The kinetic temperature in complex plasmas is often measured using particle tracking velocimetry. Here, we introduce a criterion which minimizes the probability of faulty tracking of particles with normally distributed random displacements in consecutive frames. Faulty particle tracking results in a measurement bias of the deduced velocity distribution function and hence the deduced kinetic temperature. For particles with a normal velocity distribution function, mistracking biases the obtained velocity distribution function towards small velocities at the expense of large velocities, i.e., the inferred velocity distribution is more peaked and its tail is less pronounced. The kinetic temperature is therefore systematically underestimated in measurements. We give a prescription to mitigate this type of error.