"Pure" academic research rules on data collection do not apply directly to operations research (OR). OR data collection should be viewed in terms of objective, cost, and effectiveness. For the model formulation objective, proper data strategies emphasize multiple views of the operating system to identify the "relatednesses" to be depicted. For the objective of estimating parameters or testing predictions, bias, precision, and level of confidence of results are effectiveness concepts to be balanced against cost. Decision and utility theory, sensitivity analysis, and sequential analysis apply or OR data collection strategies and employ operational parameters to define data needed and, hence, minimize costs. (Author)