Figure 1: Variance of the test-list mean values as a function of the score p. To improve comparability, the variances were multiplied by the number of words n per test list (see Equation 13). The symbols show the variances for different combinations of words in test lists. For each combination, the mean score (10) of all words involved is used as the abscissa p. Since the scores belong to four different sound pressure levels, the symbols are grouped around the four respective mean scores p. The plotted lines show the fitted parabolas according to Equation 14 To relate the values for c2 to the combinations, see Table 2. The value c2=0.332 results from fitting to the variance of single-word recognition scores (n=1) of all words.