Minimising total survey error when measuring entrepreneurship in Tonga

This study illustrates how the Total Survey Error (TSE) paradigm can identify and help reduce multiple sources of error inherent in survey work in the developing world. Of particular concern are mode errors and coverage errors caused by the 'theoretical teledensity threshold' of doing phone surveys in developing countries. The study outlines ways to improve response rate and to avoid interviewer and measurement error. It narrates the sampling design and its limitations as well as some of the qualitative aspects of total survey quality such as, translation, ethics and budgeting. The final section discusses implications for further research in statistical auto-correlation and data gathering using PDAs.