One goal of our study is to find out the difference of phthalate exposure level between different groups of people. Therefore, we need to compare phthalate exposure level between different groups of people. As the distribution of observed urine phthalate concentration is skewed, we applied natural logrism to it. Then, we devided people into subgroups and compared the least square mean of log(phthalate concentration) between each group. Least square means are means for groups that are adjusted for other terms in the model. In this case, least square means are means of log(sum-total exposure) of subgroups, adjusting for all predictors in the model except for the predictor of interest. For example, least square mean of log(sum-total exposure) of males is the mean log(sum-total exposure) value for male, adjusting for race, poverty status, and BMI.

Plots above show least square means of log(sum-total exposure) of groups by race, poverty status and gender, and their 95% confidence intervals. The plots reveal several facts:

• Compared to other race groups, levels of the phthalate exposure were significantly higher in non_hispanic_blacks.
• Compared to the group who are not living in poverty, levels of the phthalate exposure were significantly higher in the group who are living in poverty.
• Compared to females, levels of the phthalate exposure were higher in males.