Infant mortality levels in Africa: does method of estimation matter?

"Using data from 13 Demographic and Health Surveys in Africa, this paper compares estimates of infant mortality directly calculated from data with those indirectly calculated using the Trussell (1975) and Feeney (1980) techniques. The results indicate that, on average, the directly calculated infant mortality rates were significantly lower than those estimated from indirect methods. Data errors alone can not plausibly explain such differences. Such series could in fact show increases or decreases in infant mortality rates that would result mainly from differences in methods of calculation." (SUMMARY IN ITA AND FRE)