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: normality test using the over identifying moment conditions
From
Maarten Buis <[email protected]>
To
[email protected]
Subject
Re: st: normality test using the over identifying moment conditions
Date
Thu, 31 Jan 2013 14:27:37 +0100
The answer Nick gave you is that you should not perform such a test.
If you really really want to than there are a large number of options
open to you, -swilk- would be one choice.
But what you really want to look at is graphs, e.g. -qnorm- or you can
add evelopes to that plot by combining -qenv- (type in Stata -ssc
install qenv-) and -qplot- (type in Stata -findit qplot-), after
installing these packages look at the helpfile of qenvnormal for an
example.
-- Maarten
On Thu, Jan 31, 2013 at 2:06 PM, Usman Gilani <[email protected]> wrote:
> Dear Nick,
>
> the dataset has 1000 obs.
> and I'm not using Stata this output is from R
> i tried to run this test in stata but cause of limited stata knowledge I couldn't do it.
>
> please tell me how can i do gmm test with following moment conditions in stata
>
> thanks
>
> best
>
> Gilani
> On 31 Jan 2013, at 12:30, Nick Cox <[email protected]> wrote:
>
>> There is no mention here of what command you are using. With this kind
>> of data the number of values is usually so large that any test will
>> produce results significant at conventional levels, i.e. normality
>> will be rejected even for trivial deviations from normality.
>>
>> A search of the archives will show many posts explaining why tests of
>> normality are usually a bad idea.
>>
>> Nick
>>
>> On Thu, Jan 31, 2013 at 12:21 PM, Usman Gilani <[email protected]> wrote:
>>> Hi,
>>> I'm trying to interpret the following results, with respect to "normality
>>> test using the over identifying moment conditions"
>>>
>>> where returns have normal distribution
>>> with parameter mu,sd
>>> and i have 4 moment conditions
>>>
>>>> E[r-mu/sd]=0
>>>
>>>> E[(r-mu)^2/sd-1]=0
>>>
>>>> E[(r-mu)^3/sd^3]=0
>>>
>>>> E[(r-mu)^4/sd^4-3]=0
>>>
>>> output..
>>> gel(g = g, x = returns, tet0 = c(f3$estimate[1], f3$estimate[2]))
>>>
>>> Type of GEL: EL
>>>
>>> Coefficients:
>>> Estimate Std. Error t value Pr(>|t|)
>>> mean -0.01168 0.05614 -0.20805 0.83519
>>> sd 1.77591 0.03965 44.79218 0.00000
>>>
>>> Lambdas:
>>> Estimate Std. Error t value Pr(>|t|)
>>> Lambda[1] -0.09743 0.03912 -2.49028 0.01276
>>> Lambda[2] 0.65728 0.02443 26.90505 0.00000
>>> Lambda[3] 0.03247 0.01304 2.48961 0.01279
>>> Lambda[4] -0.10954 0.00407 -26.90423 0.00000
>>>
>>> Over-identifying restrictions tests: degrees of freedom is 2
>>> statistics p-value
>>> LR test 2.3341e+02 2.0730e-51
>>> LM test 7.2417e+02 5.5954e-158
>>> J test 7.2417e+02 5.5954e-158
>>>
>>> Convergence code for the coefficients: 0
>>>
>>> Convergence code for the lambdas: 0
>>>
>>>
>>> does the J-test p-value rejecting the null E[g(theta,x)]=0, and which moment
>>> condition is true under normality
>>>
>> *
>> * 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/
--
---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany
http://www.maartenbuis.nl
---------------------------------
*
* 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/