STATS CATEGORY ORDER: improvement potentials

[ThW]

Hi @ll and especially hi Jon,

I would like to thank you for the extension command STATS CATEGORY ORDER, that helps a great deal at my work with Statistics. But I am still not perfectly happy with one thing: Whenever I´d like to sort the categories of a variable by count the resulting macro attends only to categories which were mentioned by our respondents. Would it be possible to improve the command by optionally insert OTHERNM and MISSING between the sorted categories and my special values?

E.g.
My desired bottom of the table should be the category 7. All other categories should be sorted by count. But no one of my respondents mentioned categories 2 or 3.

My resulting macro definition is:

DEFINE !catsort ()
4.0 5.0 1.0 6.0 7.0
!ENDDEFINE.

... but I want it to look like this:

DEFINE !catsort ()
4.0 5.0 1.0 6.0 OTHERNM 7.0
!ENDDEFINE.

Thanks a lot!
Th W
Research Manager

[Jon K Peck]

I suppose that you would collect the othernm and missing from the value labels and missing value codes.  Is that right?  I think this is doable except for range missing values.  If you can send me a specific example, I'll look into it.


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




From:        ThW <[hidden email]>
To:        [hidden email],
Date:        08/08/2013 03:22 PM
Subject:        [SPSSX-L] STATS CATEGORY ORDER: improvement potentials
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Hi @ll and especially hi Jon,

I would like to thank you for the extension command STATS CATEGORY ORDER,
that helps a great deal at my work with Statistics. But I am still not
perfectly happy with one thing: Whenever I´d like to sort the categories of
a variable by count the resulting macro attends only to categories which
were mentioned by our respondents. Would it be possible to improve the
command by optionally insert OTHERNM and MISSING between the sorted
categories and my special values?

E.g.
My desired bottom of the table should be the category 7. All other
categories should be sorted by count. But no one of my respondents mentioned
categories 2 or 3.

My resulting macro definition is:

DEFINE !catsort ()
4.0 5.0 1.0 6.0 7.0
!ENDDEFINE.

... but I want it to look like this:

DEFINE !catsort ()
4.0 5.0 1.0 6.0 OTHERNM 7.0
!ENDDEFINE.

Thanks a lot!
Th W
Research Manager




--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/STATS-CATEGORY-ORDER-improvement-potentials-tp5721530.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

[Jignesh Sutar]

Hi Jon, I tried to use this extension for the first time just now and as I

understand it the inputs are either:

1) Categorical variable(s) - for each categorical variable inputted a macro is

returned with the values ordered based on frequency counts.

2) Multiple Response Set(s) (MRSETS) - For each MRS inputted, a new MRS set is

returned with the variables making up that set re-ordered based on frequency

counts.

But what if I wanted to enter a set of binary variables (i.e. v1 v1 v3 v4) and

for the macro to return a order list such as "v2+v3+v1+v4" for use in CTABLES?

I've explored using MRSETS many times in the past but have made it a point to

avoid using MRSETS set and instead use individual dichotomous variables directly

as scale variables and request the SUM statistic for counts and MEAN for

proportion (or % yes) and VALIDN for base. The reason being because particular

variables may be limited to a sub-sample only and using MRSETS there is no way of

identifying and distinguishing these variables in anyway.

With raw dichotomous variables VALIDN can also be requested so that it can be

made explicitly evident the base for each dichotomous variable (useful only if

they vary of course).

Most of the studies I work on generally have such designs so hence preference to

take this approach. However, I was hoping to utilize this module to sort dichotomous

variables in table which otherwise can only be done via using MRSETS.

Could this be implemented in in this module or is there something out there already which tackles this?

Cheers

Jignesh

On 9 August 2013 00:05, Jon K Peck <[hidden email]> wrote:

I suppose that you would collect the othernm and missing from the value labels and missing value codes.  Is that right?  I think this is doable except for range missing values.  If you can send me a specific example, I'll look into it.


