A value of r=0.68 is actually fairly large. It means that r^2 = 46.2% of the variance in your dependent variable can be explained by a linear relationship between the independent and dependent variables. In your case, almost half (46%) of the "motivation" of your employees can be "explained" by their degree of involvement with decision-making.
The actual *significance* of r depends on how many data points are in your sample, and can be quantified using what is called the "t-test". If certain assumptions about the data are true (e.g., the errors are normally distributed), then one can determine what the probability of getting a particular value of r is, simply by chance. That is, we ask what the probability of getting r = 0.68 is, if, in reality, the data are completely uncorrelated (r is actually equal to zero).
First, we calculate the t-statistic for the data:
t = r * sqrt((N-2)/(1-r^2))
where N is the number of observations (data points) in your sample, and N-2 is the number of "degrees of freedom" in your sample. (It's two less than N because one has to assume that the mean and variance of the real population are equal to the mean and variance of your sample.)
Then one compares the value of t calculated above with probability of obtaining that value for a t-distribution with the same number of degrees of freedom. That's the probability of getting the observed r value by chance. You then have to assess whether that probability is significant.
There are actually two cases to be considered here. If one's hypothesis about the relationship between the independent and dependent variables has a particular sign or direction (e.g., higher involvement will result in higher motivation) then one does a "1-tailed" comparison,whereas if the hypothesis makes no assumption as to the sign of the correlation (e.g., involvement will affect motivation, either positively or negatively), then one does a "2-tailed" comparison. In your case, I think the 1-tailed comparison is appropriate.
The little web applet at the second source below will do these calculations for you. In your case, for the sake of illustration, if you had data on 10 employees (N = 10), then the 1-tailed probability of getting a value of r=0.68 by chance, if there is actually no correlation, is 1.5%. (The 2-tailed probability is 3.0%). Most people would consider this highly significant, and one can "reject" the hypothesis that there is no correlation between the variables.
Before you do any of this, however, you should plot your data and look at it! Do the data seem to cluster around a straight line, or is there curvature in the cloud of data? Are there "odd" (outlier) data points? Are all the data, with few exceptions, clustered in one "cloud"? Remember that the usual correlation coefficient *only* ascertains the degree of linear (straight-line) correlation between your variables. If the data have an obvious non-linear relationship, then a linear correlation coefficient is not the appropriate measure to use to determine if there is a statistically significant relationship between your variables.
Finally, you need to remember that correlation is *not* a proof of causality. That is, just because A and B are correlated does not mean that A causes B, or that B causes A. For instance, if C causes A and C causes B, but A has no effect of B, A and B will still be acausally correlated because of their mutual dependence on C.
For example, I can imagine that only highly paid managers are involved in decision-making, and that highly paid employees are highly motivated. One would then find that decision-makers are highly motivated, yet the real "cause" of their motivation is the fact that they are highly paid, not that they are involved in decisions.