The best measures of attachment in children tend to be behavioral (e.g., the Strange Room) but other researchers use parental ratings for classification (e.g., Gunnar) or a variant of the Attachment Q Sort (see van IJzendoorn, et al., 2004 in Child Development). Given your concurrent interest in parenting styles, using parent ratings is likely to artificially inflate any associations and leave your data vulnerable to bias but the Q Sort may be less problematic. Good luck.