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




From:        ThW <[hidden email]>
To:        [hidden email],
Date:        08/08/2013 03:22 PM
Subject:        [SPSSX-L] STATS CATEGORY ORDER: improvement potentials
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Hi @ll and especially hi Jon,

I would like to thank you for the extension command STATS CATEGORY ORDER,
that helps a great deal at my work with Statistics. But I am still not
perfectly happy with one thing: Whenever I´d like to sort the categories of
a variable by count the resulting macro attends only to categories which
were mentioned by our respondents. Would it be possible to improve the
command by optionally insert OTHERNM and MISSING between the sorted
categories and my special values?

E.g.
My desired bottom of the table should be the category 7. All other
categories should be sorted by count. But no one of my respondents mentioned
categories 2 or 3.

My resulting macro definition is:

DEFINE !catsort ()
4.0 5.0 1.0 6.0 7.0
!ENDDEFINE.

... but I want it to look like this:

DEFINE !catsort ()
4.0 5.0 1.0 6.0 OTHERNM 7.0
!ENDDEFINE.

Thanks a lot!
Th W
Research Manager




--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/STATS-CATEGORY-ORDER-improvement-potentials-tp5721530.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

[Jignesh Sutar]

There could be other uses for this feature also. Say for example you wanted to produce a Jaccards' distance matrix, with this feature you could have the variables ordered by those with the highest counts. 

On 21 November 2014 at 15:45, Jignesh Sutar <[hidden email]> wrote:

Hi Jon, I tried to use this extension for the first time just now and as I

understand it the inputs are either:

1) Categorical variable(s) - for each categorical variable inputted a macro is

returned with the values ordered based on frequency counts.

2) Multiple Response Set(s) (MRSETS) - For each MRS inputted, a new MRS set is

returned with the variables making up that set re-ordered based on frequency

counts.

But what if I wanted to enter a set of binary variables (i.e. v1 v1 v3 v4) and

for the macro to return a order list such as "v2+v3+v1+v4" for use in CTABLES?

I've explored using MRSETS many times in the past but have made it a point to

avoid using MRSETS set and instead use individual dichotomous variables directly

as scale variables and request the SUM statistic for counts and MEAN for

proportion (or % yes) and VALIDN for base. The reason being because particular

variables may be limited to a sub-sample only and using MRSETS there is no way of

identifying and distinguishing these variables in anyway.

With raw dichotomous variables VALIDN can also be requested so that it can be

made explicitly evident the base for each dichotomous variable (useful only if

they vary of course).

Most of the studies I work on generally have such designs so hence preference to

take this approach. However, I was hoping to utilize this module to sort dichotomous

variables in table which otherwise can only be done via using MRSETS.

Could this be implemented in in this module or is there something out there already which tackles this?

Cheers

Jignesh

On 9 August 2013 00:05, Jon K Peck <[hidden email]> wrote:

I suppose that you would collect the othernm and missing from the value labels and missing value codes.  Is that right?  I think this is doable except for range missing values.  If you can send me a specific example, I'll look into it.


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




From:        ThW <[hidden email]>
To:        [hidden email],
Date:        08/08/2013 03:22 PM
Subject:        [SPSSX-L] STATS CATEGORY ORDER: improvement potentials
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Hi @ll and especially hi Jon,

I would like to thank you for the extension command STATS CATEGORY ORDER,
that helps a great deal at my work with Statistics. But I am still not
perfectly happy with one thing: Whenever I´d like to sort the categories of
a variable by count the resulting macro attends only to categories which
were mentioned by our respondents. Would it be possible to improve the
command by optionally insert OTHERNM and MISSING between the sorted
categories and my special values?

E.g.
My desired bottom of the table should be the category 7. All other
categories should be sorted by count. But no one of my respondents mentioned
categories 2 or 3.

My resulting macro definition is:

DEFINE !catsort ()
4.0 5.0 1.0 6.0 7.0
!ENDDEFINE.

... but I want it to look like this:

DEFINE !catsort ()
4.0 5.0 1.0 6.0 OTHERNM 7.0
!ENDDEFINE.

Thanks a lot!
Th W
Research Manager




--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/STATS-CATEGORY-ORDER-improvement-potentials-tp5721530.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD