To questions about the who, what and why of item omission on tests in short-response format, very few answers have been provided from a combination of the discipline areas of psychology and educational measurement. In this research study, an empirical approach is taken subsequent to the proposal of a theoretical model. It posits that the three clusters constituting 'presage' have effects, some direct, some indirect, some positive, some negative, on the 'product' - short-response omit rate - and also influence the hidden (and therefore unable-to-be-measured) 'process' - the interaction between item and person. Data were obtained on the 1997 QCS Test population (N = 29 273). A 120-item questionnaire was administered to a random sample of 1908 students, interviews were conducted with selected students, and information was extracted from the test construction matrix. The results are analysed in terms of background and psychological characteristics of the candidate and features of the testing process. It is concluded that the predeterminants of the propensity to omit short-response items include sex of candidate, type of school attended, test-irrelevant thinking, academic self-concept, test-taking strategies, and self-imposed difficulty. One of the subsidiary findings is concerned with attitudes to high-stakes testing, another with the consequences of the contextualisation of test items for a certain type of student.