art Kendall
Social Research Consultants

On 4/13/2014 2:47 PM, John F Hall [via SPSSX Discussion] wrote:

Art

Thanks for this: I’m clearly not the only one who works on Sundays!

John

From: Art Kendall [mailto:[hidden email]]
Sent: 13 April 2014 14:33
To: John F Hall; [hidden email]
Cc: 'Bruce Weaver'; 'Aidan Kelly'
Subject: Re: History of measurment levels in SPSS
The change may have occurred at different time on different platforms.
For example, the 1972 version of SPSS on DEC 10s, did not require 16 spaces to indicate a continuation line. An ASCII <tab> could be used. On these machines card images were treated exactly as if they were input from a teletype, i.e, a carriage return/linefeed pair indicated a new line. An extra pair indicated a blank line to end a command.
A few years later, on the DEC10 the 8 character name length was retained and % and # were allowed as initial characters. I am not sure about @ as an initial character.

Names like #LicBeds and %PopVote were legitimate names.
You could get a lot of info into 8 characters that way. # for count, % for percents, and maybe @ as "per" or rate like @Dozen @KiloWat.

A "c", "C", or "!" indicated a continuation line.

Level of measurement was not in the metadata until there were PCs. IIRC it was mid 90's that the data view was introduced. I am not sure if level of measurement was always in the data view.

Hope this helps.
Art Kendall
Social Research Consultants
On 4/13/2014 8:00 AM, John F Hall wrote:
I’m working, and writing a commentary for newbies, on the 1975 SSRC Survey Unit Multipurpose Survey (the first of its kind in the UK) recently uploaded at my request, but originally deposited at UKDS (Essex) by the late Prof. Cathie Marsh, my then colleague and trainee researcher. (See http://discover.ukdataservice.ac.uk/catalogue/?sn=680&type=Data%20catalogue#documentation).

Measurement levels originally specified, if any, have been lost during processing at UKDS (apparently normal as they produce Stata and *.tab versions as well, plus documentation). The ones in this file are those automatically assigned by SPSS as Nominal or Scale, depending on the range between the minimum and maximum encountered. Scrolling up and down the Level column there are no variables specified as Ordinal. They all need to be checked and corrected.

Can anyone remember if variable level specifications were available as long ago as that? I can only remember integer and alpha from INPUT FORMAT commands.

Variable names are all lower case: variable and value labels are a mixture of mostly UPPER and a few lower case. Can anyone remember when lower/mixed case was first allowed?

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
If you reply to this email, your message will be added to the discussion below:
http://spssx-discussion.1045642.n5.nabble.com/History-of-measurment-levels-in-SPSS-tp5725407p5725415.html

To start a new topic under SPSSX Discussion, email [hidden email]
To unsubscribe from SPSSX Discussion, click here.
NAML

Art Kendall
Social Research Consultants

Rick Oliver-3

Re: History of measurement levels in SPSS

In reply to this post by Jon K Peck

The ordinal measurement level is never assigned automatically. Variables without a specified or previously saved measurement level are assigned either the scale (continuous) or nominal measurement level.

Variable names are not case-sensitive, but the case in which they were originally defined is preserved. So if you define a variable name as Fred, it will be displayed as "Fred", but you can refer to that variable in command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and value labels have always been case-sensitive since they are quoted strings.

I think Jon is correct concerning when measurement level was introduced. Preserving case of defined variable names may have been at the same time or a few releases earlier.

Rick Oliver
Senior Information Developer
IBM Business Analytics (SPSS)
E-mail: [hidden email]

From: Jon K Peck/Chicago/IBM@IBMUS
To: [hidden email],
Date: 04/13/2014 07:53 AM
Subject: Re: History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Can anyone remember if variable level specifications were available as long ago as that? I can only remember integer and alpha from INPUT FORMAT commands.

Variable names are all lower case: variable and value labels are a mixture of mostly UPPER and a few lower case. Can anyone remember when lower/mixed case was first allowed?

John F Hall (Mr)

[Retired academic survey researcher]

Email: johnfhall@...

Website: www.surveyresearch.weebly.com

SPSS start page: www.surveyresearch.weebly.com/1-survey-analysis-workshop

Rich Ulrich

Re: History of measurement levels in SPSS

I remember one side-effect of the original implementation of preserving
the case of variable names: For at least one release I used, I could set
up separate variables named DRUG, drug, Drug, .... So long as the
capitalization was different, the variable was different.

I remember being astounded by that, and writing an example to prove it.
That would have been on a mainframe, probably about 1980, running on
a DEC-10.

--
Rich

Date: Mon, 14 Apr 2014 08:08:16 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

The ordinal measurement level is never assigned automatically. Variables without a specified or previously saved measurement level are assigned either the scale (continuous) or nominal measurement level.

Variable names are not case-sensitive, but the case in which they were originally defined is preserved. So if you define a variable name as Fred, it will be displayed as "Fred", but you can refer to that variable in command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and value labels have always been case-sensitive since they are quoted strings.

I think Jon is correct concerning when measurement level was introduced. Preserving case of defined variable names may have been at the same time or a few releases earlier.
[...]

Call

Send SMS

Add to Skype

You'll need Skype CreditFree via Skype

Jon K Peck

Re: History of measurement levels in SPSS

Rick Oliver-3

Re: History of measurement levels in SPSS

Case-preserving for display purposes came long after 1980. I am also skeptical that variable names were ever case-sensitive.

Rick Oliver
Senior Information Developer
IBM Business Analytics (SPSS)
E-mail: [hidden email]

From: Jon K Peck/Chicago/IBM@IBMUS
To: [hidden email],
Date: 04/14/2014 10:42 AM
Subject: Re: History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Maguin, Eugene

Re: History of measurement levels in SPSS

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rick Oliver
Sent: Monday, April 14, 2014 11:52 AM
To: [hidden email]
Subject: Re: History of measurement levels in SPSS

Jon K Peck

Re: History of measurement levels in SPSS

For over ten years now SPSS has provided change information for syntax in the CSR. This particular change, which had no functional effect, predates that.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621

From: "Maguin, Eugene" <[hidden email]>
To: [hidden email],
Date: 04/14/2014 10:05 AM
Subject: Re: [SPSSX-L] History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

I don’t mean this as a criticism of spss, rather an expression of surprise. Solely based on long ago working for a couple of different aircraft companies where there was a change history recorded on the face of every drawing and certification report, I would have expected something similar in commercial software production. Gene Maguin

From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Rick Oliver
Sent: Monday, April 14, 2014 11:52 AM
To: [hidden email]
Subject: Re: History of measurement levels in SPSS

Case-preserving for display purposes came long after 1980. I am also skeptical that variable names were ever case-sensitive.

Rick Oliver
Senior Information Developer
IBM Business Analytics (SPSS)
E-mail: oliverr@...

From: Jon K Peck/Chicago/IBM@IBMUS
To: [hidden email],
Date: 04/14/2014 10:42 AM
Subject: Re: History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Date: Mon, 14 Apr 2014 08:08:16 -0500
From: oliverr@...
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

The ordinal measurement level is never assigned automatically. Variables without a specified or previously saved measurement level are assigned either the scale (continuous) or nominal measurement level.

Variable names are not case-sensitive, but the case in which they were originally defined is preserved. So if you define a variable name as Fred, it will be displayed as "Fred", but you can refer to that variable in command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and value labels have always been case-sensitive since they are quoted strings.

I think Jon is correct concerning when measurement level was introduced. Preserving case of defined variable names may have been at the same time or a few releases earlier.
[...]

Call
Send SMS
Add to Skype
You'll need Skype CreditFree via Skype

Rick Oliver-3

Re: History of measurement levels in SPSS

In reply to this post by Maguin, Eugene

There is a change history from release 12 on. Prior to that, information on the exact release in which a feature was either introduced or modified is unfortunately not available.

http://pic.dhe.ibm.com/infocenter/spssstat/v22r0m0/index.jsp

Measurement level and case-preserving variable names came somewhere after release 5 and prior to 12, both probably closer to the latter.

Rick Oliver
Senior Information Developer
IBM Business Analytics (SPSS)
E-mail: [hidden email]

From: "Maguin, Eugene" <[hidden email]>
To: [hidden email],
Date: 04/14/2014 11:09 AM
Subject: Re: History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Art Kendall

Re: History of measurement levels in SPSS

In reply to this post by Rich Ulrich

At that time many IBM systems used TSO or Wlybur for editing, but they still submitted card images.

