Re: st: calculating transitions in survey data (SVY)

[Austin Nichols <austinnichols@gmail.com>]

Marilena <mekomo@essex.ac.uk>:

Ah--those are usually called "waves" or "rounds" of a longitudinal
survey. The uneven spacing complicates your interpretation of
transition rates.  If you are willing to assume the transition matrix
satisfies the Markov properties, you can take a root of the matrix to
get a set of estimated monthly rates, i.e. assume the matrix M_27
measuring where people are after 27 months is m^27 where m is the
matrix of month-to-month transitions. And M_24 measuring where people
are 24 months later is m^24 which means the comparison of [9 months to
3 years] to [3 years to 5 years] is m^27/m^24, or more generally
perhaps you want to estimate a monthly probability of each transition
adjusting for time at risk if ages differ across individuals in the
survey.

On Thu, Jan 30, 2014 at 6:58 AM, Komodromou, Maria E <mekomo@essex.ac.uk> wrote:
> Thanks.
>
> Sorry i should have explained. I have a birth cohort survey and sweep means each time the survey is repeated ie in 9 months 3 years 5 years after the birth of the child. I am interested to see transitions in and out of employment of parents when the child is 9 months (sweep 1) 3 years (sweep2) 5 years (sweep3).
>
> Hope this clarifies it.
>
> Sent from my iPhone
>
> On 30 Jan 2014, at 11:44, "Austin Nichols" <austinnichols@gmail.com> wrote:
>
>> Marilena <mekomo@essex.ac.uk>:
>>
>> I don't know what you mean by sweep. In labor economics, "sweep"
>> usually refers to the range of wage levels affected by a policy e.g. a
>> minimum wage.  But to measure transition rates in employment, tab
>> employment versus lagged employment. In that tab, you probably want
>> column proportions; if you tab lagged employment versus employment,
>> you probably want row proportions (to see the fraction of people
>> transitioning from employed to unemployed, for example).  Note that
>> this works for multistate variables as well, e.g. if you have many
>> labor force status categories in your employment variable.
>>
>> On Thu, Jan 30, 2014 at 5:57 AM, Komodromou, Maria E <mekomo@essex.ac.uk> wrote:
>>> Hi Austin
>>>
>>> Thank you so much. Just to see if I understand correctly I basically generate a lag for each sweep and I include it with my variables in svy:tab?
>>>
>>> Best,
>>>
>>> Marilena
>>>
>>> On 29 Jan 2014, at 21:59, "Austin Nichols" <austinnichols@gmail.com> wrote:
>>>
>>>> see
>>>>
>>>> http://www.stata.com/statalist/archive/2011-04/msg01110.html
>>>>
>>>> On Wed, Jan 29, 2014 at 3:36 PM, Komodromou, Maria E <mekomo@essex.ac.uk> wrote:
>>>>> Hi,
>>>>>
>>>>> I am using the millennium cohort survey which given its design i use the survey (svy) commands. i would like to calculate some transitions between sweeps (i.e in/out of employment, changes in marital status etc). Does anyone know a command that I can utilise with svy?
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Marilena
>>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/