Evaluating Main Effect
Political Orientation and Taste Study
- The study includes two variables: political orientation (ConLib) and taste condition (Conditioned).
- ConLib was not statistically significant, but Conditioned was significant as a main effect.
- The analysis will focus on following up on the Conditioned variable only.
- Mean squares error will be used for the error term in planned comparisons.
- Corrected total sums of squares will be used to calculate eta squared.
Main Effect of Taste
- The effect of liberal versus conservative (political orientation) was not significant.
- The means for liberals and conservatives showed no statistically meaningful differences in moral judgments.
- The main effects were observed in the bitter versus water versus sweet taste conditions.
- The hypothesis was that the bitter condition would lead to more extreme moral judgments.
Pairwise and Complex Comparisons
- The approach involves first comparing water and sweet conditions.
- If no significant difference is found, these two conditions will be combined and compared to the bitter condition.
- The means for water and sweet conditions are nearly identical.
- An F ratio and p-value will still be computed and reported even if the effect isn't significant.
Pairwise Comparison: Water vs. Sweet
- Compute psi, which is the difference between the means of water and sweet conditions.
- psi = 0.0097 (high precision due to slight differences in means)
- Calculate the sums of squares for the comparison using the formula:
SS_{comparison} = \frac{N \times psi^2}{Contrast\ Sum\ of\ Squares} - N is the sample size associated with the means being compared (N = 40 in this case).
- Because it is a single degree of freedom comparison, the sums of squares value is also the mean squares value. MS = SS
F Ratio
- Compute the F ratio using the formula:
F = \frac{MS{comparison}}{MS{error}} - The error term (mean squares error) is taken directly from the SPSS output.
- The result was not statistically significant.
- Differences observed are likely due to sampling error.
Effect Size: Eta Squared
- Compute eta squared as a measure of effect size:
η^2 = \frac{SS{effect}}{SS{total}} - Use the corrected total sums of squares from the SPSS output.
- In this case, eta squared is 0, indicating no effect.
Complex Comparison: Bitter vs. Combination of Water and Sweet
- Compute psi using contrast weights.
- Contrast weights used: 2 (bitter), -1 (water), -1 (sweet).
- The formula is: \text{psi} = c1 \times \text{mean}1 + c2 \times \text{mean}2 + c3 \times \text{mean}3 where c_i are the contrast weights.
- Means are obtained from the SPSS output.
Sums of Squares
Calculate the sums of squares using the appropriate formula.
The resulting sums of squares value is also the mean squares value (single degree of freedom comparison).
F Ratio for the Complex Comparison
Calculate the F ratio using the formula:
F = \frac{MS{comparison}}{MS{error}}
The calculated F ratio exceeds the critical value (3.92).
The bitter condition is significantly different from the combination of water and sweet conditions.
Effect Size: Eta Squared for the Complex Comparison
Compute eta squared to determine the proportion of variance accounted for.
η^2 = \frac{SS{effect}}{SS{total}}
Eta squared = 0.3031, indicating that approximately 30% of the variation in moral judgment is accounted for by the difference between bitter and the combined water/sweet conditions. This indicates a fairly strong effect.