I also recall a DEC10 version being case sensitive. It was a 36 bit word system that relied heavily on ASCII. A single word held 5 7bit characters (A5) and a check bit.

Art Kendall
Social Research Consultants

On 4/14/2014 11:36 AM, Rich Ulrich [via SPSSX Discussion] wrote:

I remember one side-effect of the original implementation of preserving
the case of variable names: For at least one release I used, I could set
up separate variables named DRUG, drug, Drug, .... So long as the
capitalization was different, the variable was different.

I remember being astounded by that, and writing an example to prove it.
That would have been on a mainframe, probably about 1980, running on
a DEC-10.

--
Rich

Date: Mon, 14 Apr 2014 08:08:16 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

The ordinal measurement level is never assigned automatically. Variables without a specified or previously saved measurement level are assigned either the scale (continuous) or nominal measurement level.

Variable names are not case-sensitive, but the case in which they were originally defined is preserved. So if you define a variable name as Fred, it will be displayed as "Fred", but you can refer to that variable in command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and value labels have always been case-sensitive since they are quoted strings.

I think Jon is correct concerning when measurement level was introduced. Preserving case of defined variable names may have been at the same time or a few releases earlier.
[...]

<div onmouseout="SkypeClick2Call.MenuInjectionHandler.hideMenu(event)" onmouseover="SkypeClick2Call.MenuInjectionHandler.showMenu(this, event)" style="display: none;" id="skype_c2c_menu_container" class="skype_c2c_menu_container">
Call

Send SMS

Add to Skype

You'll need Skype CreditFree via Skype

If you reply to this email, your message will be added to the discussion below:
http://spssx-discussion.1045642.n5.nabble.com/History-of-measurment-levels-in-SPSS-tp5725407p5725427.html

To start a new topic under SPSSX Discussion, email [hidden email]
To unsubscribe from SPSSX Discussion, click here.
NAML

Art Kendall
Social Research Consultants

Rich Ulrich

Re: History of measurement levels in SPSS

In reply to this post by Rick Oliver-3

I remember being amazed and I remember writing a small test
that proved it; I don't think it was some other package.

Maybe that feature/bug never existed beyond the DEC-10 version;
or, even, maybe it never existed beyond a test of the DEC-10 version
at the University of Pittsburgh. - That is not beyond the range of what is
possible -

STORY.
When I started using SPSS at the Pitt Computer Center in 1974, SPSS was a
mainframe package; there were no PCs. SPSS was not native to DEC-10,
and therefore each version needed adaptations or conversions. Pitt had an
excellent mainframe operation at that time. Partly as evidence of that, they
held the franchise from SPSS for converting to DEC-10 and other 36-bit
environments. (I think "36-bit word" was the controlling factor.) As to
excellence: I visited my alma mater, the University of Texas, in 1977 (probably),
and the person I talked to at the computer center was highly pleased at the
SPSS and BMDP upgrade kits they received from Pitt -- the mag-tapes always
came with full instructions and always worked the first time. It was later that
SPSS took the conversions in-house.

A fellow named Mike Matzek (sp?) at Pitt was in charge of conversions for
both SPSS and BMD. I came to know him through my various SPSS questions.
The on-site Help Desk at Pitt required that basically everyone, including systems
programmers, take a turn. It was a very pleasant change to ask a question and
get back, instead of "What's SPSS do?", an comment like, "Oh, yeah, I noticed that
the manual is ambiguous there, when I was testing out the conversion."

This was all years before I ever heard the term for a package, "beta release".
I do find it conceivable that Mike could have let Pitt users test a new release.
I know we paid attention to new versions, because they implied new manuals;
but I don't remember today whether there were messages about minor fixes that
today would be called "patches".

--
Rich Ulrich

Date: Mon, 14 Apr 2014 10:52:03 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

Case-preserving for display purposes came long after 1980. I am also skeptical that variable names were ever case-sensitive.

Rick Oliver
Senior Information Developer
IBM Business Analytics (SPSS)
E-mail: [hidden email]

From: Jon K Peck/Chicago/IBM@IBMUS
To: [hidden email],
Date: 04/14/2014 10:42 AM
Subject: Re: History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Before my time, but I am skeptical. Before SPSS preserved the casing of variable names, DRUG and Drug etc would both have referred to the SAME variable. The change was only cosmetic and had no functional effect.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621<span class="skype_c2c_container" dir="ltr" tabindex="-1" onmouseover="SkypeClick2Call.MenuInjectionHandler.showMenu(this, event)" onmouseout="SkypeClick2Call.MenuInjectionHandler.hideMenu(event)" skype_menu_props="{"numberToCall":"+17203425621","isFreecall":false,"isMobile":false,"isRtl":false}"><img class="skype_c2c_logo_img" src="resource://skype_ff_extension-at-jetpack/skype_ff_extension/data/call_skype_logo.png">720-342-5621

From: Rich Ulrich <[hidden email]>
To: [hidden email],
Date: 04/14/2014 09:29 AM
Subject: Re: [SPSSX-L] History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Call

Send SMS

Add to Skype

You'll need Skype CreditFree via Skype

Mike

Re: History of measurement levels in SPSS

up to 1000 variables instead of the 500 vars in regular SPSS plus
other features) and a "miniversion" of SPSS (called SPSSG) for
small systems. Some of these may have had certain capabilities
that were absent from others.

(2) In the manual that I link to above, there is a review of some
of the procedures that SPSS-10 can do and there is one procedure
that I was unaware of; I quote the manual:

|TETRACHORIC VARIABLES = variable list (low value, high value)/

| CORRELATIONS = variable list WITH variable list /

| variable list WITH variable list / ...

|TETRACHORIC is an SPSSprocedure which computes tetrachoric

|correlation coefficients between dichotomous variables.

This is not in the 1975 manual but it might be in the 1979 update

which was for SPSS version 7 and 8; the reference for this book

is the following:

Hull, C. H., & Nie, N. H. (1979). Statistical Package for the Social Sciences: New Procedures and Facilities for Releases 7 and 8. Mac Graw Hill Book Company.

I admit that I don't think I have ever seen the above reference.

But it possible that versions 7 and 8 as well as SPSS-10 had

certain capabilities that disappeared in subsequent versions.

(3) There is a bit of a surprise in seeing SPSS version 7 and 8

because these do not correspond to the current numbering of

SPSS version. I believe that after version 8 SPSS engaged

in the (ill-conceived?) name change to SPSSX (which is why

this list is SPSSX-L) and then started renumbering the version

from either 1 or 2 (a copy of what appears to be the manual for

SPSSX version 2 is listed on books.google.com and can be

seen here:

http://books.google.com/books?ei=OGhMU6W1B9HMsQSryYDgCw&id=YWNVAAAAMAAJ&dq=spssx&focus=searchwithinvolume&q=tetrachoric

NOTE: the tetrachoric procedure appears to have been removed.

There would be another SPSSX release (ver 3) and then SPSS

went back to it's original name. The 1990 "White" manual contains

version 4 of SPSS and the numbering of releases/version continue

to the present day.

I guess one would have to look at the all SPSS manuals both from

the company those produced by university computer centers,

such as the Vax version of SPSS-X release 2.1; see:

http://books.google.com/books?id=PWBVAAAAMAAJ&dq=spssx&source=gbs_similarbooks )

to determine what may have once been possible to do.

So, it is possible that certain capabilities and procedures once

were available in SPSS but disappeared for whatever reason over

the years. It is surprising that SPSS once had procedure for

tetrachorics correlations but then removed it, causing programmers

to write their own macros/syntax to calculate it. One wonders

what the story is behind that.

-Mike Palij

New York University

[hidden email]

----- Original Message -----
From: Rich Ulrich
To: [hidden email]
Sent: Monday, April 14, 2014 3:44 PM
Subject: Re: History of measurement levels in SPSS

I remember being amazed and I remember writing a small test
that proved it; I don't think it was some other package.

