High employee turnover rates constitute a major challenge to effective aid provision. This study examines how features of humanitarian work and aid workers' individual characteristics affect retention within one humanitarian organisation, Médecins Sans Frontières (MSF) Holland. The study extends existing research by providing new theoretical explanations of employment opportunities and constraints and by engaging in the first large-scale quantitative analysis of aid worker retention. Using a database of field staff (N=1,955), a logistic regression is performed of the likelihood of reenlistment after a first mission. The findings demonstrate that only 40 per cent of employees reenlist for a second mission with MSF Holland, and that workplace location and security situation, age, and gender have no significant effect. Individuals are less likely to reenlist if they returned early from the first mission for a personal reason, are in a relationship, are medical doctors, or if they come from highly developed countries. The paper reflects on the findings in the light of policy.

No abstract

EDITOR'S SUMMARY A novel approach to evaluating the impact of nonprofit organizations is proposed, combining social network and linguistic analysis. The authors examined data from nonprofit organizations' websites and site hyperlinks to other organizations. They identified 369 sites of organizations that measure social impact and captured inbound and outbound weblinks to construct a relational structure. Keywords from the sites were categorized as reflecting scientific, civil society or managerial domains and located the entity in a triangular semantic space. While most organizations fell into one of the three communities, some were in an interstitial space spanning these domains. The interstitial organizations used a mix of terminology, were densely interconnected and connected extensively to organizations across domains, but there were few direct connections among the three domains. The resulting map integrates cultural and relational dimensions and reveals hidden patterns and clusters. The approach can be used with other social systems combining rich text with relational data.

No abstract

No abstract