Improving Anemia Status of Female Garment Workers in Bangladesh: An Evaluation of Pilot Intervention in Four Factories

Research Design and Methods Impact Survey Design To determine the impact of this program on the main outcomes (i.e., prevalence of anemia and dietary intake practices) we will use a quasi-experimental, controlled, pre-post design for both sets of interventions (Figure 1).

We are using a quasi-experimental design as it is not possible to identify a large pool of factories from which to randomly sample and assign factories to a control or intervention group. To minimize possible selection bias, we will:

i. To the extent possible, factories will be selected based on specified criteria, to minimize major differences in response to the intervention among them (as described in selection of factories section).

ii. Select individual women for pre-post studies based on matching or equivalence on key characteristics (eg, age); iii. Use difference-in-difference analysis to test impacts.

Intervention Description Two types of interventions will be tested against controls in a total of 4 factories. Among factories that already provide a hot lunch to workers, one intervention and one control factory will be selected, and among factories that do not provide hot lunches, one intervention and one control factory will be selected.