Maybe that feature/bug never existed beyond the DEC-10 version;
or, even, maybe it never existed beyond a test of the DEC-10 version
at the University of Pittsburgh. - That is not beyond the range of what is
possible -

STORY.
When I started using SPSS at the Pitt Computer Center in 1974, SPSS was a
mainframe package; there were no PCs. SPSS was not native to DEC-10,
and therefore each version needed adaptations or conversions. Pitt had an
excellent mainframe operation at that time. Partly as evidence of that, they
held the franchise from SPSS for converting to DEC-10 and other 36-bit
environments. (I think "36-bit word" was the controlling factor.) As to
excellence: I visited my alma mater, the University of Texas, in 1977 (probably),
and the person I talked to at the computer center was highly pleased at the
SPSS and BMDP upgrade kits they received from Pitt -- the mag-tapes always
came with full instructions and always worked the first time. It was later that
SPSS took the conversions in-house.

A fellow named Mike Matzek (sp?) at Pitt was in charge of conversions for
both SPSS and BMD. I came to know him through my various SPSS questions.
The on-site Help Desk at Pitt required that basically everyone, including systems
programmers, take a turn. It was a very pleasant change to ask a question and
get back, instead of "What's SPSS do?", an comment like, "Oh, yeah, I noticed that
the manual is ambiguous there, when I was testing out the conversion."

This was all years before I ever heard the term for a package, "beta release".
I do find it conceivable that Mike could have let Pitt users test a new release.
I know we paid attention to new versions, because they implied new manuals;
but I don't remember today whether there were messages about minor fixes that
today would be called "patches".

--
Rich Ulrich

Date: Mon, 14 Apr 2014 10:52:03 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

Case-preserving for display purposes came long after 1980. I am also skeptical that variable names were ever case-sensitive.

Rick Oliver
Senior Information Developer
IBM Business Analytics (SPSS)
E-mail: [hidden email]

From:        Jon K Peck/Chicago/IBM@IBMUS
To:        [hidden email],
Date:        04/14/2014 10:42 AM
Subject:        Re: History of measurement levels in SPSS
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

Before my time, but I am skeptical. Before SPSS preserved the casing of variable names, DRUG and Drug etc would both have referred to the SAME variable. The change was only cosmetic and had no functional effect.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621720-342-5621

From:        Rich Ulrich <[hidden email]>
To:        [hidden email],
Date:        04/14/2014 09:29 AM
Subject:        Re: [SPSSX-L] History of measurement levels in SPSS
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

I remember one side-effect of the original implementation of preserving
the case of variable names: For at least one release I used, I could set
up separate variables named DRUG, drug, Drug, .... So long as the
capitalization was different, the variable was different.

I remember being astounded by that, and writing an example to prove it.
That would have been on a mainframe, probably about 1980, running on
a DEC-10.

--
Rich

Date: Mon, 14 Apr 2014 08:08:16 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

The ordinal measurement level is never assigned automatically. Variables without a specified or previously saved measurement level are assigned either the scale (continuous) or nominal measurement level.

Variable names are not case-sensitive, but the case in which they were originally defined is preserved. So if you define a variable name as Fred, it will be displayed as "Fred", but you can refer to that variable in command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and value labels have always been case-sensitive since they are quoted strings.

I think Jon is correct concerning when measurement level was introduced. Preserving case of defined variable names may have been at the same time or a few releases earlier.
[...]

Call
Send SMS
Add to Skype
You'll need Skype CreditFree via Skype

Call
Send SMS
Add to Skype
You'll need Skype CreditFree via Skype

Jon K Peck

Re: History of measurement levels in SPSS

Still before my time but SPSS went from SPSS 9 to SPSS-X. Clever, no? And restarted the version numbering. The X system was a massive eXtension of the old SPSS. In fact, the new central system was so big, that in 1983-84 when the PC system was first built (which is when I joined SPSS - the Inc), we had to build it off the old SPSS central system in order to meet the memory limitations of the target PC. The first PC version ran in 320K bytes of memory, including DOS, although it could use up to a grand total of 640K if you could afford that many chips. There was, of course, no virtual memory on the PC then. The memory limit required a large amount of memory management work on the SPSS code and module architecture. It wasn't until the first Windows version with its greater memory capacity that we were able to switch to the X code base.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621

From: Mike Palij <[hidden email]>
To: [hidden email],
Date: 04/14/2014 05:21 PM
Subject: Re: [SPSSX-L] History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

I have not run SPSS on a DEC-10 but started to use it on a
Sperry UNIVAC 1100 in the late 1970s and later on various
other machines (I preferred to use BMDP because I dealt
primarily with experimental designs and SPSS' had very
limited ANOVA capabilities in comparison; Ivor Francis
reviewed ANOVA software in the Dec 1973 JASA and
notes that SPSS did not have an ANOVA procedure!).
However, my memory was that on the different systems,
SPSS was indifferent to the use of capitals/lower case
for variable names, treating DRUG/drug/Drug the same.

However, this does not mean that the PDP-10 version might
not have had this capability. I did a web search for this
version and came across a PDF of a manual developed by
the computer center at the Australian University of Queensland.
This manual describes the statistical packages available at the
university, including the PDP-10 version of SPSS. You
should be able to access the PDF from the following link:
http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CCgQFjAA&url=http%3A%2F%2Fbitsavers.informatik.uni-stuttgart.de%2Fwww.computer.museum.uq.edu.au%2Fpdf%2FMNT-3%2520Statistical%2520Packages%2520Revised%2520Edition%2520August%25201981.pdf&ei=qVNMU5zXMvffsASrnIK4Dw&usg=AFQjCNH9mk3pcSQUuwcp7zEOYqIKcOMNEg&sig2=_aMz04KIeYO51wieVzzFJw
I quote a short section of what was called "SPSS-10":

|5.2 The PDP-10 Implementation of SPSS
|
|The PDP-10 implementation of SPSS (SPSS-10) was produced
|at the University of Pittsburgh and is a conversion of the National
|Opinion Research Centre's SPSSH Version 7.01. SPSS-10
|closely follows the documentation in:
|
|Nie, Hull, Jenkins, Steinbrunner, Bent
|Statistical Package for the Social Sciences
|(Second Edition)
|McGraw-Hill Book Company
|(1975)
|
|and
|
|Hull and Nie
|SPSS Update
|New Procedures and Facilities for Releases 7 and 8
|McGraw-Hill Book Company
|(1979)
|
|Any differences between the PDP-10 implementation and that
|described in the McGraw-Hill manual are documented below.
|A more comprehensive list of differences may be obtained from
|DOC: by .
|.PRINT DOC:SPSS.DOC
|and
|.PRINT DOC:SPSS8.DOC

There are a couple of aspects about SPSS that many may not
know or simply forgotten, such as:

(1) During the 1970s there were a couple of different versions of
SPSS. In the appendices of the maroon 1975 SPSS manual,
there is the "regular" SPSS (SPSSH) on different systems (but
no DEC PDP systems) as well as a "maxiversion" (which allowed
up to 1000 variables instead of the 500 vars in regular SPSS plus
other features) and a "miniversion" of SPSS (called SPSSG) for
small systems. Some of these may have had certain capabilities
that were absent from others.

(2) In the manual that I link to above, there is a review of some
of the procedures that SPSS-10 can do and there is one procedure
that I was unaware of; I quote the manual:

|TETRACHORIC VARIABLES = variable list (low value, high value)/
| CORRELATIONS = variable list WITH variable list /
| variable list WITH variable list / ...
|
|TETRACHORIC is an SPSSprocedure which computes tetrachoric
|correlation coefficients between dichotomous variables.

This is not in the 1975 manual but it might be in the 1979 update
which was for SPSS version 7 and 8; the reference for this book
is the following:

Hull, C. H., & Nie, N. H. (1979). Statistical Package for the Social Sciences: New Procedures and Facilities for Releases 7 and 8. Mac Graw Hill Book Company.

I admit that I don't think I have ever seen the above reference.
But it possible that versions 7 and 8 as well as SPSS-10 had
certain capabilities that disappeared in subsequent versions.

(3) There is a bit of a surprise in seeing SPSS version 7 and 8
because these do not correspond to the current numbering of
SPSS version. I believe that after version 8 SPSS engaged
in the (ill-conceived?) name change to SPSSX (which is why
this list is SPSSX-L) and then started renumbering the version
from either 1 or 2 (a copy of what appears to be the manual for
SPSSX version 2 is listed on books.google.com and can be
seen here:
http://books.google.com/books?ei=OGhMU6W1B9HMsQSryYDgCw&id=YWNVAAAAMAAJ&dq=spssx&focus=searchwithinvolume&q=tetrachoric
NOTE: the tetrachoric procedure appears to have been removed.

There would be another SPSSX release (ver 3) and then SPSS
went back to it's original name. The 1990 "White" manual contains
version 4 of SPSS and the numbering of releases/version continue
to the present day.

