Arvey and Faley (1988) demonstrate how the US legal/regulatory environment has played a significant role in shaping the selection practices of American firms. As organisational boundaries increasingly cut across national boundaries, then the effect of regulatory environments will similarly cross these borders. Nevertheless, regulation continues to operate at the national level. For example, Rowe, Williams and Day (1994) consider that the legal pressures in North America mean that the goal of selection is as much about meeting the standards for equal opportunity and human rights as it is about providing a reliable and productive workforce. They argue, that because of the threat of litigation, employers often use less threatening devices such as application blanks and interviews rather than psychological tests. While the North American regulatory environment has thus clearly influenced test use, the opposite is the case in Italy. For example, Shackleton and Newell (1994) argue that in Italy there is no recognised code of practice guiding recruitment and selection and therefore interview questions tend to be more searching than in countries where regulation is stronger, for example in Germany.
The use of more sophisticated selection methods can clearly be expensive. For example, an assessment centre will typically involve considerable resources in terms of time of line managers and personnel representatives (and perhaps outside consultants) and money (e.g. to provide facilities for the centre and to purchase test materials). Countries where resources are more scarce, are therefore more likely to be inhibited for financial reasons from using these more sophisticated methods. In support of this, Eleftheriou and Robertson (1999) found that although the Greek firms in their sample tended to use more subjective and intuitive selection methods, they were in fact aware that other methods were available that were potentially more valid. When asked what methods they would use if there were no budgetary constraints, respondents were more likely to suggest they would use the more valid methods compared to the actual use of the less valid methods.