The repo contains two programmes that help simplify operations with dates
Working with dates in Stata is easy if using the built-in datetime functionality.
Yet, while it is straight forward to compute difference between dates in days, there is no easy way to compute
difference between dates in years, months and days, or their combinations. In addition,
there is no easy way to calculate the date resulting from adding/subtracting a specific number of years/months to a date.
This repo contains two functions that provide such functionality.
To load dateUtilities, include the following line in your do file:
qui do "https://raw.githubusercontent.com/goshevs/datediff/master/dateUtilities.ado"
dateDiff calculates the difference between two dates in years, months and/or days.
dateDiff varlist [if] [in], GENerate(string) [Type(string) replace]
Required arguments
| argument | description |
|---|---|
| varlist | two variables in Stata date format (%td); within observatins, dates in the first variable should precede dates in the second variable |
| GENerate | specifies the user-required output format; see below for details |
Optional arguments
| argument | description |
|---|---|
| type | specified the type of calculation to be conducted, see below for details |
| replace | replaces the varaibles specified in GENerate |
GENerate has the following general syntax:
gen(years = varname1 months = varname2 days = varname3) were varname1 to varname3 are
user-specified names of variables to be created. These variables will contain years, months and days respectively.
GENerate can also be specified in the following ways:
gen(years = varname): result is reported in yearsgen(months = varname): result is reported in monthsgen(days = varname): result is reported in daysgen(years = varname1 months = varname2): result is reported in years and monthsgen(months = varname2 days = varname3): result is reported in months and days
type can take the following values:
age: computation uses the logic of age calculation (the default). For example, one year is the period from Jan 15, 2018 to Jan 14, 2019.time: computation uses the logic of time calculation. For example, one year is the period from Jan 15, 2018 to Jan 15, 2019.
dateShift shifts a date forward or backward in time by a user-specified number of years, months and/or days.
dateShift varlist(max=1) [if] [in], GENerate(name) step(string) ///
[type(string) replace INConsistent(name)]
Required arguments
| argument | description |
|---|---|
| varlist | a datetime variable name |
| GENerate | name of a new variable that will contain the output |
| step | specifies the user-provided input format; see below for details |
Optional arguments
| argument | description |
|---|---|
| type | specified the type of calculation to be conducted, see below for details |
| replace | replaces the variables specified in GENerate |
| INConsistent | flag for inconsistent backward shifts; see below for details |
step has the following general syntax:
step(years = varname1 months = varname2 days = varname3) were varname1 to varname3 are
user-specified names of variables that contain years, months and days respectively.
step can also be specified in the following ways:
step(years = varname): move dates by yearsstep(months = varname): move dates by monthsstep(days = varname): move dates by daysstep(years = varname1 months = varname2): move dates by years and monthsstep(months = varname2 days = varname3): move dates by months and days
Any variable specified in step has to be present in the dataset and has to have valid values.
Valid values are positive or negative integers. Mixed input within an observation,
e.g. negative months and positive days, is not supported at this time. However, it is
possible to shift different observations in different directions
and by different amounts.
type can take the following values:
age: computation uses the logic of age calculation (the default), i.e. the length of the shift includes the start date. Therefore, one year is defined as the period from Jan 15, 2018 to Jan 14, 2019.time: computation uses the logic of time calculation, i.e. the length of the shift does not include the start date. Therefore, one year in this case is defined as the period from Jan 15, 2018 to Jan 15, 2019.
Note: Backward shifts could exhibit minor inconsistencies due to the different number of
days months have. These inconsistencies manifest in a small number of corner cases
such as end-of-the-month start dates that translate to end dates that do not exist (such
as February 30th or June 31st). In such cases, the end date is set to the last valid date prior to the non-existent date.
For example, an age shift of October 30th backwards by one month yields September 30th. To flag inconsistent shifts,
use option INConsistent.
Additional examples are available in examples.do
*** Move bdates by the specified values in variables year, month and day, using age principle of calculation
dateShift bdate, gen(newvar) step(years = year months = month days = day)
*** Compute date difference in years, months and days, using age principle of calculation
dateDiff bdate newvar, gen(years=years months = months days = days) replace
*** Move bdates by the specified values in variable month, using age principle of calculation
dateShift bdate, gen(newvar) step(months = month) replace
*** Compute date difference in months, using age principle of calculation
dateDiff bdate newvar, gen(months = months) replace
*** Additional examples
*** Move bdates by the specified values in variables year and day, using time principle of calculation
dateShift bdate, gen(newvar) step(years = year days = day) type(time) replace
*** Compute date difference in years, using time principle of calculation
dateDiff bdate newvar, gen(years=years1) type(time) replace
*** Compute date difference in months, using age principle of calculation
dateDiff bdate newvar, gen(months = months1) replace
*** Compute date difference in days, using age principle of calculation
dateDiff bdate newvar, gen(days=days1) replace
*** Compute date difference in years and months, using time principle of calculation
dateDiff bdate newvar, gen(years=years2 months = months2) type(time) replace
*** Compute date difference in months and days, using age principle of calculation
dateDiff bdate newvar, gen(months = months3 days = days3) replace