Some issues associated with the use of moderated regression

Stone, Eugene F.; Hollenbeck, John R.

doi:10.1016/0030-5073(84)90003-5

Cited by 192 publications

(104 citation statements)

References 11 publications

Supporting

Mentioning

100

Contrasting

Order By: Relevance

“…In the moderated regression, if a significant change in variance explained was observed when the demographic moderator variable (cross-product) was added to the regression model containing the fairness predictors and the demographic variables, then the demographic variable was deemed to have a significant moderating influence on the strength of the relationship between the predictors and the criterion (Cohen and Cohen, 1975;Pedhazur, 1982). Because significant increments in variance explained by cross-product vectors are difficult to interpret as indexes of 'practical importance' (Pedhazur, 1982;Stone and Hollenbeck, 1984), subgroup analysis was also conducted. Separate full and reduced model regression analysis were conducted for subgroups identified as significant moderators.…”

Section: Discussionmentioning

confidence: 99%

Why justice matters in compensation decision making

Scarpello

Jones

1996

J. Organiz. Behav.

View full text Add to dashboard Cite

SummaryUsing agency theory and the stakeholder fairness concept as the conceptual base, this study confirmed three agency theory hypotheses about differential relationships between four sets of pay procedures and evaluations of pay, supervision and the employing organization. Education and seniority related variables were also found to moderate the relationships between procedural justice perceptions and evaluations of supervision and the employing organization. The study used a stratified random sample of 612 occupationally heterogeneous employees of a large County government in South Eastern United States. Results suggest that agency theory provides a parsimonious explanation for why justice matters in compensation decision making.

show abstract

Section: Discussionmentioning

confidence: 99%

Why justice matters in compensation decision making

Scarpello

Jones

1996

J. Organiz. Behav.

View full text Add to dashboard Cite

show abstract

“…While the square of the correlation coefficient r 2 XY indicates the percentage of Y variance explained by X variance, the corresponding regression coefficient indicates the expected change in Y for a unit change in X (r XY = b Y X only when σ Y = σ X ). While it has been suggested that testing for interaction in linear regression yields information regarding correlation interactions (Stone and Hollenbeck 1984), this has been clearly shown to be incorrect as the regression interaction tells us nothing regarding whether or not correlations differ across a moderator variable (Arnold 1982(Arnold , 1984. Indeed, Arnold (1982) contains applied examples where an interaction exists for the regression but not the correlation, and vice versa.…”

Section: Remark 6 Correlation Interactions Are Not the Same As Regrementioning

confidence: 99%

Key Results of Interaction Models with Centering

Afshartous

Preston

2011

Journal of Statistics Education

123

View full text Add to dashboard Cite

We consider the effect on estimation of simultaneous variable centering and interaction effects in linear regression. We technically define, review, and amplify many of the statistical issues for interaction models with centering in order to create a useful and compact reference for teachers, students, and applied researchers. In addition, we investigate a sequence of models that have an interaction effect and/or variable centering and derive expressions for the change in the regression coefficients between models from both an intuitive and mathematical perspective. We demonstrate how these topics may be employed to motivate discussion of other important areas, e.g., misspecification bias, multicollinearity, design of experiments, and regression surfaces. This paper presents a number of results also given elsewhere but in a form that gives a unified view of the topic. The examples cited are from the area of medical statistics.1

show abstract

“…The most noteworthy effect was that they were larger than at the incumbent level (e.g., average r = .39 vs..19 in Table 4, and .46 vs..25 in Table 5). The (Champoux & Peters, 1980;Peters & Champoux, 1979;Stone & Hollenbeck, 1984;Zedeck, 1971). In this method, the job design scale and preferences/tolerances measure were added first to the equation to predict the outcome composite, then the incremental contribution of the interaction term was tested.…”

Section: Research Questionmentioning

confidence: 99%

Interdisciplinary approaches to job design: A constructive replication with extensions.

Campion¹

1988

Journal of Applied Psychology

204

251

View full text Add to dashboard Cite

This study replicated Campion and Thayefs (1985) research, which drew from many disciplines (e.g., psychology, engineering, human factors, physiology) to demonstrate four approaches to job design and their corresponding outcomes: motivational approach with satisfaction outcomes, mechanistic approach with efficiency outcomes, biological approach with comfort outcomes, and perceptual/motor approach with reliability outcomes. This study extended the research in five ways. First, it used an expanded sample of 92 jobs and 1,024 respondents from a different industry. Second, a selfreport measure was developed and evaluated, because many jobs cannot be analyzed observationally. Third, method bias was addressed by not finding evidence of priming effects, by demonstrating strong relationships even when within-subject bias was avoided, and by relating job design to independent opinion survey data. Fourth, reliability of aggregate responses was demonstrated, and relationships at the job level of analysis were larger than at the individual level. Fifth, neither individual differences in terms of preferences/tolerances for types of work nor demographics moderated job design-outcome relationships. It was concluded that different approaches to job design influence different outcomes, each approach has costs as well as benefits, trade-offs may be needed, and both theory and practice must be interdisciplinary in perspective.

show abstract

Some issues associated with the use of moderated regression

Cited by 192 publications

References 11 publications

Why justice matters in compensation decision making

Why justice matters in compensation decision making

Key Results of Interaction Models with Centering

Interdisciplinary approaches to job design: A constructive replication with extensions.

Contact Info

Product

Resources

About