problem with missing cases in Multiple Response

[Spice Weasel]

Dear listers

I'm analyzing the data from a survey of social service agencies that
deliver programs to school-aged children and youth.  There are 45
agencies that deliver the particular program that I'm interested in.
Each agency is 'nested' within 1 of 7 regions.

One of the things that I'm interested in is what dedicated staff (by
profession) is assigned to the program.  There are 10 professions,
coded 0 if the agency does not have this profession and 1 if it does.

When I carry out CROSSTABS by Region separately for each profession,
the output shows 45 valid cases and no missing cases, as it should.

CROSSTABS dedicated_staff_teachers dedicated_staff_EAs
dedicated_staff_addiction_counsellors
    dedicated_staff_attendance_counsellors dedicated_staff_CYWs
dedicated_staff_psychiatrists dedicated_staff_psychologists
     dedicated_staff_social_workers
dedicated_staff_other_health_professionals dedicated_staff_other by
Region
     /CELLS COLUMN COUNT.

However, when I create a Multiple Response set
('$CurrentDedicatedStaff') and CROSSTAB  that by Region, I end up with
only 40 valid cases and 5 missing cases.  Even more curious, when I
compare the counts in the cells to the counts in the separate
CROSSTABS  I have an exact match, except for the total number of
agencies per Region which of course means that the % are all
incorrect.  For example, there's a difference between 6 of 6 agencies
(100%) and 6 of 8 agencies (75%).

Here's my syntax:

MULT RESPONSE GROUPS=$CurrentDedicatedStaff 'Current Dedicated Staff'
(dedicated_staff_teachers
    dedicated_staff_EAs dedicated_staff_addiction_counsellors
dedicated_staff_attendance_counsellors
    dedicated_staff_CYWs dedicated_staff_psychiatrists
dedicated_staff_psychologists
    dedicated_staff_social_workers dedicated_staff_other_health_professionals
    dedicated_staff_other (1))
  /VARIABLES=Region(1 7)
  /TABLES=$CurrentDedicatedStaff BY Region
  /CELLS=COLUMN
  /BASE=CASES.

I can use the values generated by the CROSSTABS to write my report.
The purpose of running the MULT RESPONSE was to generate a 'nice'
table that I could incorporate (cut and paste)  into the report. Since
the % are incorrect, I can't use the table.

Can anyone see what I'm doing wrong here?  Any other suggestions as to
how to generate the table I need would be appreciated.

Thanks

Pat

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

[Jerabek Jindrich]

Dear Pat,

I guess the reason of your problem is this: some agencies have none profession coded as 1. When there is no variable coded 1 in the multiresp set, such case is by SPSS treated as missing.

Suggestion: Make an aditional variable, coded 1 when none of the professions is present.

COMPUTE prof11 = 0.
IF not any (1, prof1 to prof10) prof11 = 11. /*pls replace with your variable names */
VAR LAB prof11 "none ot the professions".

Add the Prof11 "none" to you multiresp set.

HTH
Jindra

> ------------ Původní zpráva ------------
> Od: Pat C <[hidden email]>
> Předmět: problem with missing cases in Multiple Response
> Datum: 27.7.2010 15:47:35
> ----------------------------------------
> Dear listers
>
> I'm analyzing the data from a survey of social service agencies that
> deliver programs to school-aged children and youth.  There are 45
> agencies that deliver the particular program that I'm interested in.
> Each agency is 'nested' within 1 of 7 regions.
>
> One of the things that I'm interested in is what dedicated staff (by
> profession) is assigned to the program.  There are 10 professions,
> coded 0 if the agency does not have this profession and 1 if it does.
>
> When I carry out CROSSTABS by Region separately for each profession,
> the output shows 45 valid cases and no missing cases, as it should.
>
> CROSSTABS dedicated_staff_teachers dedicated_staff_EAs
> dedicated_staff_addiction_counsellors
>     dedicated_staff_attendance_counsellors dedicated_staff_CYWs
> dedicated_staff_psychiatrists dedicated_staff_psychologists
>      dedicated_staff_social_workers
> dedicated_staff_other_health_professionals dedicated_staff_other by
> Region
>      /CELLS COLUMN COUNT.
>
> However, when I create a Multiple Response set
> ('$CurrentDedicatedStaff') and CROSSTAB  that by Region, I end up with
> only 40 valid cases and 5 missing cases.  Even more curious, when I
> compare the counts in the cells to the counts in the separate
> CROSSTABS  I have an exact match, except for the total number of
> agencies per Region which of course means that the % are all
> incorrect.  For example, there's a difference between 6 of 6 agencies
> (100%) and 6 of 8 agencies (75%).
>
> Here's my syntax:
>
> MULT RESPONSE GROUPS=$CurrentDedicatedStaff 'Current Dedicated Staff'
> (dedicated_staff_teachers
>     dedicated_staff_EAs dedicated_staff_addiction_counsellors
> dedicated_staff_attendance_counsellors
>     dedicated_staff_CYWs dedicated_staff_psychiatrists
> dedicated_staff_psychologists
>     dedicated_staff_social_workers dedicated_staff_other_health_professionals
>     dedicated_staff_other (1))
>   /VARIABLES=Region(1 7)
>   /TABLES=$CurrentDedicatedStaff BY Region
>   /CELLS=COLUMN
>   /BASE=CASES.
>
> I can use the values generated by the CROSSTABS to write my report.
> The purpose of running the MULT RESPONSE was to generate a 'nice'
> table that I could incorporate (cut and paste)  into the report. Since
> the % are incorrect, I can't use the table.
>
> Can anyone see what I'm doing wrong here?  Any other suggestions as to
> how to generate the table I need would be appreciated.
>
> Thanks
>
> Pat
>
> =====================
> 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

