An MD&DI August 1997 Column
The salary estimator should not be used to determine real salaries or salary ranges. Some will no doubt ignore this warning--but they do so at their own risk.
Of all the articles we publish each year, few are more popular than our annual salary survey review. I learned this early in my career at MD&DI when I decided to publish the survey results every other year instead of annually. By the next year, the number of phone calls I'd received asking when the salary survey would appear had shown me my error. We've published it every year since then.
Nothing so popular can be without controversy, however. This seems to be especially true of the salary estimator that we include with each article (as well as on the Web in an interactive format at http://www.devicelink.com/career/).
The idea behind the salary estimator is simple, even if the terminology isn't. By means of multiple-regression analysis, our survey company, Readex, is able to identify a number of key variables that help determine an individual's salary. The resulting worksheet allows readers to answer a few questions and calculate a predicted salary based on those variables.
The worksheet is a lot of fun to play with--and that's the main point. We don't intend it to be used by human resources departments or anyone else to determine real salaries or salary ranges. Some will no doubt ignore this warning--but they do so at their own risk.
The caveats supplied in the article are worth repeating here. The salary prediction model is described as "moderately powerful," and the emphasis should be on the word moderately. The model explains about half of the variation in salaries for the people we surveyed. This is so because the survey cannot accurately measure all of the significant factors that affect salaries, most notably individual job performance. As a result, about a third of the time, the model's prediction of an individual's salary will be off by 27% or more. So if you find that your salary is less than predicted, please don't leap to the conclusion that you are in fact undercompensated.
The worksheet isn't just a toy, however. As noted in the article, the model "provides a sense of the relative importance of each factor in predicting salary." This year, for instance, it tells you that the size of the company you work for is likely to affect your salary, and that if you work in quality assurance, your contributions may well be less valued than those of other job functions.
The fact that we're sharing this information should not be construed to mean we are promoting them or that we necessarily approve of them. It's just what the numbers tell us.
Occasionally our readers don't realize this fact, and react with dismay or outrage at what the model says. In last year's worksheet, for instance, gender was listed as a factor determining salary. Noting this fact, one reader wrote to complain that we were promoting this inequity by including it in the model. Instead, the reader said, we should have simply omitted it from the calculation.
If we were presenting the model as a tool for determining salary, the reader would be absolutely right. But that is not at all the case. The purpose of the regression analysis is to identify the factors that contribute to salary, whether justified or not. To hide any factor could cover up a serious industrywide problem. If race had been listed in the worksheet, the reader said, there would have been a public outcry. As it happens, our survey does ask about race. And rest assured that if race were a factor, we would not cover up that fact.
This year, by the way, our analysis did not identify gender as a significant factor in determining salaries. I think its absence from the model this time would be less notable had we hidden it last year.
In any event, I encourage you to try out the worksheet--and, of course, to use it wisely.