Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# Re: st: Metan for random effects

 From Aggie Chidlow To statalist@hsphsun2.harvard.edu Subject Re: st: Metan for random effects Date Sat, 10 Apr 2010 20:20:27 +0100

```Dear Henry,
Thank you for the link it is much appreciated.
The information in there ia actually similar to the book I have been
reading titled: Meta-Analysis in Stata, Edited by Sterne in 2009.

I think I understand the mechanics of the metan command. However, I am
having a problem to us is for my or work. So, if you could suggest
anything I would be very thankful.

In my current research I am investigating "if"  and "how" researchers
in 4 top journals reported (or not) a particular framework in mail
survey studies between 2000 - 2009. The framework consists of 5 steps.
I coded the steps individually  as a dummy variable e.g. Step One (0=
Reported; 1=Not Reported) and I also codded them together (Tables 1
and 2 respectively  below). Out of 279 studies  with mail survey as a
data collection procedure, 113 report the framework and 166 don't
report it at all.

Table1
First Step     |  Name of the journal the article appeared
|        J1         J2        J3           J4
|     Total
---------------+--------------------------------------------+----------
Not Reported    |        25         21         17         16 |        79
Reported       |         6          8           11           8 |        33
---------------+--------------------------------------------+----------
Total |        31         29         28         24 |       112

Table 2
|  Name of the journal the article appeared
All Steps together     |        J1         J2          J3
J4      |     Total
----------------------+--------------------------------------------+----------
Not Mention         |        68         41         32
25      |       166
One Step mention         |        17         14         17         12
|        60
Two Steps mention       |         8          9          8
5      |        30
Three Steps mention     |         4          5          2
6      |        17
Four Steps mention       |         1          2          1
1      |         5
Five Steps mention        |         1          0          0
0      |         1
----------------------+--------------------------------------------+----------
Total |        99         71         60
49       |       279

I know that in order to proceed with "metan" I have to calculate the
effect size.
I also know that there are different effect sizes for continous and
binary data (as stated in the link you suggested).  But HOW do I
calculate the effect size (i.e. Odd ratio or risk ratios) for a dummy
variable in Table 1, for example? That's where I get confused?

On Wed, Apr 7, 2010 at 2:42 PM, Henry <jakanyada@gmail.com> wrote:
> Hi Aggie,
> Could this link be of help to you?
> http://www.blackwellpublishing.com/medicine/bmj/systreviews/pdfs/chapter18.pdf
> HTH,
> Henry
>
> On Wed, Apr 7, 2010 at 1:41 PM, Aggie Chidlow
>> Hi All,
>>
>>
>> Can somebody advise me how to calculate odd ratios or risk ratios for
>> *
>> *   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/
>

*
*   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/
```