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: Panel data correlation
From
Alfonso Sanchez-Penalver <[email protected]>
To
"[email protected]" <[email protected]>
Subject
Re: st: Panel data correlation
Date
Tue, 1 Apr 2014 14:05:38 -0400
Nick makes two great points, one in each email. There are two dimensions in your data: countries, and time. What correlation do you want? If it's the correlation across countries you can do it by period of time and compare. If it's the correlation across time then you do it by country and compare. The second point he makes is that there is not much information in five periods to give much credit to a correlation coefficient across time in your data.
Alfonso Sanchez-Penalver
> On Apr 1, 2014, at 1:50 PM, Nick Cox <[email protected]> wrote:
>
> Sorry, but I can't add to my previous answer as I don't think you are
> posing a precise question.
>
> What's different may be what makes sense to your project, but pooling
> countries and years just mixes different kinds of variation and would
> produce something difficult to interpret. The point is usually made in
> broader discussions of cross-sectional and panel analyses.
> Nick
> [email protected]
>
>
>> On 1 April 2014 18:43, Anderson Macedo de Jesus <[email protected]> wrote:
>> Nick,
>> I have two variables and I need to check the correlation between them in a panel data, but I have a panel data for 135 countries described as follows:
>>
>> countries years CA PFI
>> Brazil 1990 12 20
>> Brazil 1995 10 5
>> Brazil 2000 40 1
>> Brazil 2005 4 40
>> Brazil 2010 20 30
>> Netherlands 1990 100 30
>> ...
>>
>> I have been dealing with these data for a while and I could notice that it is a bit different to deal with panel data. So, I would like to know whether or not I can use the simple -correlate- command or there is something different. Thats my question. I don't want to make a big mistake
>> Thanks
>> Anderson
>>
>>
>>> On Apr 1, 2014, at 7:31 PM, Nick Cox <[email protected]> wrote:
>>>
>>> The Pearson correlation between two variables can be calculated with
>>> -correlate-. You could restrict that to different years or even
>>> different countries, although I wouldn't pay anything for correlations
>>> based on 5 values. So, what's the precise question?
>>>
>>> Nick
>>> [email protected]
>>>
>>>
>>>> On 1 April 2014 18:25, Anderson Macedo de Jesus <[email protected]> wrote:
>>>>
>>>> I need to check the correlation btw two variables in a panel data in a long format (135 countries with 5 different years for each variable). How can I do that?
>>>> As I could realize it is not possible to run the pearson correlation, right? So, I could I do that?
>>> *
>>> * 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/
>>
>>
>> *
>> * 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/
>
> *
> * 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/
*
* 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/