I guess one would have to look at the all SPSS manuals both from
the company those produced by university computer centers,
such as the Vax version of SPSS-X release 2.1; see:
http://books.google.com/books?id=PWBVAAAAMAAJ&dq=spssx&source=gbs_similarbooks )
to determine what may have once been possible to do.

So, it is possible that certain capabilities and procedures once
were available in SPSS but disappeared for whatever reason over
the years. It is surprising that SPSS once had procedure for
tetrachorics correlations but then removed it, causing programmers
to write their own macros/syntax to calculate it. One wonders
what the story is behind that.

-Mike Palij
New York University
mp26@...

----- Original Message -----
From: Rich Ulrich
To: [hidden email]
Sent: Monday, April 14, 2014 3:44 PM
Subject: Re: History of measurement levels in SPSS

I remember being amazed and I remember writing a small test
that proved it; I don't think it was some other package.

Maybe that feature/bug never existed beyond the DEC-10 version;
or, even, maybe it never existed beyond a test of the DEC-10 version
at the University of Pittsburgh. - That is not beyond the range of what is
possible -

STORY.
When I started using SPSS at the Pitt Computer Center in 1974, SPSS was a
mainframe package; there were no PCs. SPSS was not native to DEC-10,
and therefore each version needed adaptations or conversions. Pitt had an
excellent mainframe operation at that time. Partly as evidence of that, they
held the franchise from SPSS for converting to DEC-10 and other 36-bit
environments. (I think "36-bit word" was the controlling factor.) As to
excellence: I visited my alma mater, the University of Texas, in 1977 (probably),
and the person I talked to at the computer center was highly pleased at the
SPSS and BMDP upgrade kits they received from Pitt -- the mag-tapes always
came with full instructions and always worked the first time. It was later that
SPSS took the conversions in-house.

A fellow named Mike Matzek (sp?) at Pitt was in charge of conversions for
both SPSS and BMD. I came to know him through my various SPSS questions.
The on-site Help Desk at Pitt required that basically everyone, including systems
programmers, take a turn. It was a very pleasant change to ask a question and
get back, instead of "What's SPSS do?", an comment like, "Oh, yeah, I noticed that
the manual is ambiguous there, when I was testing out the conversion."

This was all years before I ever heard the term for a package, "beta release".
I do find it conceivable that Mike could have let Pitt users test a new release.
I know we paid attention to new versions, because they implied new manuals;
but I don't remember today whether there were messages about minor fixes that
today would be called "patches".

--
Rich Ulrich

Date: Mon, 14 Apr 2014 10:52:03 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

Case-preserving for display purposes came long after 1980. I am also skeptical that variable names were ever case-sensitive.

Rick Oliver
Senior Information Developer
IBM Business Analytics (SPSS)
E-mail: [hidden email]

From: Jon K Peck/Chicago/IBM@IBMUS
To: [hidden email],
Date: 04/14/2014 10:42 AM
Subject: Re: History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Before my time, but I am skeptical. Before SPSS preserved the casing of variable names, DRUG and Drug etc would both have referred to the SAME variable. The change was only cosmetic and had no functional effect.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621720-342-5621

From: Rich Ulrich <[hidden email]>
To: [hidden email],
Date: 04/14/2014 09:29 AM
Subject: Re: [SPSSX-L] History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

I remember one side-effect of the original implementation of preserving
the case of variable names: For at least one release I used, I could set
up separate variables named DRUG, drug, Drug, .... So long as the
capitalization was different, the variable was different.

I remember being astounded by that, and writing an example to prove it.
That would have been on a mainframe, probably about 1980, running on
a DEC-10.

--
Rich

Date: Mon, 14 Apr 2014 08:08:16 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

The ordinal measurement level is never assigned automatically. Variables without a specified or previously saved measurement level are assigned either the scale (continuous) or nominal measurement level.

Variable names are not case-sensitive, but the case in which they were originally defined is preserved. So if you define a variable name as Fred, it will be displayed as "Fred", but you can refer to that variable in command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and value labels have always been case-sensitive since they are quoted strings.

I think Jon is correct concerning when measurement level was introduced. Preserving case of defined variable names may have been at the same time or a few releases earlier.
[...]

Call
Send SMS
Add to Skype
You'll need Skype CreditFree via Skype

Call
Send SMS
Add to Skype
You'll need Skype CreditFree via Skype

David Marso

Re: History of measurement levels in SPSS

Administrator

Jon K Peck wrote

Still before my time but SPSS went from SPSS 9 to SPSS-X. Clever, no? And
restarted the version numbering. The X system was a massive eXtension of
the old SPSS. In fact, the new central system was so big, that in 1983-84
when the PC system was first built (which is when I joined SPSS - the
Inc), we had to build it off the old SPSS central system in order to meet
the memory limitations of the target PC. The first PC version ran in 320K
bytes of memory, including DOS, although it could use up to a grand total
of 640K if you could afford that many chips. There was, of course, no
virtual memory on the PC then. The memory limit required a large amount
of memory management work on the SPSS code and module architecture. It
wasn't until the first Windows version with its greater memory capacity
that we were able to switch to the X code base.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621

From: Mike Palij <[hidden email]>
To: [hidden email],
Date: 04/14/2014 05:21 PM
Subject: Re: [SPSSX-L] History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

I have not run SPSS on a DEC-10 but started to use it on a
Sperry UNIVAC 1100 in the late 1970s and later on various
other machines (I preferred to use BMDP because I dealt
primarily with experimental designs and SPSS' had very
limited ANOVA capabilities in comparison; Ivor Francis
reviewed ANOVA software in the Dec 1973 JASA and
notes that SPSS did not have an ANOVA procedure!).
However, my memory was that on the different systems,
SPSS was indifferent to the use of capitals/lower case
for variable names, treating DRUG/drug/Drug the same.

However, this does not mean that the PDP-10 version might
not have had this capability. I did a web search for this
version and came across a PDF of a manual developed by
the computer center at the Australian University of Queensland.
This manual describes the statistical packages available at the
university, including the PDP-10 version of SPSS. You
should be able to access the PDF from the following link:
http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CCgQFjAA&url=http%3A%2F%2Fbitsavers.informatik.uni-stuttgart.de%2Fwww.computer.museum.uq.edu.au%2Fpdf%2FMNT-3%2520Statistical%2520Packages%2520Revised%2520Edition%2520August%25201981.pdf&ei=qVNMU5zXMvffsASrnIK4Dw&usg=AFQjCNH9mk3pcSQUuwcp7zEOYqIKcOMNEg&sig2=_aMz04KIeYO51wieVzzFJw

I quote a short section of what was called "SPSS-10":

|5.2 The PDP-10 Implementation of SPSS
|
|The PDP-10 implementation of SPSS (SPSS-10) was produced
|at the University of Pittsburgh and is a conversion of the National
|Opinion Research Centre's SPSSH Version 7.01. SPSS-10
|closely follows the documentation in:
|
|Nie, Hull, Jenkins, Steinbrunner, Bent
|Statistical Package for the Social Sciences
|(Second Edition)
|McGraw-Hill Book Company
|(1975)
|
|and
|
|Hull and Nie
|SPSS Update
|New Procedures and Facilities for Releases 7 and 8
|McGraw-Hill Book Company
|(1979)
|
|Any differences between the PDP-10 implementation and that
|described in the McGraw-Hill manual are documented below.
|A more comprehensive list of differences may be obtained from
|DOC: by .
|.PRINT DOC:SPSS.DOC
|and
|.PRINT DOC:SPSS8.DOC

There are a couple of aspects about SPSS that many may not
know or simply forgotten, such as:

(1) During the 1970s there were a couple of different versions of
SPSS. In the appendices of the maroon 1975 SPSS manual,
there is the "regular" SPSS (SPSSH) on different systems (but
no DEC PDP systems) as well as a "maxiversion" (which allowed
up to 1000 variables instead of the 500 vars in regular SPSS plus
other features) and a "miniversion" of SPSS (called SPSSG) for
small systems. Some of these may have had certain capabilities
that were absent from others.

(2) In the manual that I link to above, there is a review of some
of the procedures that SPSS-10 can do and there is one procedure
that I was unaware of; I quote the manual:

|TETRACHORIC VARIABLES = variable list (low value, high value)/
| CORRELATIONS = variable list WITH variable list /
| variable list WITH variable list / ...
|
|TETRACHORIC is an SPSSprocedure which computes tetrachoric
|correlation coefficients between dichotomous variables.

