Meta-Analysis. Insights gained from meta-analysis can be useful to resolve several issues faced in combining surveys, as survey heterogeneity, planning data collection, and pooling data across surveys (Xxxxxx 1999). Xxxxx and Xxxxxxxxx (1990) define meta-analysis as “the practice of using statistical methods to combine the outcomes of a series of different experiments or investigations”. It implies four steps: identifying all relevant studies; assessing study quality; dealing with study heterogeneity; and summarizing the results. Xxxx (1998b), in the context of constructing an average birth rate for a continent using separate country birth rates, proposes three options for pooling multinational samples that are directly comparable to the three main meta-analytic models for combining study effect sizes: fixed effects (equal weight to each country’s estimate); equal effects (all subjects are independent and of equal importance); and random effects (weighed averages of the study proportions). When similar data are collected in several sites (cities, provinces or districts of one country) of a combined population, but not in all of the sites, alternative treatments of them are possible (Xxxx 1999a). In combining separate sites three decisions must be made: the allocation of sample sizes, whether the samples should be combined and what weighting to use. These are expressed as follows by Xxxx.
Meta-Analysis. Insecure attachment
Meta-Analysis. Parental care
Meta-Analysis. Parental overprotection
Meta-Analysis. Self-evaluation
Meta-Analysis. Alexithymia
Meta-Analysis. Understanding mental states
Meta-Analysis. Social dominance
Meta-Analysis. Parental care Twenty-five studies were identified that measured perceived parental care (Xxxxxxx, 2011; Xxxxx et al., 2003; Xxxxx, Xxxxxxxx, Xxxx, & Xxxxxxx, 2000; Xxxxx, Xxxxxx, Xxxxx, & Xxxxxx, 1990; Xxxxxxx, Xxxxxx, Lerer, Xxxxxx, & Xxxxxx, 2008; Xxxxx et al., 2012; Xx Xxxxxxxx, Xxxxxxxxx, Xxxxx, Xxxx, & Xxxxxxx, 2003; Xxxx, Xxxxx, Xxxxxx, Yellowlees, & Xxxxxxxx, 2011; Xx Xxxxxxx et al., 1998; Xxxxxxx, Xxxxxxx, Xxxxx, & Xxxx, 2010; Xxxxxxxx et al., 2011; Xxxxx, 1984; Xxxxxxx & Laporte, 2002; Xxxx & Xxxxxxx, 1992; Laporte & Xxxxxxx, 2007; Xxxxx, Xxxxxx, & Xxxxxx, 2000; Xxxxxx, Xxxxxxxxxxx, & Xxxxxxxx, 1988; Xxxxxx & Kroger, 1992; Xxxxxxx, Xxxxx-Xxxxxxxx, Xxx, & Xxxxxxx, 1992; Xxxxxxx, Xxx xxx Xxxx, Xxxxxxxxx, & Xxxxxxxx, 1989; Xxxxxxxx-Xxxxx et al., 2005; Xxxxxxx et al., 2010; Xxxxxxxxxxxx & Perris, 1994; Xxxx, Xxxxxxxxx, & Xxxxxx, 2007; Xxxxxxxxx et al., 2000). Each study measured parental care using the Parental Bonding Instrument (PBI). The studies were included in the meta-analysis shown in the forest plot in Figure 2. People with ED reported lower levels of perceived parental care with a moderate ES 0.53 (95%CI: 0.41, 0.65, p < .001) relative to HC. This overall effect remained moderate .51 (p < .001) after adjusting using the meta-regression and there was no evidence of publication bias (Begg’s test p=.080; Xxxxx’x test p=.413).
Meta-Analysis. Parental overprotection The same twenty-six studies as in the prior analysis were included in the meta-analysis of perceived parental overprotection shown in Figure 3 (Xxxxxxx, 2011; Xxxxx et al., 2003; Xxxxx et al., 2000; Xxxxx et al., 1990; Xxxxxxx et al., 2008; Xxxxx et al., 2012; Xx Xxxxxxxx et al., 2003; Xxxx et al., 2011; Xx Xxxxxxx et al., 1998; Xxxxxxx et al., 2010; Xxxxxxxx et al., 2011; Xxxxx, 1984; Xxxxxxx & Laporte, 2002; Xxxx & Xxxxxxx, 1992; Laporte & Xxxxxxx, 2007; Xxxxx et al., 2000; Xxxxxx et al., 1988; Xxxxxx & Kroger, 1992; Xxxxxxx et al., 1992; Xxxxxxx et al., 1989; Xxxxxxxx-Xxxxx et al., 2005; Xxxxxxx et al., 2010; Xxxxxxxxxxxx & Perris, 1994; Xxxx et al., 2007; Xxxxxxxxx et al., 2000). People with ED reported higher levels of overprotection with a small ES 0.33 (95%CI: 0.21, 0.45, p < .001) relative to HC. This overall effect remained small .29 (p = .022) following adjustments made to the overall effect by the meta-regression. We found evidence of publication bias (Xxxx’x test p=.025; Xxxxx’x test p=.036). The trim and fill method estimated that twenty studies were missing from the analysis. The adjusted ES remained small .14 (95%CI: 0.01, 0.27, p = .033) after correcting for publication bias using this method.
