Evaluation of the methodology for analyzing the patterns and determinants of breast feeding and mortality in the Near East

This report evaluates data analysis methodologies used in the pilot project. Omissions in breastfeeding data make analysis of breastfeeding trends in Jordan impossible. To provide overall information on breastfeeding practices, however, the use of the current status method of estimating mean duration of breastfeeding in categories comparable to those used in the infant mortality analysis is strongly recommended, as are checks on the multivariate analysis used to control sample selection bias and repetition of the analysis in categories comparable to the infant mortality analysis for the individual variables. Data on pill use, maternal labor force participation, and education should be interpreted with caution. Misreporting duration of breastfeeding has been alleviated by separately analyzing dichotomous dependent variables employing an X function. A standard statistical estimation procedure has been employed to define dichotomous variables, but the analysis needs to be amplified and its results presented in more comprehensible language. The prototype analysis of infant mortality is on the right track, but could be improved in several ways: (1) by finding ways to avoid age misstatement in neonatal and postnatal deaths; (2) by comparing mortality rates obtained in this study with those available from other sources; (3) by extending the analysis to include child and infant mortality; (4) by modifying the multivariate analysis to assess the total effect of the variables; and (5) by converting multivariate analysis results to a more comprehensible format. To improve the project overall, the authors recommend: (1) increasing coordination between the two components by using either weighted or unweighted samples for the multivariate analysis, using comparable socioeconomic and demographic variables, and sharing common data problems; (2) noting cross-country differences in defining variables; (3) introducing a variable representing a region of a country in both project components; (4) making results more comprehensible to non-experts; and (5) submitting a combined final report for comparison purposes and as a policy aid. Appended is a list of contacts.