THE RESTRICTIONS OF EXISTING MOTIVATION MODELS AND NET-EFFECTS THINKING

The research by Lawrence and Nohria (2002) and Nohria et al. (2008) contribute greatly to developing a comprehensive motivation theory that incorporates many key research fields. However, we suggest that the theory concerning employee motivation and organization performance may be advanced by adopting an alternative research method approach over the conventional quantitative analysis on which current theories are based. 

As it stands, researchers view their primary task as one of assessing the relative importance of causal variables drawn from the various employee motivation theories. In the perfect situation, the relevant theories emphasize different motivation variables and make clear and unequivocal statements about how these variables are connected to relevant outcomes. However, in practice, motivation theories are imprecise when it comes to specifying both causal conditions and outcomes, and they tend to be even more vague when it comes to stating how the causal conditions are connected to outcomes (i.e., what are the conditions that must be met for motivated employees and organizational performance? What are the justifications for the conditions chosen?). Therefore, researchers usually develop only general lists of potential relevant causal conditions, better known as contingent moderating and/or mediating variables, based on the broad definitions of what they find in competing motivation theories (Ragin, 2008). The main analytic task is typically viewed as one of assessing the relative importance of the relevant moderators and mediators. If the moderators/mediators associated with a particular motivation theory prove to be the best predictors of the outcomes (i.e., variables that provide the highest percentage of variance explained), then this model is used for informing existing theory

or developing new theory. These models are symmetric by design and the correlation coefficient is the measure for developing conclusions based on general patterns of association (Ragin, 2008). This method of conducting quantitative analysis is the default procedure in the social sciences today, one that researchers fall back on, often for lack of knowledge of a clear alternative. 

Specifically, in conventional quantitative research (e.g., multiple regression analysis, structural equation modeling), independent variables are seen as analytically separable causes of the outcome under investigation (Woodside, 2013). Typically, each causal variable is thought to have an independent capacity to influence level, intensity or probability of the dependent variable. These methods assume that the effects of the independent variables are both linear and additive, meaning that the impact of a given independent variable on the dependent variable is assumed to be the same regardless of the values of other independent variables (Woodside, 2013). This is known as net-effects estimation, which assumes that the impact of a given independent variable is the same not only across other independent variables but also across all their different combinations (Ragin, 2008). To estimate the net effect of a given independent variable, the researcher offsets the impact of rival causal conditions by subtracting from the estimate the effect of each variable any explained variation in the dependent variable it shares with other causal variables (Ragin, 2008; Woodside, 2013).

When confronted with arguments that cite combined conditions (e.g., that a recipe of some sort must be satisfied), the usual recommendation is that researchers model combinations of conditions as interaction effects and test for the significance of the incremental contribution of “statistical interaction” to explain variation in the dependent variable (e.g., Baker & Cullen, 1993; Drazin & Van de Ven, 1985; Miller, 1988). When there is interaction, the size of the effect of an independent variable on a dependent variable depends upon the values of one or more other independent variables. However, as explained in Ragin (1987, 2008), estimation techniques designed for linear-additive models often come up short when assigned the task of estimating complex interaction effects. Large samples are required, and there are still many controversies and difficulties surrounding the use of any variable in multiplicative interaction models (Ragin, 2008). 

When used exclusively, Ragin (1987, 2000, 2008) points out the problem of symmetric models and net-effects thinking. First, the evaluation of net effects is dependent on model specification and can be swayed by the correlation among the moderating/mediating variables. Limiting the number of correlated variables and a chosen variable may have a substantial net effect on the outcome, but this variable may not have a net effect in the presence of other correlated variables (see also Woodside, 2013). Second, and most importantly, the estimation of net effects is highly dependent on the correct specification of the research model, and this is dependent upon strong theory and deep knowledge, which are often lacking in the application of net-effect methods. Therefore, how meaningful is a specified research model that does not have strong theory? And, how much credibility is there in the conclusions that are derived from the specified research model? 

While powerful and rigorous, the net-effects approach is limited. Consequently, it is reasonable to consider an alternative approach, one with strengths that complement those of symmetric models and net-effects methods. In addition to assessing net effects, researchers could examine how different causal conditions among employee motivations combine to explain organizational performance. Specifically, the net effects approach, with its heavy emphasis on calculating the effect of each independent variable in order to isolate its independent impact, can be counterbalanced and complemented with an approach using set theory that explicitly considers the combinations and configurations of various conditions.

References 

• Baker, D. D., & Cullen, J. B. (1993). Administrative reorganization and configurational context: The contingent effects of age, size, and change in size. Academy of Management Journal, 36(6), 1251–1277.
• Drazin, R., & Van de Ven, A. H. (1985). Alternative forms of fit in contingency theory. Administrative Science Quarterly, 30, 514–539.
• Lawrence, P. R., & Nohria, N. (2002). Driven: How human nature shapes our choices. San Francisco: Jossey-Bass.
• Miller, D. (1988). Relating Porter’s business strategies to environment and structure: Analysis and performance implications. Academy of Management Journal, 31(2), 280–308. 
• Michael, T. L., Robyn, L. R.(2016). Understanding employee motivation and organizational performance: Arguments for a set-theoretic approach. Journal of Innovation & Knowledge.
• Nohria, N., Groysberg, B., & Lee, L. (2008). Employee motivation: A powerful new model. Harvard Business Review, 86(7/8), 78–83. 
• Ragin, C. C. (2008). Redesigning social inquiry: Fuzzy sets and beyond. Chicago: University of Chicago Press.
• Ragin, C. (1987). The comparative method: Moving beyond qualitative and quantitative methods. Berkeley: University of California.
• Ragin, C. C. (2000). Fuzzy-set social science. University of Chicago Press.  
• Woodside, A. G. (2013). Moving beyond multiple regression analysis to algorithms: Calling for adoption of a paradigm shift from symmetric to asymmetric thinking in data analysis and crafting theory. Journal of Business Research, 66(4), 463–472.
• Woodside, A. G. (2013). Moving beyond multiple regression analysis to algorithms: Calling for adoption of a paradigm shift from symmetric to asymmetric thinking in data analysis and crafting theory. Journal of Business Research, 66(4), 463–472.