Failure to add to 1—to be “justified” to 1—occurs frequently in reported sets of rounded data. The probabilities for different rules of rounding vectors and tables to give “justified” results under varying assumptions concerning the probabilistic structure of the data are computed. This extends the pioneering work of Mosteller Youtz, and Zahn (1967) and Diaconis and Freedman (1979) wh< assessed the probability that conventionally rounded proportions add to 1. In particular, it is shown that other rules of rounding can improve upon the conventional rule.
- rounding vectors rounding tables rounding proportionss probability metrics