Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: how to do a weighted transition matrix
From
Nick Cox <[email protected]>
To
[email protected]
Subject
Re: st: how to do a weighted transition matrix
Date
Fri, 29 Apr 2011 19:58:47 +0100
-xttrans- does not support weights, so I don't see the point to your
question about -xttrans-.
Austin Nichols gave hints in his email about how to use weights in this area.
Nick
On Fri, Apr 29, 2011 at 7:50 PM, Hey Sky <[email protected]> wrote:
> Dear Nick
>
> thanks for your time and kind answer.
>
> I think now the question connects to how Stata deals weight.
>
> what does it mean when I use aweight to the command tab? it seems it should not
> change the result since the help file says aweight is inversely proportional to
> the VARIANCE of an observation, usually represent averages, but not the value
> itself. but it does have difference when with/out using the weight.
>
>
> do you think the weight for xttrans in Stata, if it has the option, has the same
> principle as tab? that is, the weight multiplies the value? if not, what would
> it be to get the weighted value with command tab?
>
> thanks for your time and kind answer again
>
> Nan
> from Montreal
>
>
>
> ----- Original Message ----
> From: Nick Cox <[email protected]>
> To: [email protected]
> Sent: Fri, April 29, 2011 1:54:45 PM
> Subject: Re: st: how to do a weighted transition matrix
>
> Thanks for the detail. From what you tell us, I gather that the
> weighted matrix of transition frequencies is just the actual matrix
> multiplied by the constant. The transition probabilities are
> unaffected by a multiplicative constant.
>
> Nick
>
> On Fri, Apr 29, 2011 at 5:37 PM, Hey Sky <[email protected]> wrote:
>> Dear Nick
>>
>> thanks for your reply.
>>
>> my research topic is about the transition probability from one state to
>>another,
>> e.g., white/blue collar job and self-employment within an unbalanced panel.
> all
>> the three states are mutural exclusive dummy variables. the weight is a
>>constant
>> number but not integer. it adjust the percentage of different people groups,
>> e.g., there is too many refugee in the survey but less other immigrants so
> that
>> the stats bureau uses the weight to adjust it back to represent the real
>>world.
>>
>>
>> the following created data are almost the same with the actural one, though I
>>am
>> not sure about the value of the weight. I checked it once long time ago and
>> vaguely remember it is just a constant number (I cannot check not due to the
>> data is not at hand). I used xttrans to get the transition matrix but I have
>> been required to use weights otherwise I cannot take the results to the
> public.
>>
>>
>> hope I make my question clear and thanks for any suggetsions.
>>
>> clear
>> input id time w b s weight
>> 1 1 1 0 0 1.8
>> 1 2 1 0 0 1.8
>> 1 3 0 1 0 1.8
>> 2 1 0 1 0 1.8
>> 2 2 0 1 0 1.8
>> 2 3 0 0 1 1.8
>> 3 1 0 0 1 1.8
>> 3 2 0 1 0 1.8
>> 3 3 . . . 1.8
>> 4 1 0 0 1 1.8
>> 4 2 0 1 0 1.8
>> 4 3 1 0 0 1.8
>> end
>>
>>
>> Nan
>> from Montreal
>>
>>
>>
>>
>> ----- Original Message ----
>> From: Nick Cox <[email protected]>
>> To: [email protected]
>> Sent: Fri, April 29, 2011 12:06:46 PM
>> Subject: Re: st: how to do a weighted transition matrix
>>
>> More explanation would help. Where do the weights come from? Do they
>> weight the transitions? The earlier state? The later state? Are they
>> integers or not? What is your data structure? Which variables hold
>> which information? Do you expect a precise answer to a vague question?
>>
>> At its easiest, this sounds like a transition matrix multiplied
>> elementwise by a weights matrix, so save the matrices independently
>> and do the multiplication in Mata.
>>
>> Nick
>>
>> On Fri, Apr 29, 2011 at 4:57 PM, > Nan from Montreal, a.k.a. Hey Sky
>> <[email protected]> wrote:
>>
>>> I have to do a weighted transition matrix but it is not an option for the
>>> command xttrans.
>>>
>>> I have searched the archives but did not find any discussion about
>>> this question.
>>> did anyone experience this before and any (other) method to deal with it?
>>
>> *
>> * 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/
>>
>
> *
> * 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/
>
*
* 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/