In this chapter, the authorsâ€™ consider a reasonable model of error that is often appropriate and shows how that model of error directs them to a certain aggregate index. They assume that the errors have a normal distribution. The authors also assume that a frequency distribution for the error tickets would have the general shape of the normal curve depicted. One of the most troubling possible error distributions is one that has the general shape of the normal distribution except that it has "thicker" tails. The formal assumption is that each error is independent of the value of any other error. The final important assumption is that the errors are unbiased or, equivalently, that the expected value or mean of the distribution from which the errors are sampled equals zero. Depending on the substantive domain, ensuring that the errors are random involves aspects of experimental design, construction of survey questionnaires, interviewing procedures and instrument reliability.