bayesmh - error message: "errors in model specification"

[Marcos Almeida]

Dear Statalisters,

I've tried to repeat the example presented by Stata here (https://www.youtube.com/watch?v=0F0QoMCSKJ4) under a very simple data set:

. list

     +-------+
     | heads |
     |-------|
  1. |     0 |
  2. |     0 |
  3. |     1 |
  4. |     0 |
  5. |     0 |
     |-------|
  6. |     1 |
  7. |     1 |
  8. |     0 |
  9. |     0 |
 10. |     1 |
     +-------+

What intrigued me is: whenever I perform these commands

. bayesmh heads, likelihood(dbernoulli({theta})) prior({theta}, beta(1,1))

in the command window, it works fine.

Burn-in ...
Simulation ...

Model summary
------------------------------------------------------------------------------
Likelihood:
  heads ~ bernoulli({theta})

Prior:
  {theta} ~ beta(1,1)
------------------------------------------------------------------------------

Bayesian Bernoulli model                         MCMC iterations  =     12,500
Random-walk Metropolis-Hastings sampling         Burn-in          =      2,500
                                                 MCMC sample size =     10,000
                                                 Number of obs    =         10
                                                 Acceptance rate  =      .4823
Log marginal likelihood = -7.8194591             Efficiency       =      .2291
 
------------------------------------------------------------------------------
             |                                                Equal-tailed
             |      Mean   Std. Dev.     MCSE     Median  [95% Cred. Interval]
-------------+----------------------------------------------------------------
       theta |  .4187117   .1342192   .002804   .4152274   .1746616   .6876875
------------------------------------------------------------------------------

I can change the alpha and beta parameters, and it still works perfectly:

. bayesmh heads, likelihood(dbernoulli({theta})) prior({theta}, beta(30,30))
  
Burn-in ...
Simulation ...

Model summary
------------------------------------------------------------------------------
Likelihood:
  heads ~ bernoulli({theta})

Prior:
  {theta} ~ beta(30,30)
------------------------------------------------------------------------------

Bayesian Bernoulli model                         MCMC iterations  =     12,500
Random-walk Metropolis-Hastings sampling         Burn-in          =      2,500
                                                 MCMC sample size =     10,000
                                                 Number of obs    =         10
                                                 Acceptance rate  =      .4245
Log marginal likelihood = -6.9942639             Efficiency       =      .2376
 
------------------------------------------------------------------------------
             |                                                Equal-tailed
             |      Mean   Std. Dev.     MCSE     Median  [95% Cred. Interval]
-------------+----------------------------------------------------------------
       theta |  .4864921   .0591096   .001213   .4884325    .370204   .6015751
------------------------------------------------------------------------------

However, if I try to use the drop-down menu, exactly as demonstrated by Chuck Huber in the tutorial, I got the message "errors in model specification" whenever clicking on "Create" to define the priors of model parameters.

I'm using Stata 14.1 on Windows 10, and the error message (only under the point-and-click menu) persisted even after I updated the Stata today.

I suspect it might be a bug, probably related to the menu commands.

Thank you in advance for any help!

Kind regards,

Marcos

[Daniel Hernandez]

I have the same problem but using Mac Book Pro OS X El Capitan

Anybody knows what to do?

Thanks

[Marcos Almeida]

Maybe it is a bug with bayesmh or some sort of conflict inside the software, for it seems not to be related to the operating system, be it Windows or Mac.

Let's hope StataCorp notice this and give us good news!

[Nikolay Balov (StataCorp)]

Dear Marcos and Daniel,

Yes, this is a bug in the dialogue system related to bayesmh which we are going to fix at earliest possibility.

Thank you for your patience,
Nikolay

[Marcos Almeida]

Dear Nikolay,

Thank you so much for the reply and the auspicious information!

Best,

Marcos

[Daniel Hernandez]

Thank you both

[Chris Martin]

Hello All, This bug has apparently re-appeared. I'm using Stata/SE 17.0 and I have exactly the same problem as Marcos in #1.