this will take care of multiple incomes of the same value
***************************** start of poverty.ado ************************
*! program to calculate mean income around a given poverty level
*! syntax: poverty incomeVar povertyLevel window
*! example: poverty gdppc 12000 500
program define poverty, sortpreserve
version 9.2
args inc inclev n
/*
inc = income variable
inclev = poverty level (number)
n = window (how many incomes above and below poverty level to include)
*/
tempvar diff match id
sort `inc'
gen `id' = _n
gen byte `match' = 0
gen `diff' = abs(`inc' - `inclev')
qui summ `diff'
qui replace `match' = 1 if `diff' == r(min)
qui count if `match'==1
if r(N) > 1 {
di
di as res "Note: More than one income matches the poverty level of `inclev'"
}
qui summ `id' if `match' == 1
local min = r(min)
local max = r(max)
if (`min' - `n' <= 0) | (`max' + `n' > _N) {
di as err "There are not enough observations around the poverty level"
exit
}
di
di as res "Summary of income for households above and below the
poverty level of `inclev'"
summ `inc' if (_n >= `min' - `n') & (_n <= `max' + `n')
end
exit
***************************** end of poverty.ado ************************
On 7/22/08, Sergiy Radyakin <[email protected]> wrote:
> Dear Ana,
>
> probably something like this to get started:
>
> sysuse auto, clear
> sort price
> local price_line=5079 /* 228.22 */
> local window=5
> tempvar obsnum
> gen `obsnum'=_n
> sum `obsnum' if price==`price_line'
> sum price in `=`=r(min)'-`window''/`=`=r(max)'+`window''
>
> Note that there might be more than 1 household right at the level of
> poverty line. The above code will take all of them, plus window
> households on the sides
>
> Also note that a concept of "being at the poverty line" sounds
> strange. Typically there are poor (inc<povline) and non-poor
> (inc>=povline) households. Also the chances of a household having
> income exactly equal to the poverty line are rather low. Also avoid
> direct comparison (income==value) since income is stored as decimals,
> and you may run into precision problems if your poverty line is not a
> round number, but rather generated automatically in your code, etc,
> etc, etc.
>
> Just in case: ADePT Poverty program will do a lot of standard poverty
> tabulations:
> www.worldbank.org/adept
>
> Best regards,
> Sergiy Radyakin
>
> On 7/22/08, Ana R. Rios <[email protected]> wrote:
>> Dear Stata users,
>>
>> Could someone suggest a code that will perform the following?
>> - I have defined a poverty level based on income per capita (incpc). I
>> would like to perform calculations on x number of households above/below
>> poverty.
>>
>> Here is an example: lets assume that the poverty level is an income per
>> capita equal to 228.22. I would like to calculate the mean income per
>> capita for a total sample of 11 households: 5 households above poverty, 5
>> households below povery and the household at the poverty level. That is,
>> the mean income per capita for observations 8 to 18.
>>
>> obs househol incpc
>> 1 18109 15.87111
>> 2 19705 52.32142
>> 3 37503 75
>> 4 25309 80
>> 5 33520 91.14099
>> 6 22905 100
>> 7 4113 150
>> 8 37520 167.7384
>> 9 33102 190
>> 10 28614 192.205
>> 11 28504 227.3124
>> 12 37616 227.3575
>> 13 37620 228.22
>> 14 33518 235.8528
>> 15 33315 236.0431
>> 16 4205 250
>> 17 5210 250
>> 18 23814 252.6725
>> 19 28506 256.8579
>> 20 30912 265
>>
>>
>> Thank you for your time.
>>
>> Regards,
>> Ana Rios
>>
>>
>>
>>
>> *
>> * 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/
