2
$\begingroup$

How to interpret white noise test from Stata? Have a white noise means that a variable does not have autocorrelation.

Can I say here that all of my variables has a white noise (does not have autocorrelation)? How can I see it?

 wntestq a

 Portmanteau test for white noise
 ---------------------------------------
  Portmanteau (Q) statistic =    28.0682
  Prob > chi2(40)           =     0.9221

  wntestq b

 Portmanteau test for white noise
 ---------------------------------------
  Portmanteau (Q) statistic =   162.3201
  Prob > chi2(40)           =     0.0000


 wntestq  c

 Portmanteau test for white noise
 ---------------------------------------
  Portmanteau (Q) statistic =   162.6615
  Prob > chi2(40)           =     0.0000


 wntestq  d

 Portmanteau test for white noise
 ---------------------------------------
  Portmanteau (Q) statistic =   192.8795
  Prob > chi2(40)           =     0.0000


 wntestq  e

 Portmanteau test for white noise
 ---------------------------------------
  Portmanteau (Q) statistic =   182.2451
  Prob > chi2(40)           =     0.0000


 wntestq  f

 Portmanteau test for white noise
 ---------------------------------------
  Portmanteau (Q) statistic =   126.0059
  Prob > chi2(40)           =     0.0000
$\endgroup$
  • 1
    $\begingroup$ If a P-value reported as 0.9221 has the same interpretation as one reported as 0.0000, then anything goes. $\endgroup$ – Nick Cox Nov 17 '14 at 9:40
  • 1
    $\begingroup$ Note that a White test is not the same as a white noise test. $\endgroup$ – Nick Cox Nov 17 '14 at 9:40
  • 2
    $\begingroup$ Well documented at stata.com/manuals13/tswntestq.pdf $\endgroup$ – Nick Cox Nov 17 '14 at 9:43
3
$\begingroup$

In the Portmanteau test, the null hypothesis is that the variable follows a white noise process. And just like in any other statistical inference, a p-value that is smaller than the significance level, α (the probability of rejecting a true null hypothesis, usually set at α=0.05) means that you reject the null and conclude that the variable is not a white noise. On the other hand, if the p-value is larger than α, then you cannot reject the null, implying that the variable is indeed a white noise process. In your case, the results of the test show that the variable a is the only variable that follows a white noise process.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.