자기효능감과 성과의 관계에 대한 재연구

• 영문명: Self-Efficacy and Performance Revisited - Controlling for Individual Differences
• 발행기관: 한국인사조직학회
• 저자명: 이동섭(Dongseop Lee)/김기태(Kitae Kim)/조봉순(Bongsoon Cho)
• 간행물 정보: 『인사조직연구』인사·조직연구 제16권 제4호, 75~107쪽, 전체 33쪽
• 주제분류: 경제경영 > 경영학
• 파일형태: PDF
• 발행일자: 2008.12.31

국문 초록

본 연구는 과업성과의 예측변수로서의 자기효능감과 개인성격 변수들의 상대적 효과를 검증한다. 사회인지이론(social cognitive theory)은 성과를 비롯한 다양한 행위 변수를 설명하는 요인으로서 자기효능감의 중요성을 강조하였으며, 지난 30년간 많은 연구의 결과가 이를 지지하였다. 하지만 최근 Judge, Jackson, Shaw, Scott, & Rich(2007)는 메타분석결과에 의거한 경로분석을 통해, 개인성격 변수가 포함될 경우 자기효능감의 효과는 사라진다는 주장을 전개하였다. 본 논문은 Judge et al.의 연구에 내재된 방법론적인 한계를 감안할 때, 자기효능감의 과잉변수화 주장에 대해서는 보다 신중한 해석이 필요하다는 점을 지적하고, 그 대안으로 자기효능감과 Big 5 성격 변수를 동시에 포함하는 개별연구(primary study)를 통해 고유한 자기효능감 효과의 존재 유무를 살펴보았다. 156명의 대학생을 표본으로 하여 2회에 걸쳐 실시된 본 연구의 자료 분석결과에 의하면, Judge et al.(2007)에서와는 달리, 성격 변수의 효과는 통계적으로 유의하지 않은 반면, 자기효능감은 과업성과에 대해 일관적으로 유의한 예측력을 보였다.

영문 초록

The purpose of the study is to empirically examine the relative impacts of Big 5 personality factors and self efficacy on task performance, controlling for mental ability and experience. Self-efficacy, the central variable of human regulatory mechanisms in social cognitive theory, has been proposed to be one of the most powerful cognitive determinants of human behavior(Bandura, 1986, 1997, 2006a; Locke & Latham, 1990). Research for over two decades has shown that self-efficacy has an impact on a variety of motivational and behavioral outcomes in clinical, athletic sport, educational, and various organizational settings (Holden, 1991; Holden, Moncher, Schinke, & Barker, 1990; Multon, Brown, & Lent, 1991; Stajkovic & Luthans, 1998). However, Judge, Jackson, Shaw, Scott, and Rich (2007) recently called the validity of these findings into question. They suggested that the unique contribution of self-efficacy in predicting performance may become negligible if individual traits, which have been largely neglected in social cognitive theory, are considered simultaneously. Using meta-analytic path analysis on the relationships among Big 5 personality traits, mental ability, experience, self-efficacy, and task performance, they reported a substantially reduced incremental predictive validity of self-efficacy when individual difference variables were controlled for. On the basis of this finding, they concluded that given the significant relationships of personality and intelligence to self-efficacy, task-specific self-efficacy may simply be a superfluous byproduct of a smart and generally confident person (Judge et al., 2007). We argue that the conclusion made by Judge et al.(2007) is premature and potentially erroneous. In testing the incremental predictive validity of self-efficacy on performance with individual difference variables controlled for, Judge et al. combined psychometric meta-analysis and path analysis, a procedure we found has critical limitations. First, the biggest problem in the procedure Judge et al. used concerns that relative predictive power between Big 5 personality traits and self-efficacy, which is the most important aspect of their model, had not been examined by any of the individual studies included in their meta-analytic synthesis. In other words, the “relative” part of data is entirely missing, and this makes their conclusion about relative predictive power in fact based on missing data. Missing data is considered one of the most critical problems in this procedure because estimation of certain relationships (e.g., path coefficients) in the model may become difficult or distorted due to missing data (Becker & Schram, 1994; Viswesvaran & Ones, 1995). Second, due to the systematic missing data in Judge et al.(2007), it may be possible that different parts of the model (e.g., the relationships between Big 5 traits and performance, between self-efficacy and performance, and between Big 5 traits and self-efficacy) are based on substantially differing populations. This variation in meta-analytic correlations is a critical issue, for the test of a path model can be affected even when only one of the variables included in the path model is moderated by a certain other variable (Viswesvaran & Ones, 1995). A series of moderator analyses conducted by Judge et al. themselves showed that the relationship between self-efficacy and performance was stronger, for example, when self-efficacy was measured with grid type measures rather than with Likert type measures, when performance domain was relatively specifically identifiable task performance rather than overall job performance, and when performance was assessed objectively rather than subjectively. We view these results as indicating numerous sources of systematic variation in their path model. Given the scarcity of research properly evaluating the relative predictive validity of self-efficacy, more primary studies including both self-efficacy and various

키워드

자기효능감 Big 5 성격 변수 과업성과 Self-Efficacy Big 5 Personality Traits Task Performance