The above average effect in an end-user computing context

Abstract

This thesis investigates how the above-average effect presents in the ubiquitous, fast-changing domain of end-user computing (EUC). EUC is mandatory in many workplaces but can be performed with different levels of skill. The above-average effect has been the subject of many studies in many different domains, as has end-user computing. This study brings these two areas together using an original approach to understand how processes, such as the above-average effect, interact with personal factors to influence perceptions of EUC skill level in self and others. The Above-Average Effect is a social bias found in many domains considered routine, or vaguely defined. This bias involves making an unwarranted, positive assessment of the difference between one’s abilities and knowledge and those of an ‘average person’. Explanations for this effect include self-enhancement, focalism, ego-centrism and the Dunning-Kruger Effect. The focus of this study was on the relationship of personal factors, such as age, sex, expertise and personality, with the occurrence of the Above-Average Effect in the context of end-user computing. This context has several characteristics that should make it an ideal setting for the occurrence and investigation of the Above-Average Effect. First, there are few opportunities for end-users to observe directly the ability of others, which contrasts with settings such as driving in which near-continuous observation of others’ skill level is possible. Second, end-user computing has undergone continuous change that many users may not notice if they perform routine tasks. Third, end-user computing roles and jobs cover a range of skill levels, the full extent of which may not be clear to many users. Measures of personal factors, demonstrated skill, self-perceptions of end-user computing skill and perceptions of the average end-user’s skill were taken from a sample of employed computer end-users. Both objective and subjective measures were used to compare self-reports with demonstrated skill and to test eight hypotheses addressing the Above-Average Effect and the Dunning-Kruger Effect. A results based testing system was developed and validated specifically for assessing end-user skill typical of workplace computing. Measurement of perceptions was undertaken using a visual analogue scale. Findings confirmed expectations that the Above-Average Effect is present in the end-user computing domain. In this domain, users often are unaware of the extent of the domain, their own skill level within it or the skill level of other end-users. Unexpectedly, however, it was found that variables that previous studies had found to be associated with the Above-Average Effect in this study were not significantly associated with the Above-Average Effect when analysed in combination. This suggests the presence of previously unidentified interactions between these variables that lessen the strength of the Above-Average Effect, specific to the domain of end-user computing. Evidence to support the operation of the Dunning-Kruger Effect as an explanation for the occurrences of the Above-Average Effect was mixed. Findings revealed a significant relationship between self-assessment and estimations of the breadth of the domain. However, there was no support for an association between a person’s estimation of the breadth of the domain and the above-average effect. Likewise, there was no support for an association between a combination of personal and expertise factors and the Dunning-Kruger Effect. As for the Above-Average Effect, this raises questions as to the types of interactions that lead to reduced evidence of the Dunning-Kruger Effect. It was concluded that (1) the Above-Average Effect is present in end-user computing; (2) interactions between variables individually associated with the Above-Average Effect may moderate the effect; (3) interactions between variables individually associated with the Dunning-Kruger Effect may moderate the effect, and (4) the domain of EUC has differences that make it stand out from other domains the AAE typically occurs in. Possible explanations for and implications of these findings for theoretical development of the Above-Average Effect and Dunning-Kruger Effect are considered, especially in domains that are commonplace, constantly changing, and that incorporate a wide range of levels of expertise. Implications for skill development and training in end-user computing are also discussed. Based on the findings, further work is recommended to explore the Above-Average Effect and its relationship to other variables, especially in ubiquitous, fast changing domains such as end-user computing. End-user computing is a vast and fast changing domain that, due to its wide use, is often misunderstood in terms of complexity and range of use. This study contributes to understanding the AAE in an area not otherwise investigated for this bias. This bias leads to overestimation of skill and knowledge which can present potential problems for accuracy and efficiency of use. This is significant because these skills are critical to modern workplaces. A further contribution extends to the instrumentation developed. This study has proved the worth of such instruments for use in social settings and shows the VAS provides a more accurate measure of perceptions than do discrete scales.