Thank you Richard and Stas for your responses and assistance. It is an
especially welcomed response when it is what you hoped to hear.
In
response to a couple of the questions below: My population of interest
is those 55 and older. I broke them into so many age categories because
I'm looking for the differences at the different points in age past 55.
When I run this using age as a continuous variable I am just able to
illustrate a continual decrease in the dependent variable (which has
already been well documented). It's been an important finding that for
different variations of this dependent variable (courses taken for work
or personal reasons) that the lack of significant differences can be
illustrated between different ages and the reference group (55-59 year
olds). Your point is well taken about the cleaner use of self-created
dummy variables.
I am certainly open to other suggestions and
immensely appreciate all the advice I have received throughout my
dissertation from this listserv.
Carissa
> Date: Sun, 25 Jan 2009 12:58:03 -0500
> To: [email protected]; [email protected]
> From: [email protected]
> Subject: RE: st: subpop question
>
> At 12:30 PM 1/25/2009, Carissa Moffat Miller wrote:
>>1. Below is my output, note _IAgeCat7_7 dropped due to collinearity.
>>This is the category for anyone under 55 so it shouldn't have been in
>>the model, but does this message indicate another problem?
>>
>> 2. I
>>have also run the model with 6 category age just coding age less than
>>55 as missing but this seems akin to misusing if commands instead
>>of the subpop with svy.
>
> Given your sample selection, I would expect _IAgeCat7_7 to be dropped
> (all selected sample members have a value of 0 on it), so I don't see
> any problems.
>
>>Can someone tell me how to include age
>>in the model but also focus on the subpopulation of interest accounting
>>for the survey design? Thank you, Carissa
>
> I'm not sure I understand the question. Are you saying that you
> would like to analyze all age groups while paying particular
> attention to those over 55? If so, it would seem like interaction
> effects can be used. But perhaps I misunderstand you. Also, do you
> really need to break 55+ up into a bunch of age categories? Could
> you just use a dummy coded 1 if 55+, 0 otherwise?
>
> As a sidelight, I hate the way output looks when you use xi. If you
> want something that is quick and dirty it is fine, but if I want
> output that is nice and easy to read I like to create the dummies
> myself with names that are easy to understand.
>
>
> -------------------------------------------
> Richard Williams, Notre Dame Dept of Sociology
> OFFICE: (574)631-6668, (574)631-6463
> HOME: (574)289-5227
> EMAIL: [email protected]
> WWW: http://www.nd.edu/~rwilliam
>
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
