Analyses were stratified by ethnicity and a formal interaction test was used to establish p-values for the null hypothesis of no difference in association (linear or non-linear) between the ethnic groups (exposure ethnicity interaction).
The results indicate a substantive condition by participant ethnicity interaction, in which African Americans assign greater credibility to high ethnic identity spokespeople, while Caucasian respondents found low ethnic identity spokespersons more credible.
However, there was no time by ethnicity interaction indicating that while both groups improved over time, the ethnic difference remained after the intervention.
There was not a significant treatment by ethnicity interaction for any of the lipids.
The final model included age group (p<0.0001), sex (p<0.0001), race and ethnicity (p=0.0158), and the age group-by-race and ethnicity interaction (p=0.0072).
As the age squared by ethnicity interaction did not improve the model fit and the term itself was not significant, we dropped this term from all the models.
All models were adjusted for ethnicity and income In multivariable analyses including neighborhood factor × ethnicity interaction terms, Recreation × Ethnicity was significant for both timing of breast and pubic hair development (Table 5).
Moreover, to calculate formal ethnicity interaction in the relationship with physical activity and socioeconomic position, we ran additional sex-stratified models, including interaction terms for ethnicity and socioeconomic position.
In women, we observed the reverse for the socioeconomic position measures: there was no ethnicity interaction for active commuting by level of education, whereas we found a statistically significant differential association by ethnicity for occupational class.
