Statistical Methods for Psychology

the conclusions. The calculation of the appropriate tests was carried out the same way it

was earlier, by running a reduced analysis of variance using only the Phase 1 (or Phase 2)

cells. Here again we are using separate error terms to test the Shock and No-shock effects,

thus reducing problems with the sphericity assumption. (Again, just because the analyses

also give simple effects due to Groups and Cycles is no reason to feel an obligation to

interpret them. If they don’t speak to issues raised by the experimental hypotheses, they

should neither be reported nor interpreted unless you take steps to minimize the increase in

the experimentwise error rate.)

Section 14.9 Two Within-Subjects Variables and One Between-Subjects Variable 493

Table 14.12 SPSS output of the analysis of conditioned suppression data

Tests of Between-Subjects Effects
Measure: MEASURE_1
Transformed Variable: Average

Tests of Within-Subjects Effects
Measure: MEASURE_1

Source

Type III Sum

of Squares df Mean Square F Sig.

Cycle Sphericity

Assumed

2726.974 3 908.991 12.027 .000

Cycle * Group Sphericity

Assumed

1047.073 6 174.512 2.309 .044

Error(Cycle) Sphericity

Assumed

4761.328 63 75.77

Phase Sphericity

Assumed

11703.130 1 11703.130 129.855 .000

Phase * Group Sphericity

Assumed

4054.385 2 2027.193 22.493 .000

Error(Phase) Sphericity

Assumed

1892.609 21 90.124

Cycle * Phase Sphericity

Assumed

741.516 3 247.172 4.035 .011

Cycle *

Phase *

Group

Sphericity

Assumed

1273.781 6 212.297 3.466 .005

Error(Cycle *

Phase)

Sphericity

Assumed

3859.078 63 61.255

Source

Type III Sum

of Squares df Mean Square F Sig.

Intercept 156009.005 1 156009.005 208.364 .000

Group 4616.760 2 2308.380 3.083 .067

Error 15723.359 21 748.731