[Spice Weasel]

That did it. It makes perfect sense but I never would have worked it
out on my own.

Thanks

Pat Cit


2010/7/27 Jerabek Jindrich <[hidden email]>:

> Dear Pat,
>
> I guess the reason of your problem is this: some agencies have none profession coded as 1. When there is no variable coded 1 in the multiresp set, such case is by SPSS treated as missing.
>
> Suggestion: Make an aditional variable, coded 1 when none of the professions is present.
>
> COMPUTE prof11 = 0.
> IF not any (1, prof1 to prof10) prof11 = 11. /*pls replace with your variable names */
> VAR LAB prof11 "none ot the professions".
>
> Add the Prof11 "none" to you multiresp set.
>
> HTH
> Jindra
>
>
>
>> ------------ Původní zpráva ------------
>> Od: Pat C <[hidden email]>
>> Předmět: problem with missing cases in Multiple Response
>> Datum: 27.7.2010 15:47:35
>> ----------------------------------------
>> Dear listers
>>
>> I'm analyzing the data from a survey of social service agencies that
>> deliver programs to school-aged children and youth.  There are 45
>> agencies that deliver the particular program that I'm interested in.
>> Each agency is 'nested' within 1 of 7 regions.
>>
>> One of the things that I'm interested in is what dedicated staff (by
>> profession) is assigned to the program.  There are 10 professions,
>> coded 0 if the agency does not have this profession and 1 if it does.
>>
>> When I carry out CROSSTABS by Region separately for each profession,
>> the output shows 45 valid cases and no missing cases, as it should.
>>
>> CROSSTABS dedicated_staff_teachers dedicated_staff_EAs
>> dedicated_staff_addiction_counsellors
>>     dedicated_staff_attendance_counsellors dedicated_staff_CYWs
>> dedicated_staff_psychiatrists dedicated_staff_psychologists
>>      dedicated_staff_social_workers
>> dedicated_staff_other_health_professionals dedicated_staff_other by
>> Region
>>      /CELLS COLUMN COUNT.
>>
>> However, when I create a Multiple Response set
>> ('$CurrentDedicatedStaff') and CROSSTAB  that by Region, I end up with
>> only 40 valid cases and 5 missing cases.  Even more curious, when I
>> compare the counts in the cells to the counts in the separate
>> CROSSTABS  I have an exact match, except for the total number of
>> agencies per Region which of course means that the % are all
>> incorrect.  For example, there's a difference between 6 of 6 agencies
>> (100%) and 6 of 8 agencies (75%).
>>
>> Here's my syntax:
>>
>> MULT RESPONSE GROUPS=$CurrentDedicatedStaff 'Current Dedicated Staff'
>> (dedicated_staff_teachers
>>     dedicated_staff_EAs dedicated_staff_addiction_counsellors
>> dedicated_staff_attendance_counsellors
>>     dedicated_staff_CYWs dedicated_staff_psychiatrists
>> dedicated_staff_psychologists
>>     dedicated_staff_social_workers dedicated_staff_other_health_professionals
>>     dedicated_staff_other (1))
>>   /VARIABLES=Region(1 7)
>>   /TABLES=$CurrentDedicatedStaff BY Region
>>   /CELLS=COLUMN
>>   /BASE=CASES.
>>
>> I can use the values generated by the CROSSTABS to write my report.
>> The purpose of running the MULT RESPONSE was to generate a 'nice'
>> table that I could incorporate (cut and paste)  into the report. Since
>> the % are incorrect, I can't use the table.
>>
>> Can anyone see what I'm doing wrong here?  Any other suggestions as to
>> how to generate the table I need would be appreciated.
>>
>> Thanks
>>
>> Pat
>>
>> =====================
>> 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