This is not in the 1975 manual but it might be in the 1979 update
which was for SPSS version 7 and 8; the reference for this book
is the following:

Hull, C. H., & Nie, N. H. (1979). Statistical Package for the Social
Sciences: New Procedures and Facilities for Releases 7 and 8. Mac Graw
Hill Book Company.

I admit that I don't think I have ever seen the above reference.
But it possible that versions 7 and 8 as well as SPSS-10 had
certain capabilities that disappeared in subsequent versions.

(3) There is a bit of a surprise in seeing SPSS version 7 and 8
because these do not correspond to the current numbering of
SPSS version. I believe that after version 8 SPSS engaged
in the (ill-conceived?) name change to SPSSX (which is why
this list is SPSSX-L) and then started renumbering the version
from either 1 or 2 (a copy of what appears to be the manual for
SPSSX version 2 is listed on books.google.com and can be
seen here:
http://books.google.com/books?ei=OGhMU6W1B9HMsQSryYDgCw&id=YWNVAAAAMAAJ&dq=spssx&focus=searchwithinvolume&q=tetrachoric
NOTE: the tetrachoric procedure appears to have been removed.

There would be another SPSSX release (ver 3) and then SPSS
went back to it's original name. The 1990 "White" manual contains
version 4 of SPSS and the numbering of releases/version continue
to the present day.

I guess one would have to look at the all SPSS manuals both from
the company those produced by university computer centers,
such as the Vax version of SPSS-X release 2.1; see:
http://books.google.com/books?id=PWBVAAAAMAAJ&dq=spssx&source=gbs_similarbooks
)
to determine what may have once been possible to do.

So, it is possible that certain capabilities and procedures once
were available in SPSS but disappeared for whatever reason over
the years. It is surprising that SPSS once had procedure for
tetrachorics correlations but then removed it, causing programmers
to write their own macros/syntax to calculate it. One wonders
what the story is behind that.

-Mike Palij
New York University
[hidden email]

----- Original Message -----
From: Rich Ulrich
To: [hidden email]
Sent: Monday, April 14, 2014 3:44 PM
Subject: Re: History of measurement levels in SPSS

I remember being amazed and I remember writing a small test
that proved it; I don't think it was some other package.

Maybe that feature/bug never existed beyond the DEC-10 version;
or, even, maybe it never existed beyond a test of the DEC-10 version
at the University of Pittsburgh. - That is not beyond the range of what
is
possible -

STORY.
When I started using SPSS at the Pitt Computer Center in 1974, SPSS was a
mainframe package; there were no PCs. SPSS was not native to DEC-10,
and therefore each version needed adaptations or conversions. Pitt had an
excellent mainframe operation at that time. Partly as evidence of that,
they
held the franchise from SPSS for converting to DEC-10 and other 36-bit
environments. (I think "36-bit word" was the controlling factor.) As to
excellence: I visited my alma mater, the University of Texas, in 1977
(probably),
and the person I talked to at the computer center was highly pleased at
the
SPSS and BMDP upgrade kits they received from Pitt -- the mag-tapes always

came with full instructions and always worked the first time. It was
later that
SPSS took the conversions in-house.

A fellow named Mike Matzek (sp?) at Pitt was in charge of conversions for
both SPSS and BMD. I came to know him through my various SPSS questions.
The on-site Help Desk at Pitt required that basically everyone, including
systems
programmers, take a turn. It was a very pleasant change to ask a question
and
get back, instead of "What's SPSS do?", an comment like, "Oh, yeah, I
noticed that
the manual is ambiguous there, when I was testing out the conversion."

This was all years before I ever heard the term for a package, "beta
release".
I do find it conceivable that Mike could have let Pitt users test a new
release.
I know we paid attention to new versions, because they implied new
manuals;
but I don't remember today whether there were messages about minor fixes
that
today would be called "patches".

--
Rich Ulrich

Date: Mon, 14 Apr 2014 10:52:03 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

Case-preserving for display purposes came long after 1980. I am also
skeptical that variable names were ever case-sensitive.

Rick Oliver
Senior Information Developer
IBM Business Analytics (SPSS)
E-mail: [hidden email]

From: Jon K Peck/Chicago/IBM@IBMUS
To: [hidden email],
Date: 04/14/2014 10:42 AM
Subject: Re: History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Before my time, but I am skeptical. Before SPSS preserved the casing of
variable names, DRUG and Drug etc would both have referred to the SAME
variable. The change was only cosmetic and had no functional effect.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621720-342-5621

From: Rich Ulrich <[hidden email]>
To: [hidden email],
Date: 04/14/2014 09:29 AM
Subject: Re: [SPSSX-L] History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

I remember one side-effect of the original implementation of preserving
the case of variable names: For at least one release I used, I could set
up separate variables named DRUG, drug, Drug, .... So long as the
capitalization was different, the variable was different.

I remember being astounded by that, and writing an example to prove it.
That would have been on a mainframe, probably about 1980, running on
a DEC-10.

--
Rich

Date: Mon, 14 Apr 2014 08:08:16 -0500
From: [hidden email]
Subject: Re: History of measurement levels in SPSS
To: [hidden email]

The ordinal measurement level is never assigned automatically. Variables
without a specified or previously saved measurement level are assigned
either the scale (continuous) or nominal measurement level.

Variable names are not case-sensitive, but the case in which they were
originally defined is preserved. So if you define a variable name as Fred,
it will be displayed as "Fred", but you can refer to that variable in
command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and
value labels have always been case-sensitive since they are quoted
strings.

I think Jon is correct concerning when measurement level was introduced.
Preserving case of defined variable names may have been at the same time
or a few releases earlier.
[...]

Call
Send SMS
Add to Skype
You'll need Skype CreditFree via Skype

Call
Send SMS
Add to Skype
You'll need Skype CreditFree via Skype

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?"

Jon K Peck

Re: History of measurement levels in SPSS

I was actually a principal developer on the OS/2 product, which used the Review editor that I designed as its user interface. The Mac version followed suit. Both of those had a character-mode interface, but they set the pattern for the two-process, frontend-backend architecture that evolved into what we use today. The OS/2 product, though, went nowhere, because nobody - at least in the end user community - chose to run that OS. Well, I guess more copies of the SPSS OS/2 version were sold than of the DEC Pro version :-)

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621

From: David Marso <[hidden email]>
To: [hidden email],
Date: 04/15/2014 06:03 AM
Subject: Re: [SPSSX-L] History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

