Skip to main content
. Author manuscript; available in PMC: 2021 Jul 1.
Published in final edited form as: Obesity (Silver Spring). 2020 May 22;28(7):1224–1234. doi: 10.1002/oby.22817

Table 5.

Adjusted least square means for between-group comparisons: diet changes

Within-group change (intervention) Within-group change (control) Between-group difference (intervention - control)
Mean Change (95% CI) Mean Change (95% CI) Mean Change (95% CI) p-value
Among all social network members
REAP-S sum score
Very close relationship (intervention n=111, control n=62) −0.23 (−1.33, +0.86) +0.62 (−0.70, +1.93) −0.85 (−2.34, +0.64) 0.264
Somewhat or not very close relationship (intervention n=72, control n=61) +0.66 (−0.72, +2.03) −0.97 (−2.42, +0.48) +1.63 (−0.08, +3.34) 0.063
Spatially close (intervention n=106, control n=82) −0.06 (−1.19, +1.08) +0.18 (−1.06, +1.42) −0.24 (−1.67, +1.19) 0.746
Not spatially close (intervention n=77, control n=41) +0.14 (−1.09, +1.36) −0.69 (−2.18, +0.81) +0.82 (−0.90, +2.54) 0.348
Social network members who had a very close relationship with trial participants
REAP-S sum score
Spatially close (intervention n=60, control ln=37) −0.05 (−1.47, +1.38) +1.81 (+0.08, +3.55) −1.86 (−3.85, +0.13) 0.066
Not spatially close (intervention n=51, control n=25) +0.13 (−1.45, +1.71) +0.10 (−1.90, +2.10) +0.10 (−2.11, +2.30) 0.932
Social network members who were spatially close to trial participants
REAP-S sum score
Very close relationship (intervention n=57, control n=37) −0.36 (−1.76, +1.03) +1.37 (−0.28, +3.02) −1.73 (−3.64, +0.17) 0.075
Somewhat or not very close relationship (intervention n=46, control n=45) +0.51 (−1.26, +2.28) −0.78 (−2.50, +0.95) +1.29 (−0.72, +3.29) 0.208

Significant p-values are indicated in bold. All models controlled for social network members’ age, sex, education, relationship status, employment status, self-rated health, trial participants’ program attendance, and baseline value of outcome of interest as fixed effects. Trial program sites and trial participants’ identification number were treated as random effects in all models.