About the Salary Estimator's Statistical Model
Return to the salary estimator
Note: The worksheet applies only to full-time professionals in firms manufacturing medical devices or in vitro diagnostics, with salaries ranging from $30,000 to $160,000.
The model provides a sense of the relative importance of each factor in predicting salary. It is not used appropriately as a basis for comparison with any individual’s situation.
This worksheet was originally published as part of an article in Medical Device & Diagnostic Industry ("Job Satisfaction Remains High in Device Industry," Gregg Nighswonger, October 99, p. 95). The information presented in that article provides a broad framework for evaluating employee compensation in the U.S. medical device industry, based on the responses to MD&DI's annual salary survey for 1999. Most of the responses analyzed are linear—that is, they use cross-tabulation to compare one variable, such as job function, and analyze it simultaneously with a second variable, such as number of employees supervised. To account for the influence of additional variables, multivariate analysis is used to examine all variables simultaneously, account for their interdependence, and identify those that have the highest degree of predictive power. In any given year, some factors may not help to predict salary—usually because they correlate strongly with other variables already included in the model.
The variables that were determined to have the greatest predictive value formed the basis for this year's salary approximation worksheet. Statistically speaking, the model is only moderately powerful—it explains a little more than 50% of the variation in salary (adjusted R-square = 0.541) and is significant by the F-test at p<0.000. About one-third of the time, its prediction of an individual's salary will be off by 25% or more from the actual value. Predictions tend to be most accurate for salary values in the middle of the range.
