Predicting examiner recommendations on Ph.D. theses

This paper investigates relationships between candidate and examiner characteristics, the texts of examiner reports on Ph.D. theses and examiner recommendations made on theses. The data were related to 804 examiner reports on 301 theses submitted at three Australian universities. Thesis topics ranged across ten Broad Fields of Study or discipline areas. Simple bivariate analyses were first undertaken to identify candidate, examiner and university variables significantly related to the examiner recommendation—three candidate variables, examiner location and, in some cases, the university attended were identified. Next, these variables were regressed on examiner recommendation. The text categories were then factor analysed to confirm five constructs identified in previous work—positive summation, negative summation, prescription, formative evaluation and dialogic elements. These were added to the regression equation. A multilevel regression analysis with examiner recommendation as response variable indicated that four of the five constructs (not including the dialogic elements construct), holding a scholarship and two examiner-country variables explained a total of 43% of the variance in examiner recommendation on the thesis. Implications of the results are discussed.