crosstabs comparison question

[Maguin, Eugene]

I have two crosstabs, each for the same variable pair, but for two different samples. What I’d like to do is to characterize the difference between the two crosstabs as wholes and cell by cell. Is it possible to do this with the Hiloglinear command? I’ve fit several different models but even the closest model, v1  v2 sample, doesn’t reproduce the sample-specific expected cell counts/total percents. Is there a better option? (I confess I know very little about these types of models; they never existed in the psychology graduate stats sequence and they don’t often figure now.)

 

Thanks, Gene maguin

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

[Jon Peck]

I don't know about HILOG, but for a visual, if you don't have a huge number of categories, a combined bar chart might be helpful in making the comparison.

To do this, you would combine the two samples and turn on split files - compare groups for the sample indicator and run the crosstab selecting only the total percentages for the cell statistics (or some other appropriate percent).  Alternatively, put the sample indicator in the layer, which then can give you statistics for each sample and the whole sample in a single table.

Then double click the table, select all the cells, right click, and choose Create Graph > Bar.

Edit the chart to remove the totals category if it was selected or to rearrange the categories.  You could also use the pivot trays to rearrange the table as desired before doing the graph.

For another visual comparison, you could run the two crosstabs without split files and then use STATS TABLE CALC to replace the cells in one table with the difference between the two tables, assuming that the category values are the same.  Then use SPSSINC MODIFY TABLES to color code positive and negative differences.  That would require a fancy use of TABLE CALC to bring the two tables together, but I can  help with that if you want to go that route.

On Wed, Mar 14, 2018 at 11:56 AM, Maguin, Eugene <[hidden email]> wrote:

I have two crosstabs, each for the same variable pair, but for two different samples. What I’d like to do is to characterize the difference between the two crosstabs as wholes and cell by cell. Is it possible to do this with the Hiloglinear command? I’ve fit several different models but even the closest model, v1  v2 sample, doesn’t reproduce the sample-specific expected cell counts/total percents. Is there a better option? (I confess I know very little about these types of models; they never existed in the psychology graduate stats sequence and they don’t often figure now.)

 

Thanks, Gene maguin

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

[David Marso]

You could look at V1, V2, sample and see if there is any "interaction" of
sample with either V1, V2 or V1 x v2.  I don't have LOGLINEAR or HILOGLINEAR
on my installation.



-----
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?"
--
Sent from: http://spssx-discussion.1045642.n5.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

[Rich Ulrich]

Remember BMDP?

After I switched to SPSS for everything else, I used BMDP for multi-way

tables like this one, and for simple Repeated Measures. That was partly

for the standard options available, but mostly for the fine layout of results.

Wiki tells me that BMDP  "is no longer available as of 2017". 

--

Rich Ulrich

---

From: SPSSX(r) Discussion <[hidden email]> on behalf of David Marso <[hidden email]>
Sent: Thursday, March 15, 2018 7:31:40 AM
To: [hidden email]
Subject: Re: crosstabs comparison question

 

You could look at V1, V2, sample and see if there is any "interaction" of
sample with either V1, V2 or V1 x v2.  I don't have LOGLINEAR or HILOGLINEAR
on my installation.



-----
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?"
--
Sent from: http://spssx-discussion.1045642.n5.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

[David Marso]

I used BMDP back in the late 80's.  It was a Multivariate Statistics course
using the book by Barbara Tabachnick and Linda Fidel.  The both taught at
CSUN where I too my BA and MA in Psych.  The book used SPSS and BMDP in
parallel.  
I recall that SPSS bought it and didn't pursue any further development (They
went into a buying frenzy at one point).    Looks like it was sold off to
Statistical Solutions along the way.

Meanwhile maybe Gene can post his two Xtabs and his LOGLINEAR/HILOG syntax
and we can go from there?
--

Rich Ulrich wrote

> Remember BMDP?
>
>
> After I switched to SPSS for everything else, I used BMDP for multi-way
>
> tables like this one, and for simple Repeated Measures. That was partly
>
> for the standard options available, but mostly for the fine layout of
> results.
>
>
> Wiki tells me that BMDP  "is no longer available as of 2017".
>
>
> --
>
> Rich Ulrich
>
> ________________________________
> From: SPSSX(r) Discussion &lt;

> SPSSX-L@.UGA

> &gt; on behalf of David Marso &lt;

> david.marso@

> &gt;
> Sent: Thursday, March 15, 2018 7:31:40 AM
> To:

> SPSSX-L@.UGA

> Subject: Re: crosstabs comparison question
>
> You could look at V1, V2, sample and see if there is any "interaction" of
> sample with either V1, V2 or V1 x v2.  I don't have LOGLINEAR or
> HILOGLINEAR
> on my installation.
>
>
>
> -----
> 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?"
> --
> Sent from: http://spssx-discussion.1045642.n5.nabble.com/
>
> =====================
> To manage your subscription to SPSSX-L, send a message to

> LISTSERV@.UGA

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

> LISTSERV@.UGA

>  (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD





-----
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me.
---
"Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis."
Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?"
--
Sent from: http://spssx-discussion.1045642.n5.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 Donatello]

I used BMDP as well (on a VMS system, which shows my age). The table output was great.

On Thu, Mar 15, 2018 at 2:44 PM, David Marso <[hidden email]> wrote:

I used BMDP back in the late 80's.  It was a Multivariate Statistics course
using the book by Barbara Tabachnick and Linda Fidel.  The both taught at
CSUN where I too my BA and MA in Psych.  The book used SPSS and BMDP in
parallel.
I recall that SPSS bought it and didn't pursue any further development (They
went into a buying frenzy at one point).    Looks like it was sold off to
Statistical Solutions along the way.

Meanwhile maybe Gene can post his two Xtabs and his LOGLINEAR/HILOG syntax
and we can go from there?
--

Rich Ulrich wrote
> Remember BMDP?
>
>
> After I switched to SPSS for everything else, I used BMDP for multi-way
>
> tables like this one, and for simple Repeated Measures. That was partly
>
> for the standard options available, but mostly for the fine layout of
> results.
>
>
> Wiki tells me that BMDP  "is no longer available as of 2017".
>
>
> --
>
> Rich Ulrich
>
> ________________________________
> From: SPSSX(r) Discussion &lt;

> SPSSX-L@.UGA

> &gt; on behalf of David Marso &lt;

> david.marso@

> &gt;
> Sent: Thursday, March 15, 2018 7:31:40 AM
> To:

> SPSSX-L@.UGA

> Subject: Re: crosstabs comparison question
>
> You could look at V1, V2, sample and see if there is any "interaction" of
> sample with either V1, V2 or V1 x v2.  I don't have LOGLINEAR or
> HILOGLINEAR
> on my installation.
>
>
>
> -----
> 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?"
> --
> Sent from: http://spssx-discussion.1045642.n5.nabble.com/
>
> =====================
> To manage your subscription to SPSSX-L, send a message to

> LISTSERV@.UGA

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

> LISTSERV@.UGA

>  (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD





-----
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me.
---
"Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis."
Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?"
--
Sent from: http://spssx-discussion.1045642.n5.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