Abstract

Relationship between human resources for health and mortality remains inconclusive despite numerous studies published on the topic in the last decades. This paper investigates how and why methodological trade-offs implicitly made by researchers when using macro-data can in part explains this puzzling lack of agreement. Using data from the Global Health Observatory, we build a model of the relationship between human resources and mortality, which we progressively alter by changing its scope, variables and analysis period. Then, we compare results among themselves to isolate the impact of methodological choices from other changes in the data. Results demonstrate how methodological choices linked to (1) the analysis period, (2) the definition of health inputs, health outcomes and control variables and (3) the choice of specific variables as proxy for human resources and health outcomes affects the relationship between human resources and health outputs. Results presented highlights the need for complementing existing macro-analysis with other analytical strategies, for better documenting methodological choices in research studies, as well as for further supporting countries' efforts to produce reliable and consistent data.