Suppose the variables $x$ and $y$ have a strong positive linear correlation ($r = 0.85$) when measured over a wide range of values. If a new sample is collected where the values of $x$ are restricted to a very narrow range near its mean, what is the most likely effect on the calculated correlation coefficient $r$ of the new sample?