The relative size of the treated and untreated groups or the T/UT ratio in an analysis sample often departs from the T/UT ratio in the population or original sample because of missing values and data limitations. While departures of the sample from the population T/UT ratio do not generate bias for many estimators, instrumental variable (IV) estimates are biased by such departures even when the IV is analytically valid for the population in estimating the Local Average Treatment Effect (LATE). We prove that the bias in IV estimates generated by departures of the sample from population T/UT ratio is a monotonic function of the difference between the sample and population T/UT ratios. If researchers know that their sample T/UT ratio is smaller or larger than the population T/UT ratio, then they can identify the direction of the bias of their IV estimates.  Additionally, we construct a new diagnostic procedure to detect invalid IVs. This diagnostic procedure allows researchers to identify and discard invalid IVs even when they do not know the true population T/UT ratio. Our findings are verified by empirical analyses and simulation studies. The simulations also indicate that the diagnostic procedure works best with large samples and strong IVs, the conditions that are also ideal for using IV estimation.