I've sent you a copy of the paper you requested to your registered email. Let me know if you don't get this.
The article is being bounced from stats journals to psychology / medical journals. The psychology / medical journals say 'this is far too complex for a psychology audience, have you tried a statistics journal?' The stats journals say 'yes, we know this - of course. Have you tried sending it to a psychology or medical journals as it is they who are making the mistakes?'. Catch-22 supreme.
1. If you are in what Microsoft calls an International Setting (e.g., France, Germany, etc), then you may get the following error when trying to run the software: Runtime Error - 13: Type mismatch This means that you do not have the English (United States) settings in your Regional Settings. To fix this, change Regional Settings to English (United States) under My Computer / Control Panel / Regional Settings. Once finished using the programme, however, you might want to change back to your own region again.
2. Another problem may occur in countries where a comma is used instead of a decimal place. Here's what to do:
Under Windows XP / Vista / Win 7 in English slight variations around the following works: Access the 'Regional and Language Options' under the Control Panel. Then under 'Regional Options' you select 'Customise' and then change the decimal symbol to a "." instead of a ",". Accept the changes and try the programme again (some machines may need a reboot).
In German, a colleague advised us of the following: in the German version is the order 'Regional and Language Options' (in German ?Zeit, Sprache, Region?) under the Control Panel (in German ?Systemsteuerung?). Then under 'Regional Options' (in German ?Tastaturen und Eingabemethoden ?ndern?, English ?alter keyboard and input method?) you select 'customise' (in German ?Format?, English ?format?), then ?further settings? (in German ?Weitere Einstellungen?) and then change the decimal symbol to a "." instead of a ",". Accept the alteration and it works without any problems. - Thank you Dirk!
My pleasure. I've just looked at the picture I attached and it's not that clear. It should be:
correlated t-statistic multiplied by Square-root of ((2(1-r))/n).
On a side note: We will be submitting a paper ourselves soon which demonstrates that Hedges' g is a better approximation, under most circumstances, that Dunlap's equation (the one above - which is usually recommended for correlated designs).