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: drop observations - if statement
From
Nick Cox <[email protected]>
To
"[email protected]" <[email protected]>
Subject
Re: st: drop observations - if statement
Date
Thu, 6 Mar 2014 15:19:00 +0000
I can't answer that for you.
It's not necessarily a good idea to -drop- observations just because
some values are missing. That depends on your research goals, and I
have no idea what they are. Also, missing values do little harm,
because Stata usually ignores them.
It's a conditional: if you want to -drop- data with missings,
-dropmiss- (SJ) is just one tool. But it is, necessarily, no more than
a wrapper for -drop-, so whatever it does you can do by other means.
Nick
[email protected]
On 6 March 2014 15:11, R Zhang <[email protected]> wrote:
> Nick,
>
> thanks so much! I ran di missing("."), it display 0, so . is not missing.
>
> should I still use the user written command you suggested to drop observations?
>
> -Rochelle
>
> On Thu, Mar 6, 2014 at 6:52 AM, Nick Cox <[email protected]> wrote:
>> Note also that in general Stata is highly literal when dealing with
>> strings. Indeed it could hardly be anything else.
>>
>> So, when you see instances of "." that you want to -drop- but you ask
>> that Stata -drop- instances of "", your command cannot have the
>> desired effect.
>>
>> The user-written command -dropmiss- (Stata Journal) has an option
>> -piasm-, comprehensible as "period is also string missing" for the
>> convenience of users who wish to regard "." as missing. The existence
>> of this option is evidence for the general rule, not against it!
>>
>> Nick
>> [email protected]
>>
>>
>> On 6 March 2014 10:35, Nick Cox <[email protected]> wrote:
>>> "." may seem to mean missing in two possible senses;
>>>
>>> 1. Users may employ that coding for missing, but that's a personal
>>> choice. Stata pays no special attention.
>>>
>>> 2. Stata will return numeric missing to -real(".")-
>>>
>>> . di real(".")
>>>
>>> .
>>>
>>> Notice that Stata displayed a period or stop, meaning numeric (often
>>> called system) missing.
>>>
>>> But it does that to anything that can't be translated to a number:
>>>
>>> . di real("Stata")
>>> .
>>>
>>> Is "." missing? Stata will answer this question for you
>>>
>>> . di missing(".")
>>> 0
>>>
>>> 0 means false.
>>>
>>> Nick
>>> [email protected]
>>>
>>>
>>> On 6 March 2014 03:03, R Zhang <[email protected]> wrote:
>>>> thank you Jeph!
>>>>
>>>> I did "count if missing(cusip)", it shows zero observation. but when
>>>> the data is openned in browse mode, cusip field is displayed as "." ,
>>>> is that not missing?
>>>>
>>>> On Wed, Mar 5, 2014 at 6:00 PM, Jeph Herrin <[email protected]> wrote:
>>>>> Are you sure any are missing?
>>>>>
>>>>> Try
>>>>>
>>>>> count if missing(cusip)
>>>>>
>>>>> and see what you get.
>>>>>
>>>>> hth,
>>>>> Jeph
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> On 3/5/2014 5:41 PM, R Zhang wrote:
>>>>>>
>>>>>> Hi all,
>>>>>>
>>>>>> I used the following commands to drop observations based on a string
>>>>>> variable - CUSIP- that takes missing values, i.e. delete observations
>>>>>> with missing CUSIP,
>>>>>>
>>>>>> drop if cusip==""
>>>>>>
>>>>>> cusip is str10
>>>>>>
>>>>>> (0 observations deleted)
>>>>>>
>>>>>> could you explain why "(0 observations deleted)"?
*
* 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/