Background and Purpose: The purpose of this article is to test the requirement of “local independence” in the newly developed “Maternal Health Literacy” (MaHeLi) composite scale measuring health literacy in pregnant adolescents attending antenatal care. Methods: The 20-item scale was administered to 384 adolescents aged 15–19 years attending antenatal care in Uganda during the period July–December 2013. Rasch analysis was conducted using RUMM2030. Results: Differential item functioning was observed for 3 items. One item had disordered response categories. One subscale brought substantial multidimensionality into the MaHeLi scale. Conclusions: Results support the use of a 12-item version of the MaHeLi scale. The article shows how Rasch analyses help us to identify violations of local independence in scales.