Section Five LIMITATIONS OF INDIVIDUAL PRACTICES
As noted earlier, individual human resource practices tend to be
highly intercorrelated. As a result, it is easy to overestimate
their actual impact on performance. For example, if it is found
that merit-based compensation practices have a significant
correlation coefficient of r = .40 with firm profitability, then the
statistical interpretation is that such compensation systems
account for 16 percent (r^2 ) of the variance in firm profitability.
Unfortunately, this would be a misleading interpretation
because the compensation practice is probably only one of
several other highly correlated human resource practices that
are simultaneously influencing profitability. A more accurate
interpretation would be that several high-performance practices
are responsible for 16 percent of the variance in profitability.^92
Indeed, the Delaney and Huselid study cited earlier, based on
the National Organizations Survey, found that simple
relationships between individual human resource practices and
measures of firm performance overstate their contributions
because when additional practices are added to multiple
regression equations, the size of their coefficients (magnitude
of contribution) decreases and may even become statistically
insignificant.^93