Is the "Glasgow effect" of cigarette smoking explained by socio-economic status?: A multilevel analysis