The validity of biodata as a selection tool within the Scottish accountancy profession
The aim of this thesis was to critically evaluate whether biodata could be used as a valid tool in the pre-selection process of trainee chartered accountants. Biographical details of recently qualified accountants who trained within the whole spectrum of ICAS training offices, were collected from a self-completion questionnaire. The data collected was used to develop statistical models predicting two relevant pre-selection criteria for the accounting profession, namely the ability to pass the professional examinations at the first attempt and the probability of remaining with the training provider for at least 18 months post qualification. The validity of the developed models for use within the Scottish accountancy profession was then evaluated. Phase I of the thesis developed two models which considered examination performance, one based on honours graduates (n--183) and the other based on ordinary graduates (n=366) who qualified in the years 1988-1992. An empirical approach was adopted. Whilst statistically significant explanatory powers were identified for each model, a disappointing shrinkage was experienced when the models were applied to trainees who qualified in the years 1993-94. Two possible explanations were offered, the instability of the reference group and the empirical nature of the model derivation. Phase 11 of the thesis therefore adopted a more rational approach through the formulation of a conceptual framework. General background areas were hypothesised to be relevant in the determination of the criteria, and within these general areas specific factors were highlighted and entered into the logistic regression models using data from trainees who qualified in the years 1993-4. Statistically significant models were developed for both fully-accredited honours (n---149) and ordinary graduates (n--225) and these models continued to retain their validity when applied to trainees who qualified in 1995. The model developed for non-relevant honours students (n--61) was not significant highlighting the need for further work in this area. Likewise, from the data collected, an effective tenure model was not developed. This thesis has provided an additional validity study for the technique of biodata. by identifying that rationally derived biodata models can be used to differentiate between fully-accredited trainees who pass their WAS examinations at the first attempt and those who experience failure. This work therefore provides training principals who employ fully-accredited graduates to undertake WAS training with a useful preselection tool. It also provides evidence that rationally derived models appear to retain their validity and experience less shrinkage than models which were developed using a more empirical approach. Finally, the work demonstrates that it is possible to develop biodata models which are transportable across the full spectrum of ICAS authorised training offices and that biodata can therefore be transportable from one organisation to another.