Mental and psychomotor development: At 18 months of age, the children's development was assessed with the Revised Version of Bayley Scales of Infant Development (BSID-II) (19 ) using its Mental and Psychomotor Development Indices (MDI and PDI). The children were tested in the presence of their mothers at one of the four local health centres. The Bayley Scales have not been standardized for Bangladeshi children but have been used by the same research group in several previous studies in rural (20 (
link)) and urban (9 (
link), 21 (
link)–22 (
link)) Bangladeshi children. The children's scores were in the normal range and correlated with parental education, socioeconomic status, and HOME scores in a theoretically-sensible way. Five psychologists were trained to test the children, and before beginning the study, each of them performed 10 tests on non-study children of the similar age range and was observed by a trainer. The intraclass correlations between the trainer and each psychologist ranged from r=0.88 to 0.99 (n=10) for both MDI and PDI.
Language: The children's comprehensive and expressive language development was assessed at 18 months of age using an inventory, specially developed for Bangladesh, based on the principles of the MacArthur Communicative Development Inventory: words and gestures (23 (
link)–24 ). The inventory depends on mothers’ report of their children's ability to comprehend and express words, arranged in categories (e.g. animals, body-parts, and food). There is a short version of 89 words that contains only nouns, verbs, and sounds but no gestures (25 (
link)). The Bangladeshi inventory contained 60 words arranged in the same categories in order of difficulty (Hamadani JD
et al. Personal communication, 2010). The inventory was developed after extensive piloting with mothers of young children and in consultation with Larry Fenson (Personal communication, 2003) and was then given to mothers in their homes. The test-retest reliabilities after 7–14 days in 15 mothers of children aged 18 months for comprehension and expression were (intraclass correlation) r=0.67 and 0.99 respectively.
Family care indicators: The FCI questionnaire was developed by groups of experts organized by the UNICEF with preliminary piloting for comprehension in several countries (26 ). The items were grouped into the following theoretical subgroups: ‘Varieties of play materials’ (including picture books for young children) (7 items), which classified toys by their use; ‘Sources of play materials’ (4 items), which identified where the play materials came from; and ‘Play activities’ (6 items), which identified specific types of activities done by any adult in the home with the child in the previous three days. All these items were scored: yes=1 and no=0 (presence or absence of play material or activity). Two other items—‘Household books’, i.e. the number of books in the home, excluding picture books for young children (1 item) and ‘Magazines’, i.e. the number of magazines and newspapers in the home (1 item)—were initially intended to make one subscale; however, they behaved differently in the analyses, and we decided to keep them separate.
The FCI inventory was given in the children's homes by one of four research assistants. The interviewer asked to see items concerning play materials and reading materials whereas responses to the remaining items depended on mothers’ report. Before beginning the study, each interviewer conducted five interviews and observed and scored 15 more in the presence of the trainer, and intraclass correlation for each interviewer was 0.99.
