The temporal decay of mass accretion in young stars is a fundamental tracer of the early evolution of circumstellar disks . Through population syntheses , we study how correlated uncertainties between the estimated parameters of young stars ( luminosity , temperature , mass , age ) and mass accretion rates \dot { M } _ { acc } , as well as observational selection effects , can bias the temporal decay of mass accretion rates ( \dot { M } _ { acc } \propto t ^ { - \eta } ) inferred from a comparison of measured \dot { M } _ { acc } with isochronal ages in young stellar clusters . We find that the presence of realistic uncertainties reduces the measured value of \eta by up to a factor of 3 , leading to the inference of shallower decays than the true value . This suggests a much faster temporal decay of \dot { M } _ { acc } than generally assumed . When considering the minimum uncertainties in ages affecting the Orion Nebula Cluster , the observed value \eta \sim 1.4 , typical of Galactic star forming regions , can only be reproduced if the real decay exponent is \eta \gtrsim 4 . This effect becomes more severe if one assumes that observational uncertainties are larger , as required by some fast star formation scenarios . Our analysis shows that while selection effects due to sample incompleteness do bias \eta , they can not alter this main result and strengthen it in many cases . A remaining uncertainty in our work is that it applies to the most commonly used and simple relationship between \dot { M } _ { acc } , the accretion luminosity and the stellar parameters . We briefly explore how a more complex interplay between these quantities might change the results .