" It wasn't until the first Windows version with its greater memory capacity that we were able to switch to the X code base. " Not quite right Jon, SPSS for OS/2 and the Mac version predated the Windoze version by at least a year (maybe 2) and were based on SPSS 4.0. I know this because I was the main OS/2 support person in SPSS technical support for several years beginning in Oct, 1990 -also backed up Cathy F on Mac support-. The OS/2 system was ROCK SOLID. Only calls I got were people trying to print and a funky font issue. Recall the 'good ole' days when a monster PC had 32M of RAM! My OS/2 box had 16M and I waited for 2 weeks to have a math coprocessor installed after I started in teksport. Remember they cut disks in the basement back in those days. ---- Jon K Peck wrote > Still before my time but SPSS went from SPSS 9 to SPSS-X. Clever, no? And > restarted the version numbering. The X system was a massive eXtension of > the old SPSS. In fact, the new central system was so big, that in 1983-84 > when the PC system was first built (which is when I joined SPSS - the > Inc), we had to build it off the old SPSS central system in order to meet > the memory limitations of the target PC. The first PC version ran in 320K > bytes of memory, including DOS, although it could use up to a grand total > of 640K if you could afford that many chips. There was, of course, no > virtual memory on the PC then. The memory limit required a large amount > of memory management work on the SPSS code and module architecture. It > wasn't until the first Windows version with its greater memory capacity > that we were able to switch to the X code base. > > > Jon Peck (no "h") aka Kim > Senior Software Engineer, IBM > peck@.ibm > phone: 720-342-5621 > > > > > From: Mike Palij < > mp26@ > > > To: > SPSSX-L@.uga > , > Date: 04/14/2014 05:21 PM > Subject: Re: [SPSSX-L] History of measurement levels in SPSS > Sent by: "SPSSX(r) Discussion" < > SPSSX-L@.uga > > > > > > I have not run SPSS on a DEC-10 but started to use it on a > Sperry UNIVAC 1100 in the late 1970s and later on various > other machines (I preferred to use BMDP because I dealt > primarily with experimental designs and SPSS' had very > limited ANOVA capabilities in comparison; Ivor Francis > reviewed ANOVA software in the Dec 1973 JASA and > notes that SPSS did not have an ANOVA procedure!). > However, my memory was that on the different systems, > SPSS was indifferent to the use of capitals/lower case > for variable names, treating DRUG/drug/Drug the same. > > However, this does not mean that the PDP-10 version might > not have had this capability. I did a web search for this > version and came across a PDF of a manual developed by > the computer center at the Australian University of Queensland. > This manual describes the statistical packages available at the > university, including the PDP-10 version of SPSS. You > should be able to access the PDF from the following link: >http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CCgQFjAA&url=http%3A%2F%2Fbitsavers.informatik.uni-stuttgart.de%2Fwww.computer.museum.uq.edu.au%2Fpdf%2FMNT-3%2520Statistical%2520Packages%2520Revised%2520Edition%2520August%25201981.pdf&ei=qVNMU5zXMvffsASrnIK4Dw&usg=AFQjCNH9mk3pcSQUuwcp7zEOYqIKcOMNEg&sig2=_aMz04KIeYO51wieVzzFJw> > I quote a short section of what was called "SPSS-10": > > |5.2 The PDP-10 Implementation of SPSS > | > |The PDP-10 implementation of SPSS (SPSS-10) was produced > |at the University of Pittsburgh and is a conversion of the National > |Opinion Research Centre's SPSSH Version 7.01. SPSS-10 > |closely follows the documentation in: > | > |Nie, Hull, Jenkins, Steinbrunner, Bent > |Statistical Package for the Social Sciences > |(Second Edition) > |McGraw-Hill Book Company > |(1975) > | > |and > | > |Hull and Nie > |SPSS Update > |New Procedures and Facilities for Releases 7 and 8 > |McGraw-Hill Book Company > |(1979) > | > |Any differences between the PDP-10 implementation and that > |described in the McGraw-Hill manual are documented below. > |A more comprehensive list of differences may be obtained from > |DOC: by . > |.PRINT DOC:SPSS.DOC > |and > |.PRINT DOC:SPSS8.DOC > > There are a couple of aspects about SPSS that many may not > know or simply forgotten, such as: > > (1) During the 1970s there were a couple of different versions of > SPSS. In the appendices of the maroon 1975 SPSS manual, > there is the "regular" SPSS (SPSSH) on different systems (but > no DEC PDP systems) as well as a "maxiversion" (which allowed > up to 1000 variables instead of the 500 vars in regular SPSS plus > other features) and a "miniversion" of SPSS (called SPSSG) for > small systems. Some of these may have had certain capabilities > that were absent from others. > > (2) In the manual that I link to above, there is a review of some > of the procedures that SPSS-10 can do and there is one procedure > that I was unaware of; I quote the manual: > > |TETRACHORIC VARIABLES = variable list (low value, high value)/ > | CORRELATIONS = variable list WITH variable list / > | variable list WITH variable list / ... > | > |TETRACHORIC is an SPSSprocedure which computes tetrachoric > |correlation coefficients between dichotomous variables. > > This is not in the 1975 manual but it might be in the 1979 update > which was for SPSS version 7 and 8; the reference for this book > is the following: > > Hull, C. H., & Nie, N. H. (1979). Statistical Package for the Social > Sciences: New Procedures and Facilities for Releases 7 and 8. Mac Graw > Hill Book Company. > > I admit that I don't think I have ever seen the above reference. > But it possible that versions 7 and 8 as well as SPSS-10 had > certain capabilities that disappeared in subsequent versions. > > (3) There is a bit of a surprise in seeing SPSS version 7 and 8 > because these do not correspond to the current numbering of > SPSS version. I believe that after version 8 SPSS engaged > in the (ill-conceived?) name change to SPSSX (which is why > this list is SPSSX-L) and then started renumbering the version > from either 1 or 2 (a copy of what appears to be the manual for > SPSSX version 2 is listed on books.google.com and can be > seen here: >http://books.google.com/books?ei=OGhMU6W1B9HMsQSryYDgCw&id=YWNVAAAAMAAJ&dq=spssx&focus=searchwithinvolume&q=tetrachoric> NOTE: the tetrachoric procedure appears to have been removed. > > There would be another SPSSX release (ver 3) and then SPSS > went back to it's original name. The 1990 "White" manual contains > version 4 of SPSS and the numbering of releases/version continue > to the present day. > > I guess one would have to look at the all SPSS manuals both from > the company those produced by university computer centers, > such as the Vax version of SPSS-X release 2.1; see: >http://books.google.com/books?id=PWBVAAAAMAAJ&dq=spssx&source=gbs_similarbooks> ) > to determine what may have once been possible to do. > > So, it is possible that certain capabilities and procedures once > were available in SPSS but disappeared for whatever reason over > the years. It is surprising that SPSS once had procedure for > tetrachorics correlations but then removed it, causing programmers > to write their own macros/syntax to calculate it. One wonders > what the story is behind that. > > -Mike Palij > New York University > mp26@ > > > ----- Original Message ----- > From: Rich Ulrich > To: > SPSSX-L@.UGA > Sent: Monday, April 14, 2014 3:44 PM > Subject: Re: History of measurement levels in SPSS > > > I remember being amazed and I remember writing a small test > that proved it; I don't think it was some other package. > > Maybe that feature/bug never existed beyond the DEC-10 version; > or, even, maybe it never existed beyond a test of the DEC-10 version > at the University of Pittsburgh. - That is not beyond the range of what > is > possible - > > STORY. > When I started using SPSS at the Pitt Computer Center in 1974, SPSS was a > mainframe package; there were no PCs. SPSS was not native to DEC-10, > and therefore each version needed adaptations or conversions. Pitt had an > excellent mainframe operation at that time. Partly as evidence of that, > they > held the franchise from SPSS for converting to DEC-10 and other 36-bit > environments. (I think "36-bit word" was the controlling factor.) As to > excellence: I visited my alma mater, the University of Texas, in 1977 > (probably), > and the person I talked to at the computer center was highly pleased at > the > SPSS and BMDP upgrade kits they received from Pitt -- the mag-tapes always > > came with full instructions and always worked the first time. It was > later that > SPSS took the conversions in-house. > > A fellow named Mike Matzek (sp?) at Pitt was in charge of conversions for > both SPSS and BMD. I came to know him through my various SPSS questions. > The on-site Help Desk at Pitt required that basically everyone, including > systems > programmers, take a turn. It was a very pleasant change to ask a question > and > get back, instead of "What's SPSS do?", an comment like, "Oh, yeah, I > noticed that > the manual is ambiguous there, when I was testing out the conversion." > > This was all years before I ever heard the term for a package, "beta > release". > I do find it conceivable that Mike could have let Pitt users test a new > release. > I know we paid attention to new versions, because they implied new > manuals; > but I don't remember today whether there were messages about minor fixes > that > today would be called "patches". > > -- > Rich Ulrich > > > > > Date: Mon, 14 Apr 2014 10:52:03 -0500 > From: > oliverr@.ibm > Subject: Re: History of measurement levels in SPSS > To: > SPSSX-L@.UGA > > Case-preserving for display purposes came long after 1980. I am also > skeptical that variable names were ever case-sensitive. > > Rick Oliver > Senior Information Developer > IBM Business Analytics (SPSS) > E-mail: > oliverr@.ibm > > > > From: Jon K Peck/Chicago/IBM@IBMUS > To: > SPSSX-L@.uga > , > Date: 04/14/2014 10:42 AM > Subject: Re: History of measurement levels in SPSS > Sent by: "SPSSX(r) Discussion" < > SPSSX-L@.uga > > > > > > > > Before my time, but I am skeptical. Before SPSS preserved the casing of > variable names, DRUG and Drug etc would both have referred to the SAME > variable. The change was only cosmetic and had no functional effect. > > > Jon Peck (no "h") aka Kim > Senior Software Engineer, IBM > peck@.ibm > phone: 720-342-5621720-342-5621 > > > > > From: Rich Ulrich < > rich-ulrich@ > > > To: > SPSSX-L@.uga > , > Date: 04/14/2014 09:29 AM > Subject: Re: [SPSSX-L] History of measurement levels in SPSS > Sent by: "SPSSX(r) Discussion" < > SPSSX-L@.uga > > > > > > > > I remember one side-effect of the original implementation of preserving > the case of variable names: For at least one release I used, I could set > up separate variables named DRUG, drug, Drug, .... So long as the > capitalization was different, the variable was different. > > I remember being astounded by that, and writing an example to prove it. > That would have been on a mainframe, probably about 1980, running on > a DEC-10. > > -- > Rich > > > > Date: Mon, 14 Apr 2014 08:08:16 -0500 > From: > oliverr@.ibm > Subject: Re: History of measurement levels in SPSS > To: > SPSSX-L@.UGA > > The ordinal measurement level is never assigned automatically. Variables > without a specified or previously saved measurement level are assigned > either the scale (continuous) or nominal measurement level. > > Variable names are not case-sensitive, but the case in which they were > originally defined is preserved. So if you define a variable name as Fred, > it will be displayed as "Fred", but you can refer to that variable in > command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and > value labels have always been case-sensitive since they are quoted > strings. > > I think Jon is correct concerning when measurement level was introduced. > Preserving case of defined variable names may have been at the same time > or a few releases earlier. > [...] > > Call > Send SMS > Add to Skype > You'll need Skype CreditFree via Skype > > Call > Send SMS > Add to Skype > You'll need Skype CreditFree via Skype ----- 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/History-of-measurment-levels-in-SPSS-tp5725407p5725445.htmlSent 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

