differences between log-linear models and the analysis
of variance, 641
difference scores, 197, 622
directional test, 100
discrete variables, 4
versus continuous variables, 118
discriminant analysis, 562
with dichotomous dependent variable, 571
disordinal interactions, 423
dispersion, 36
distance, 540
distribution assumptions, 531
distribution-free tests, 660
distributions of the two kinds of variables, 118
double subscripts, 30
dummy variables, 581
Dunnett’s test, 377, 395
Dunn-Sidák test, 378ˇ
Dunn’s test, 377
effective sample size, 235
effect of heterogeneous subsamples, 282–283
effect of range restrictions, 281
effect size, 104–105, 386
d-family: risks and odds, 160
estimating the effect size, 229–230
example, 159–160
odds ratios in 2 32 3 ktables, 163–164
odds ratios in 2 3 ktables, 162–163
phi (f) and Cramér’s V, 164
recombining the effect size and n, 230–231
r-family: phi and Cramér’s V, 164
effect size computation in an analysis of covariance, 609–611
d-family measure, 610–611
r-family measure, 609–610
efficiency, 46
equally weighted means, 444
equal sample sizes, 233
equiprobability model, 633
error rate per comparison (PC), 364
error rates, 364–367
error rate per comparison (PC), 364
familywise error rate (FW), 365
the null hypothesis and error rates, 365
a priori versus post hoc comparisons, 365
significance of the overall F, 366
errors of prediction, 254
errors of prediction as a function of r, 261
error variance, 259, 320
estimating required sample size, 230–232
estimating the effect size, 229–230
eta-squared (h^2 ), 344–345
evaluation of x^2 , 147
event, 114
examining the saturated model, 647
Excel (commercial software), 9
exhaustive events, 114
expected cell frequencies, 653–654
expected frequencies, 142
expected frequencies for contingency tables, 145–146
expected mean squares, 432
expected mean squares and alternative designs, 430–438
calculation for nested designs, 435–436
crossed experimental design with a random variable, 432
crossed experimental design with fixed variables, 432
nested designs, 435
expected value, 46, 323
exploratory data analysis (EDA), 5, 24
exponentiation, 568
external validity, 3factorial, 122
factorial analysis of covariance, 612–621
adjusted means, 320–621
testing adjusted means, 619–622
factorial analysis of variance, 413–455
analysis of variance applied to the effects of
smoking, 426–428
a computer example, 453–455
expected mean squares and alternative designs,
430–438
extension of the Eysenck study, 416–420
higher-order factorial designs, 446–453
interactions, 421–423
measures of association and effect size, 438–443
multiple comparisons, 428–429
notation, 415
power analysis for factorial experiments, 429–430
reporting the results, 443–444
simple effects, 423–426
structural models and expected mean squares, 420–421
unequal sample sizes, 444–446
factorial designs, 414, 586–593
full model, 587–590
reduced model, 591–593
factors, 414
factors affecting the power of a test, 227–229
basic concept, 227
power as a function of alpha, 227–228
power as a function of H 1 , 228
factors that affect the correlation, 281–283
effect of heterogeneous subsamples, 282–283
effect of range restrictions, 281
failure, 127
false discovery rate (FDR), 396
familywise error rate (FW), 365
first-order interactions, 446
first quartile Q1, 39
Fisher, R. A., 90, 94
Fisher’s arcsine transformation, 670760 Index