iedropone
Title
iedropone - an extension of the command drop with features preventing additional observations are unintentionally dropped.
Syntax
For a more descriptive discussion on the intended usage and work flow of this command please see the DIME Wiki.
iedropone [if] , [ numobs(integer) mvar(varname) mval(list of values) zerook ]
| options | Description |
|---|---|
| numobs(integer) | Number of observations that is allowed to be dropped - default is 1 |
| mvar(varname) | Allows that no observation is dropped |
| mval(list of values) | Variable for which multiple values should be dropped - must be used together with mval() |
| zerook | The list of values in mvar() that should be dropped - must be used together with mvar() |
Description
This commands might be easier to understand by following the examples below before reading the description or the explanations of the options.
iedropone has the same purpose as the Stata built-in command drop when dropping observations. However, iedropone safeguards that no additional observations are unintentionally dropped, or that changes are made to the data so that the observations that are supposed to be dropped are no longer dropped.
iedropone checks that no more or fewer observations than intended are dropped. For example, in the case that one observation has been identified to be dropped, then we want to make sure that when re-running the do-file no other observations are dropped even if more observations are added to that data set or changed in any other way.
While the default is 1, iedropone allows the user to set any another number of observation that should be dropped. If the number of observations that fit the drop condition is different, then the command will throw an error.
Options
numobs(integer) this allows the user to set the number of observation that should be dropped. The default is 1 but any positive integer can be used. The command throws an error if any other number of observations match the drop condition.
mvar(varname) and mval(list of values) allows that multiple values in one variable are dropped. These two options must be used together.
If the variable in mvar() is a string variable and some of the values in mval() includes spaces, then the list of values in mval() must be listed exactly as in example 4 below. The command loops over the values in mval() and drops the observations that satisfy the if condition and each of the value in mval(). For example:
iedropone if village == 100 , mvar(household_id) mval(21 22 23)is identical to:
iedropone if village == 100 & household_id == 21
iedropone if village == 100 & household_id == 22
iedropone if village == 100 & household_id == 23The default is that exactly one observation should be dropped for each value in mval() unless numobs() or zerook is used. If those options are used then, then they apply to all values in mval() separately.
mval(list of values) - see mvar(varname) above.
zerook allows that no observations are dropped. The default is that an error is thrown if no observations are dropped.
Stored results
Examples
Example 1.
Let’s say that we have identified the household with the ID 712047 to be incorrect and it should be dropped. Identical to drop if household_id == 712047 but it will test that exactly one observation is dropped each time the do-file runs. This guarantees that we will get an error message that no observation is dropped if someone makes a change to the ID. Otherwise we would unknowingly keep this incorrect observation in our data set.
Similarly, if a new observation is added that is the correct household with ID 712047, then both observation would be dropped without warning if we would have used drop if household_id == 712047. iedropone can be used as below to make sure that only that one observations is dropped. And if the data changes such that two observations are dropped, then the command will throw and error.
iedropone if household_id == 712047Example 2.
Let’s say we have added a new household with the ID 712047. In order to drop only one of those observations we must expand the if-condition to indicate which one of them we want to drop.
iedropone if household_id == 712047 & household_head == "Bob Smith"Example 3.
Let’s say we added a new household with the ID 712047 but we want to drop exactly both of them, then we can use the option numobs().
The command will now throw an error if not exactly two observations have the household ID 712047.
iedropone if household_id == 712047, numobs(2)Feedback, bug reports and contributions
Please send bug-reports, suggestions and requests for clarifications writing “ietoolkit iedropone” in the subject line to: dimeanalytics@worldbank.org
You can also see the code, make comments to the code, see the version history of the code, and submit additions or edits to the code through GitHub repository for ietoolkit.