Mike

Re: History of measurement levels in SPSS

During the 1980s I was in research situations that were best served

by BMDP (the mainframe version of which appeared to be ported

directly to the MS-Dos version, unlike the situation with SPSS).

But around 1993, I joined an outfit that (1) did almost all analyses

in SPSS, and (b) did not have access to a mainframe but found that

the OS/2 version of SPSS provided a reasonable alternative (the

SPSS-PC version was too limited). We continued to use SPSS

for OS/2 in the early to mid 1990s until the company decided to

go with Windows PCs (I preferred the OS/2 interface but we

no longer had any OS/2 machines; besides, I could run BMDP

on Windows to compliment/verify the analyses in SPSS-Win).

There is a review of the statistical packages for the PC in the

May 11, 1993 "PC Magazine" and the three versions of SPSS

(i.e., SPSS-PC, SPSS for OS/2, and SPSS-Win) are reviewed.

The article can be read on books.google.com at the following link:

http://books.google.com/books?id=23XxbVB2Y3gC&pg=PA264&lpg=PA264&dq=%22spss+for+os/2%22&source=bl&ots=eo-LZ5NRC2&sig=UtIxO5v44uVh0VH06bRX7tPsE5w&hl=en&sa=X&ei=dCRNU5i6O4ulsASXo4HgBg&ved=0CEkQ6AEwAw#v=onepage&q=%22spss%20for%20os%2F2%22&f=false

NOTE: SPSS-Win got a PC Mag Editor's Choice designation.

A little history reported in the article:

(1) SPSS for DOS was first released in 1984 but this was a

stripped down version of the mainframe version. Given

the expense of MS-DOS machines plus the limitations of

SPSS for DOS, it made more sense to continue to use a

mainframe version of SPSS.

(2) SPSS for OS/2 was released in 1989 (OS/2 ver 1) but

the mainframe version 4.1 was made compatible with OS/2

ver 2 around 1992-1993. It was far superior to the DOS

version but did not take advantage of all of the features that

OS/2 provided.

(3) It is interesting to note that in this 1993 article another

stat program also received an Editor Choice designation:

Systat for DOS (the Systat for Windows did make the cut).

Interesting given the future relationship that Systat would

have with SPSS in the future.

By the way, back in 1993 (could have been 1994) there

was a regression I was trying to run in SPSS for OS/2 but

SPSS crashed when I tried to run it. The analysis ran

on a VAX that I had access to, so the question was what

was the problem on OS/2? The best I could do was to localized

the problem to a constellation of variables in the equation and

something else (I can't remember what). We did notify SPSS

but either they weren't able to replicate the problem or

they did but could figure out a fix (I don't remember the details).

We just avoiding the constellation I identified in SPSS of OS/2 or

I did the analysis on a Vax.

-Mike Palij

New York University

[hidden email]

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: Tuesday, April 15, 2014 8:24 AM

Subject: Re: History of measurement levels in SPSS

I was actually a principal developer on the OS/2 product, which used the Review editor that I designed as its user interface. The Mac version followed suit. Both of those had a character-mode interface, but they set the pattern for the two-process, frontend-backend architecture that evolved into what we use today. The OS/2 product, though, went nowhere, because nobody - at least in the end user community - chose to run that OS. Well, I guess more copies of the SPSS OS/2 version were sold than of the DEC Pro version :-)

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: 720-342-5621

From: David Marso <[hidden email]>
To: [hidden email],
Date: 04/15/2014 06:03 AM
Subject: Re: [SPSSX-L] History of measurement levels in SPSS
Sent by: "SPSSX(r) Discussion" <[hidden email]>

