Re: st: Meta-analysis questions

[Rui Liu <ruiliu2010@gmail.com>]

Thank you Tiago! You were right about the randomi/random, they
produced virtually the same results. Except for one case where I have
a very small subgroup and the results varied a tiny bit.

Regarding the metabias commond, you mentioned that the error showed up
because I was trying to "use a command valid only for an older version
of the metabias", what would be the new command? Also, I re-installed
metabias this morning and it's giving me the same error message. I
believe I'm using the most current version of metabias and in the help
metabais, there is a 'by' syntax. So I'm a bit confused.

Rui



On Fri, Sep 24, 2010 at 2:32 AM, Tiago V. Pereira
<tiago.pereira@mbe.bio.br> wrote:
> Specifically, that error shows up because you are trying to use a command
> valid only for an older version of -metabias-
>
>
> I just would like to mention that I have done some confusion with the begg
> option in -metabias-.
>
> As the help file shows " graph - graph displays a Galbraith plot (the
> standard normal deviate of intervention effect estimate against its
> precision) for the original
>        Egger test or a modified Galbraith plot of Z/sqrt(V) versus
> sqrt(V) for Harbord's modified test. ****There is no corresponding
> plot
> for the
>        Peters or Begg tests.*****
>
> The message you got has nothing to do with the type of data you are
> dealing with.
>
>
>
>
>
>
>
>
>
>
>
>
>
> In practice, as long as the studies to be combined are not very small, I
> believe that both random and randomi approaches will give virtually
> identical results.
>
>  If you are using data based on logor/selogor, I guess -metan- uses
> automatically randomi (or fixedi) even if you specify fixed/random.
>
> Cheers!
>
> Tiago
>
>
>
>> Dear Rui,
>>
>> 1) Yes, the default is the fixed-effect inverse variance method if your
> data is based on logor and selogor; typing "metan logor selogor,
> randomi" will give you the DerSimonian-Laird method.
>>
>> Type
>>
>> help metan
>>
>> to see the options.
>>
>>
>> 2) As the message says, the option begg can handle only raw data (e.g.
> 2x2 tables, with a b c d counts)
>>
>> Type
>>
>> help metabias
>>
>> for further information
>>
>>
>> 3) You need to provide more information on your data set, and how the
> typename variable was coded.
>>
>> Also, try the pdfs that the following link shows :
>>
>> http://tinyurl.com/342osna
>>
>>
>> Hope this helps.
>>
>> Tiago
>>
>>
>> --------------
>>
>> Hi. I'm a new user of STATA and the meta-analysis commands and have a
> couple of questions...
>>
>> 1) For "metan logor selogor" command, is the default fixed effect? and
> if I would like to test for the DerSimonian & Laird random effects is it
> correct to type "metan logor selogor, random"?
>>
>> 2) I'm trying to do the command "metabias logor selogor, graph (begg)"
> but is giving me the following error message:
>> Note: data input format theta se_theta assumed.
>> Only Egger or Begg tests can be used with two input variables theta
> se_theta
>> r(198);
>>
>> What is causing this and how should I resolve it?
>>
>> 3) I'm trying to do the command "metafunnel logor selogor, by
>> (typename)" where typename is the type of different study designs but is
> it giving me the following error message:
>> Note: default data input format (theta, se_theta) assumed.
>> type mismatch
>> r(109);
>>
>> If i don't put in "by (typename)" it works fine and giving me the funnel
> plot. What is causing this and how should I resolve it?
>>
>> Thank you so much in advance for your guidance.
>>
>> Rui
>>
>>
>
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/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/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/