da deliziosa » 07/12/2011, 12:57
Qui i due paragrafi :
Statistical methods
Fisher’s exact and x 2 tests were used to compare categorical variables between groups, and t tests were used to compare continuous variables between groups. Partial F-tests in multiple linear regression models were used to compare continuous outcomes between groups controlling for age. Logistic regression models were used to estimate odds ratios for being on the street versus being in the orphanage, unadjusted or adjusted for age.16 Likelihood ratio deviance statistics were used for testing whether unadjusted or adjusted odds ratios differed significantly from one. Age was controlled for because it is a proxy for the time at risk for several of the outcome variables. All analyses were conducted using Stata version 7.0 software (Stata Corporation, College Station, Texas). All hypothesis tests were two tailed, with significance judged by a p value less than 0.05.
In logistic regression models controlling for age, problems with police (odds ratio (OR) 8.0, 95% confidence interval (CI) 2.4 to 26.6, p < 0.001), earning income by begging rather than shoe shining (OR = 8.4, 95% CI 1.4 to 48.3, p = 0.010), and an older age (OR = 1.19/year, 95% CI 0.96 to 1.47, p = 0.058) predicted paint thinner use. An older age also predicted alcohol use (OR = 1.15/year, 95% CI 1.1 to 1.6, p , 0.001), as did longer duration on the street (OR = 1.2/ year, 95% CI 1.0 to 1.3, p = 0.008), being on the street due to physical abuse compared to emotional abuse (OR = 17.9, 95% CI 2.0 to 162.9, p , 0.001), and earning income by begging rather than by singing on the bus (OR = 6.2, 95% CI 1.1 to 36.6, p = 0.031). The only significant predictor of glue adhesive use was sexual abuse (OR = 3.4, 95% CI 1.1 to 10.7, p <0.001)
.