ABSTRACT
Twenty scoring systems are used on NORC's occupational prestige values. The linear scoring system lies midway between the other scoring systems, which by definition excludes the most extreme scoring systems in each direction. The resulting largely random scoring systems vary among themselves on the actual values assigned to each rank, the range of values, and the size of the differences between adjacent values. The study suggests two research strategies when analyzing ordinal variables. First, a linear scoring system is assigned according to the available evidence on the distances between ranks. Second, all available rank order categories are used, rather than collapsing them into a smaller number, because the greater the number of ranks the greater the stability and confidence in the assigned scoring system. Interpretations of path coefficients do not appear to require modification by the assignment of numbers to ordinal categories.
