|
MeasurementJack Meredith, Samuel Mantel, Jr. 2003 Project Management John Wiley and Sons, New York, pp. 76-77 Measurement It is common for those who oppose a project, for whatever reason, to complain that information supporting the project is "subjective." This epithet appears to mean that the data are biased and therefore untrustworthy. To use the scoring methods discussed or to practice risk management in project selection, we need to represent though not necessarily collect expected project performance for each criterion in numeric form. If a performance characteristic cannot be measured directly as a number, it may be useful to characterize performance verbally and then, through a word/number equivalency scale, use the numeric equivalents of verbal characterization as model inputs. Subjective versus Objective These distinction between subjective and objective is generally misunderstood. All too often objective is held to be synonymous with fact and subjective is taken to be a synonym for opinion–where fact = true and opinion = false. The distinction in measurement theory is quite different, referring to the location of the standard for measurement. A measurement taken by references to an external standard is said to be "objective." Reference to a standard that is internal to the system is said to be "subjective." A yardstick, incorrectly divided into 100 divisions and labeled "meter," would be an objective but inaccurate measure. The eye of an experienced judge is a subjective measure that may be quite accurate. Quantitative versus Qualitative The distinction between quantitative and qualitative is also misunderstood. It is not the same as numeric and nonnumeric. Both quantity and quality may be measured numerically. the number of words on this page is a quantity. The color of a red rose is a quality, but it is also a wavelength that can be measured numerically, in terms of microns. The true distinction is that one may apply the law of addition to quantities but not to qualities (van Gigeh, 1978). Water, for example, has a volumetric measure and a density measure. The former is quantitative and the latter qualitative. Two one-gallon containers of water poured into one lager container gives us two gallons, but the density of the water, before and after joining the two gallons, is still the same: 1.0. Reliable versus Unreliable A data source is said to be reliable if repetitions of a measurement produce results that vary from one another by less than a prespecified amount. The distinctions is important when we consider the use of statistical data in our selection models. Valid versus Invalid Validity measures the extent to which a piece of information actually means what we believe it to mean. A measure may be reliable but not valid. Consider our mismarked 36-inch yardstick pretending to be a meter. It performs consistently, so it is reliable. It does not, however, match up accurately with other meter rules, so it would not be judged valid. To be satisfactory when used in the previous project selection models, the measures may be either subjective or objective, quantitative or qualitative, but they must be numeric, reliable and valid. Avoiding information merely because it is subjective or qualitative is an error and weakens decisions. On the other hand, including information of questionable reliability or validity in selection models, even though it may be numeric, is dangerous. It is doubly dangerous if decision makers are comfortable dealing with the selection model but are unaware of the doubtful character of some input data. A condition a colleague has referred to as GIGO–garbage in, gospel out–may prevail back to ideas |