104
career-105 related content. The outcome variable "best aspect" needed to be categorized in order to be fit for statistical analysis. Consequently, we established a set of rules for assigning observations consistently. We are confident that others would come to the same distribution of observations if they applied the same rules for categorization. The outcome variable consists of six mutually exclusive categories, each representing a component, or aspect, of a compensation and reward scheme. In addition to the outcome variable, we obtained a number of independent variables in order to assess and control for various adjusting effects. The three independent control variables are (1) level, which controls for differences in seniority; (2) region, which adjusts for the effect that respondents were based in different parts of Europe; and (3) company size, which controls for differences in the size of the respondent’s employer. Most importantly, Vault.com provided an independent continuous variable (stars) that measure each respondent’s level of compensation satisfaction, on a scale from 1 to 5. Thus, the star variable was crucial to our investigation into the possible patterns of association between compensation satisfaction and a respondent's propensity to highlight the different categories that make up the outcome variable
We built a predictive model based on a seven-step model-building approach, called purposeful selection, by Hosmer et al. (2013). This approach ensures not just the selection of significant independent variables but also the retention of confounding variables. The model-building approach tests whether assumptions are met, in order to ensure the robustness of the results. Ultimately, the predictive model was validated through the use of two strong internal validation methods, namely (1) 100 repeats of 10-fold cross-validation and (2) 1000 bootstrap resamples.
Based on theoretical insights from agency theory, social exchange theory, theories of distributive and procedural justice and self-determination theory, we developed 8 hypotheses that together would produce the insights necessary to provide an answer to the problem statement. The univariable analyses of association were then utilized to test our hypotheses.
Results
The final predictive model is able to make predictions about the likelihood that a management consultant will prefer a given compensation and reward component to the other components, given her individual characteristics. The validation tests show that the predictive model has an accuracy (ACC) of 36.74% (See table 12), which is significantly greater than the no information rate (NIR) of 27.39% (P-value that ACC > NIR: 7.80E-06***). We argue that the model provides a foundation for achieving a better understanding of the nature of association between compensation satisfaction and the different compensation and reward scheme components.
The conclusion to our problem statement is that management consultants, at the two extreme ends of the compensation satisfaction scale (1 and 5), are more likely to highlight the safety and work-life
106
balance and base salary components, respectively, than any of the alternative components, as the best aspect of their individual compensation and reward scheme. Safety and work-life balance thus decreases in importance as compensation satisfaction increases, while base salaries increases in prominence.
The first sub-question, which is concerned with which components that are most likely to be associated with high levels of compensation satisfaction, was addressed by hypotheses 1 through 6.
The second, which deals with the characteristics of a consultant and its impact on his or her probability of choosing a component, was addressed by hypotheses 7 and 8. Our findings lend support to 5 of our 8 hypotheses (H1, H4, H5, H7 and H8).
For the first sub-question, our findings show that base salary is the most likely component to be associated with high levels of compensation satisfaction. For the second sub-question, we conclude that the three independent variables company size, seniority and region provide some adjustments to the effect of compensation satisfaction. For instance, consultants employed at small consultancies are significantly more likely to highlight the incentive component relative to consultants employed at medium or large consultancies. Further, we observed that consultants with high seniority (i.e., experienced consultants) are more likely than their entry-level colleagues to highlight the incentive component as the best aspect of their individual compensation and reward scheme. Finally, the prominence of the allowance and merit pay components vary across the regions, with the allowance component being more prominent in regions 1 and 2, while the merit pay component being more prominent in regions 3 and 4.
The third and final sub-question calls for a discussion of the potential practical implications of our findings for the design of compensation and reward schemes in management consulting. We argued that our findings have three main implications for reputable management consultancies in Europe:
Focus on getting base salaries right
The base salary component has the potential to drive compensation satisfaction, but may also be a source of dissatisfaction when perceived as inadequate. The base salary component is particularly associated with high levels of compensation satisfaction. The prominence of the component thus increases with compensation satisfaction. Consequently, we suggest that management consultancies should prioritize giving their employees a competitive base salary.
Consider allowances before incentives
Our results did not find any significant association between a respondent’s level of compensation satisfaction and their propensity to highlight either the allowance or the incentive component. In other words, the chance that a respondent would highlight either of the components as the best aspect
107 remains consistent regardless of the respondent’s level of compensation satisfaction. However, we found that the allowance component is highlighted as the best aspect more often than the incentive component is. This is true across all regions and at every level of seniority that we measured. The only situation in which the incentive component ranks more prominently than the allowance component is for management consultants in small consultancies. Consequently, we suggest that consultancies should consider carefully whether to prioritize the allowance component over the incentive component.
Use safety and work-life balance for retention
At last, the safety and work-life balance component may have a positive effect on organizational performance parameters such as employee retention, reputation, and recruitment. Respondents who are generally unsatisfied with their compensation choose to remain with the firm after all, possibly due to the inherent properties of the safety and work-life balance component. Consequently, we propose that management consultancies ensure that their safety and work-life balance component is adequate in terms of providing the necessary perceived support for their employees to perform.
Generalizability of results and suggestions for further research
We argue that the robustness of our analytical framework ensures that our implications are generalizable across our population of interest. However, applying our model to other professional service industries may be fraught with problems, as structural differences in terms of working methods, culture, and regulatory framework may change the underlying dynamics that govern the associations between a respondent’s level of compensation satisfaction and her propensity to highlight the various components of her individual compensation and reward scheme.
As noted in the delimitations sub-section of the introduction, we argued that other time windows may have produced different results. Additionally, it would have been interesting to investigate how preferences and associations may evolve over the course of a longer period. At last, introducing alternative independent confounding variables into the model may provide new insights about potentially significant adjusting effects to the impact of compensation satisfaction. We invite other scholars to further investigate these extensions to our research.
108