Table 2.
Average effect on probability of choosing a new contraceptive method, by method selected; new contraceptive clients visiting family planning clinics in Salt Lake County; March 2016 to March 2017; N = 2188
| Reasons for choosing a new birth control method, “quite or extremely important” relative to “somewhat or not at all important” | LNg IUD |
Implant |
Combined oral contraceptives |
Copper IUD |
3-month injectable |
Vaginal ring |
|---|---|---|---|---|---|---|
| Average effect of choosing “quite or extremely important” on the probability of selecting a method (CI) | ||||||
| It's safe | −.006 | −.079 | .054 | .061 | −.061 | .031 |
| (−.183 to .171) | (−.242 to .084) | (−.047 to .155) | (−.058 to .181) | (−.162 to .039) | (−.008 to .070) | |
| It's the most effective method | .154 | .038 | −.177 | .027 | −.025 | −.017 |
| (.092–.216) | (−.036 to .111) | (−.270 to −.084) | (−.048 to .102) | (−.087 to .037) | (−.065 to .032) | |
| It doesn't interrupt sex | −.038 | .060 | −.004 | .012 | −.001 | −.029 |
| (−.095 to .019) | (.014–.106) | (−.050 to .043) | (−.032 to .056) | (−.038 to .037) | (−.064 to .005) | |
| It doesn't reduce my libido | −.007 | −.013 | −.006 | .031 | −.014 | .009 |
| (−.062 to .048) | (−.064 to .038) | (−.055 to .043) | (−.014 to .076) | (−.053 to .025) | (−.015 to .035) | |
| It doesn't have side effects | .054 | −.007 | −.003 | −.049 | −.013 | .018 |
| (.009–.098) | (−.051 to .037) | (−.045 to .040) | (−.093 to −.004) | (−.048 to .022) | (−.003 to .038) | |
| It is acceptable to my partner | .037 | −.033 | .002 | .006 | −.017 | .005 |
| (−.005 to .079) | (−.072 to .006) | (−.036 to .040) | (−.024 to .036) | (−.047 to .013) | (−.016 to .025) | |
| It doesn't contain hormones | −.160 | −.080 | −.027 | .303 | −.022 | −.015 |
| (−.197 to −.123) | (−.062 to −.048) | (−.062 to .009) | (.267–.339) | (−.051 to .006) | (−.033 to .004) | |
| It's recommended by my friends | .076 | .035 | −.080 | −.013 | −.014 | −.004 |
| (.033–.120) | (−.006 to .076) | (−.116 to to −.045) | (−.042 to .016) | (−.045 to .016) | (−.025 to .018) | |
| It's in line with my religious beliefs | .037 | −.077 | .031 | −.016 | .033 | −.009 |
| (−.026 to .099) | (−.127 to −.026) | (−.030 to .093) | (−.054 to .022) | (−.015 to .082) | (−.038 to .021) | |
Underlined numbers represent significance at the p <.05 value or below.
Coefficients are for null hypothesis of no effect on predicted probability of method selection.
Results are reported as average effects on the probability of method selection using the margins postestimation command in Stata 15.
All models are multinomial logistic, include full sets of control and independent variables, and are run on complete case datasets. The dependent variable is a categorical variable to identify whether that method was selected vs. other methods. Control variables include the household size, percent of federal poverty level, relationship status, race, sexual orientation, employment status, student status, unstable housing conditions, education and age category.