Measuring the quality of educational data in NWFP Pakistan

In this thesis, I demonstrate that two variables which are normally used to define data quality--validity and reliability--are insufficient for explaining problems with data quality in developing countries. I demonstrate that a third variable, rent-seeking behavior, needs to be taken into account when measuring data quality. I present evidence that field education officers in collusion with local politicians were intentionally distorting data quality in the Northwest Frontier Province of Pakistan in order to extract rents from public funds earmarked for the education sector.;I demonstrate that rent-seeking behavior cannot be taken into account from the rational system perspective, but must be understood from the perspective of educational organizations as intentionally loosely-coupled systems. I show that this view of loose coupling as an intention-filled strategy for extracting rents implies that it is not possible to overcome intentional distortions in data quality through improvements based on the rational system perspective, such as providing training, skills development, or incentives.;A review of the literature on organizational theory and educational data quality in developing countries demonstrates that research on educational data quality in developing countries suffers from a lack of solid theoretical grounding to explain the effect of rent-seeking behavior on data quality. Exploratory analysis of data quality from 1986 to 1992 in the Northwest Frontier Province of Pakistan shows that distortions in data quality were masked in the traditional manual information system, but then unmasked once the information system was computerized.;Data collected from a random sample survey on educational data quality in NWFP Pakistan shows the impact which validity, reliability, and rent-seeking behavior have on data quality in a computerized education management information system. From March to December, 1993, both quantitative and qualitative data were collected from a 5% random sample of schools as well as from field education officers in all twenty districts of NWFP Pakistan. Both quantitative and qualitative data were analyzed at the school level to show the effects of validity and reliability on data quality and at the district level to show the effect of rent-seeking behavior.