why is spss dropping a category when it shouldn't?

[Greg]

Hi everyone,

I was hoping someone will be able to provide me some advice or suggestions.

I have two variables: nativity status (1: foreign-born & 2: native-born) and citizenship status (1:citizen, 2:not-citizen, and 3: not applicable). I combined the above two variables into 2 new dummies, which look like this:

A) foreign-born & not citizen =1 ; native-born & everyone in B (fbnotcitizen=1 & others=0)
B) foreign-born & citizen = 1  ;     native-born & everyone in A (fbcitizen=1 & others=0)

as for the not-applicables, I made sure i only included the native-born.

When I run logistic regression, spss automatically drops one category. I do not want to include a reference category; I want to include both categories. I'm assuming this should not happen since i'm taking all the foreign-born and dividing them into groups (citizen & non-citizen).

I would greatly appreciate any advice/suggestion as to what to do?

Regards,
Greg

[Bruce Weaver]

grigoris wrote

Hi everyone,

I was hoping someone will be able to provide me some advice or suggestions.

I have two variables: nativity status (1: foreign-born & 2: native-born) and citizenship status (1:citizen, 2:not-citizen, and 3: not applicable). I combined the above two variables into 2 new dummies, which look like this:

A) foreign-born & not citizen =1 ; native-born & everyone in B (fbnotcitizen=1 & others=0)
B) foreign-born & citizen = 1  ;     native-born & everyone in A (fbcitizen=1 & others=0)

as for the not-applicables, I made sure i only included the native-born.

When I run logistic regression, spss automatically drops one category. I do not want to include a reference category; I want to include both categories. I'm assuming this should not happen since i'm taking all the foreign-born and dividing them into groups (citizen & non-citizen).

I would greatly appreciate any advice/suggestion as to what to do?

Regards,
Greg

Why are you not using the original variables?  What questions are you trying to answer with your analysis?

[Maguin, Eugene]

Grigoris,

If you don't have a reference category, what is the other category to be
compared against? Logistic regression is just like normal-theory (regular)
regression with dichotomous variables. If you were predicting height among
3rd and 4th graders and boys and girls, aren't you interested in three
possible comparisons: 3rd vs 4th, boys vs girls, and gender by grader
interaction? Same principle.

I'd also like to comment on the construction of your new dummies. It seems
to me that there is something unusual about their construction but I may
just not understand your intention. The two input variables hasve six
combinations: FB-C, FB-NC, FB-NA, NB-C, NB-NC, NB-NA, where FB is foreign
borh; NB is native born; C is citizen; NC is non-citizen; and NA is not
applicable.

For (A) you have (I think) FB-NC vs NB-C + NB-NC + NB-NA + everyone in B???.
For (B) you have (I think) FB-C vs NB-C + NB-NC + NB-NA + everyone in A???.

It seems like you want to compare FB-NC vs reference group and FB-C vs
reference group. So, this:

FB-NC vs NB-C + NB-NC + NB-NA + FB-C + FB-NA
FB-C vs NB-C + NB-NC + NB-NA + FB-NC + FB-NA

Perhaps you said, this and I just didn't understand.

Gene Maguin



>>>I was hoping someone will be able to provide me some advice or
suggestions.

I have two variables: nativity status (1: foreign-born & 2: native-born) and
citizenship status (1:citizen, 2:not-citizen, and 3: not applicable). I
combined the above two variables into 2 new dummies, which look like this:

A) foreign-born & not citizen =1 ; native-born & everyone in B
(fbnotcitizen=1 & others=0)
B) foreign-born & citizen = 1  ;     native-born & everyone in A
(fbcitizen=1 & others=0)

as for the not-applicables, I made sure i only included the native-born.

When I run logistic regression, spss automatically drops one category. I do
not want to include a reference category; I want to include both categories.
I'm assuming this should not happen since i'm taking all the foreign-born
and dividing them into groups (citizen & non-citizen).

I would greatly appreciate any advice/suggestion as to what to do?

Regards,
Greg
--
View this message in context:
http://old.nabble.com/why-is-spss-dropping-a-category-when-it-shouldn%27t--t
p27789073p27789073.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