" It wasn't until the first Windows version with its greater memory capacity that we were able to switch to the X code base. " Not quite right Jon, SPSS for OS/2 and the Mac version predated the Windoze version by at least a year (maybe 2) and were based on SPSS 4.0. I know this because I was the main OS/2 support person in SPSS technical support for several years beginning in Oct, 1990 -also backed up Cathy F on Mac support-. The OS/2 system was ROCK SOLID. Only calls I got were people trying to print and a funky font issue. Recall the 'good ole' days when a monster PC had 32M of RAM! My OS/2 box had 16M and I waited for 2 weeks to have a math coprocessor installed after I started in teksport. Remember they cut disks in the basement back in those days. ---- Jon K Peck wrote > Still before my time but SPSS went from SPSS 9 to SPSS-X. Clever, no? And > restarted the version numbering. The X system was a massive eXtension of > the old SPSS. In fact, the new central system was so big, that in 1983-84 > when the PC system was first built (which is when I joined SPSS - the > Inc), we had to build it off the old SPSS central system in order to meet > the memory limitations of the target PC. The first PC version ran in 320K > bytes of memory, including DOS, although it could use up to a grand total > of 640K if you could afford that many chips. There was, of course, no > virtual memory on the PC then. The memory limit required a large amount > of memory management work on the SPSS code and module architecture. It > wasn't until the first Windows version with its greater memory capacity > that we were able to switch to the X code base. > > > Jon Peck (no "h") aka Kim > Senior Software Engineer, IBM > peck@.ibm > phone: 720-342-5621 > > > > > From: Mike Palij < > mp26@ > > > To: > SPSSX-L@.uga > , > Date: 04/14/2014 05:21 PM > Subject: Re: [SPSSX-L] History of measurement levels in SPSS > Sent by: "SPSSX(r) Discussion" < > SPSSX-L@.uga
> > > > > > I have not run SPSS on a DEC-10 but started to use it on a > Sperry UNIVAC 1100 in the late 1970s and later on various > other machines (I preferred to use BMDP because I dealt > primarily with experimental designs and SPSS' had very > limited ANOVA capabilities in comparison; Ivor Francis > reviewed ANOVA software in the Dec 1973 JASA and > notes that SPSS did not have an ANOVA procedure!). > However, my memory was that on the different systems, > SPSS was indifferent to the use of capitals/lower case > for variable names, treating DRUG/drug/Drug the same. > > However, this does not mean that the PDP-10 version might > not have had this capability. I did a web search for this > version and came across a PDF of a manual developed by > the computer center at the Australian University of Queensland. > This manual describes the statistical packages available at the > university, including the PDP-10 version of SPSS. You > should be able to access the PDF from the following link: >http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CCgQFjAA&url=http%3A%2F%2Fbitsavers.informatik.uni-stuttgart.de%2Fwww.computer.museum.uq.edu.au%2Fpdf%2FMNT-3%2520Statistical%2520Packages%2520Revised%2520Edition%2520August%25201981.pdf&ei=qVNMU5zXMvffsASrnIK4Dw&usg=AFQjCNH9mk3pcSQUuwcp7zEOYqIKcOMNEg&sig2=_aMz04KIeYO51wieVzzFJw> > I quote a short section of what was called "SPSS-10": > > |5.2 The PDP-10 Implementation of SPSS > | > |The PDP-10 implementation of SPSS (SPSS-10) was produced > |at the University of Pittsburgh and is a conversion of the National > |Opinion Research Centre's SPSSH Version 7.01. SPSS-10 > |closely follows the documentation in: > | > |Nie, Hull, Jenkins, Steinbrunner, Bent > |Statistical Package for the Social Sciences > |(Second Edition) > |McGraw-Hill Book Company > |(1975) > | > |and > | > |Hull and Nie > |SPSS Update > |New Procedures and Facilities for Releases 7 and 8 > |McGraw-Hill Book Company > |(1979) > | > |Any differences between the PDP-10 implementation and that > |described in the McGraw-Hill manual are documented below. > |A more comprehensive list of differences may be obtained from > |DOC: by . > |.PRINT DOC:SPSS.DOC > |and > |.PRINT DOC:SPSS8.DOC > > There are a couple of aspects about SPSS that many may not > know or simply forgotten, such as: > > (1) During the 1970s there were a couple of different versions of > SPSS. In the appendices of the maroon 1975 SPSS manual, > there is the "regular" SPSS (SPSSH) on different systems (but > no DEC PDP systems) as well as a "maxiversion" (which allowed > up to 1000 variables instead of the 500 vars in regular SPSS plus > other features) and a "miniversion" of SPSS (called SPSSG) for > small systems. Some of these may have had certain capabilities > that were absent from others. > > (2) In the manual that I link to above, there is a review of some > of the procedures that SPSS-10 can do and there is one procedure > that I was unaware of; I quote the manual: > > |TETRACHORIC VARIABLES = variable list (low value, high value)/ > | CORRELATIONS = variable list WITH variable list / > | variable list WITH variable list / ... > | > |TETRACHORIC is an SPSSprocedure which computes tetrachoric > |correlation coefficients between dichotomous variables. > > This is not in the 1975 manual but it might be in the 1979 update > which was for SPSS version 7 and 8; the reference for this book > is the following: > > Hull, C. H., & Nie, N. H. (1979). Statistical Package for the Social > Sciences: New Procedures and Facilities for Releases 7 and 8. Mac Graw > Hill Book Company. > > I admit that I don't think I have ever seen the above reference. > But it possible that versions 7 and 8 as well as SPSS-10 had > certain capabilities that disappeared in subsequent versions. > > (3) There is a bit of a surprise in seeing SPSS version 7 and 8 > because these do not correspond to the current numbering of > SPSS version. I believe that after version 8 SPSS engaged > in the (ill-conceived?) name change to SPSSX (which is why > this list is SPSSX-L) and then started renumbering the version > from either 1 or 2 (a copy of what appears to be the manual for > SPSSX version 2 is listed on books.google.com and can be > seen here: >http://books.google.com/books?ei=OGhMU6W1B9HMsQSryYDgCw&id=YWNVAAAAMAAJ&dq=spssx&focus=searchwithinvolume&q=tetrachoric> NOTE: the tetrachoric procedure appears to have been removed. > > There would be another SPSSX release (ver 3) and then SPSS > went back to it's original name. The 1990 "White" manual contains > version 4 of SPSS and the numbering of releases/version continue > to the present day. > > I guess one would have to look at the all SPSS manuals both from > the company those produced by university computer centers, > such as the Vax version of SPSS-X release 2.1; see: >http://books.google.com/books?id=PWBVAAAAMAAJ&dq=spssx&source=gbs_similarbooks> ) > to determine what may have once been possible to do. > > So, it is possible that certain capabilities and procedures once > were available in SPSS but disappeared for whatever reason over > the years. It is surprising that SPSS once had procedure for > tetrachorics correlations but then removed it, causing programmers > to write their own macros/syntax to calculate it. One wonders > what the story is behind that. > > -Mike Palij > New York University
> mp26@ > > > ----- Original Message ----- > From: Rich Ulrich > To: > SPSSX-L@.UGA > Sent: Monday, April 14, 2014 3:44 PM > Subject: Re: History of measurement levels in SPSS > > > I remember being amazed and I remember writing a small test > that proved it; I don't think it was some other package. > > Maybe that feature/bug never existed beyond the DEC-10 version; > or, even, maybe it never existed beyond a test of the DEC-10 version > at the University of Pittsburgh. - That is not beyond the range of what > is > possible - > > STORY. > When I started using SPSS at the Pitt Computer Center in 1974, SPSS was a > mainframe package; there were no PCs. SPSS was not native to DEC-10, > and therefore each version needed adaptations or conversions. Pitt had an > excellent mainframe operation at that time. Partly as evidence of that, > they > held the franchise from SPSS for converting to DEC-10 and other 36-bit > environments. (I think "36-bit word" was the controlling factor.) As to > excellence: I visited my alma mater, the University of Texas, in 1977 > (probably), > and the person I talked to at the computer center was highly pleased at > the > SPSS and BMDP upgrade kits they received from Pitt -- the mag-tapes always > > came with full instructions and always worked the first time. It was > later that > SPSS took the conversions in-house. > > A fellow named Mike Matzek (sp?) at Pitt was in charge of conversions for > both SPSS and BMD. I came to know him through my various SPSS questions. > The on-site Help Desk at Pitt required that basically everyone, including > systems > programmers, take a turn. It was a very pleasant change to ask a question > and > get back, instead of "What's SPSS do?", an comment like, "Oh, yeah, I > noticed that > the manual is ambiguous there, when I was testing out the conversion." > > This was all years before I ever heard the term for a package, "beta > release". > I do find it conceivable that Mike could have let Pitt users test a new > release. > I know we paid attention to new versions, because they implied new > manuals; > but I don't remember today whether there were messages about minor fixes > that > today would be called "patches". > > -- > Rich Ulrich > > > > > Date: Mon, 14 Apr 2014 10:52:03 -0500 > From: > oliverr@.ibm > Subject: Re: History of measurement levels in SPSS > To: > SPSSX-L@.UGA > > Case-preserving for display purposes came long after 1980. I am also > skeptical that variable names were ever case-sensitive. > > Rick Oliver > Senior Information Developer > IBM Business Analytics (SPSS) > E-mail: > oliverr@.ibm > > > > From: Jon K Peck/Chicago/IBM@IBMUS > To: > SPSSX-L@.uga > , > Date: 04/14/2014 10:42 AM > Subject: Re: History of measurement levels in SPSS > Sent by: "SPSSX(r) Discussion" < > SPSSX-L@.uga > > > > > > > > Before my time, but I am skeptical. Before SPSS preserved the casing of > variable names, DRUG and Drug etc would both have referred to the SAME > variable. The change was only cosmetic and had no functional effect. > > > Jon Peck (no "h") aka Kim > Senior Software Engineer, IBM > peck@.ibm > phone: 720-342-5621720-342-5621 > > > > > From: Rich Ulrich < > rich-ulrich@ > > > To: > SPSSX-L@.uga > , > Date: 04/14/2014 09:29 AM > Subject: Re: [SPSSX-L] History of measurement levels in SPSS > Sent by: "SPSSX(r) Discussion" < > SPSSX-L@.uga > > > > > > > > I remember one side-effect of the original implementation of preserving > the case of variable names: For at least one release I used, I could set > up separate variables named DRUG, drug, Drug, .... So long as the > capitalization was different, the variable was different. > > I remember being astounded by that, and writing an example to prove it. > That would have been on a mainframe, probably about 1980, running on > a DEC-10. > > -- > Rich > > > > Date: Mon, 14 Apr 2014 08:08:16 -0500 > From: > oliverr@.ibm > Subject: Re: History of measurement levels in SPSS > To: > SPSSX-L@.UGA > > The ordinal measurement level is never assigned automatically. Variables > without a specified or previously saved measurement level are assigned > either the scale (continuous) or nominal measurement level. > > Variable names are not case-sensitive, but the case in which they were > originally defined is preserved. So if you define a variable name as Fred, > it will be displayed as "Fred", but you can refer to that variable in > command syntax as "FRED" or "fred" or "FrEd", etc. AFAIK, variable and > value labels have always been case-sensitive since they are quoted > strings. > > I think Jon is correct concerning when measurement level was introduced. > Preserving case of defined variable names may have been at the same time > or a few releases earlier. > [...] > > Call > Send SMS > Add to Skype > You'll need Skype CreditFree via Skype > > Call > Send SMS > Add to Skype > You'll need Skype CreditFree via Skype ----- 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/History-of-measurment-levels-in-SPSS-tp5725407p5725445.htmlSent 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