Meta-analysis of binomial data

Internet addiction in Turkey has emerged as the Internet has become widely used. In this study, it was aimed to determine the internet addiction proportion and the proportion according to geographical region and studied groups in Turkey. Meta-analysis is the synthesis of information obtained fromin dependent studies on a specific subject that reported in the sameway. Generally, fixedand random effect model are used in the synthesis of there port data in meta-analysis. It is more appropriateto choose the random effect model in syn the sizing the in formation obtained from the literatüre. The method of meta-analysis differsaccording to there ported information. For Binomaildata there are three different meta-analysis methods. The approximations to the normal distribution is used in thecase of the prevalence proportions are around 0,5, when the proportions get closerto the limits of the 0 and 1 range the logit and double arcsine transformation are used. Studies published on internet addiction in Turkey have been examined by Meta-analysis. Toobta in prevalence of internet addiction, "metaprop" function in R wasused. A meta-analysis was conducted with a total of 71 studies accordingto in clusion and exclusion criteria. 66 of the studies were carried out in 2010 and after. Random effect model was applied for normal distribution, logit and arcsin methods. The normal distribution method giving then arrowest confidence interval was chosen. The heterogeneity measuresI^2was examined, the studies were found to be highly heterogeneous (p = 0,0001, I^2=99,9%). Internet addiction rate in Turkey is 13% (95% CI: 0,11; 0,14). When the regionsare examined; A egeanand Central Anatolia regions 17%, Black Sea 14%, Marmara 13%, Mediterranean 11%, Eastern Anatolia 5% and Southeastern Anatolia 3%.When the studied group is examined; 5% formiddle school, 9% for high school, 30% for junior high schooland 13% for university.