Skip to main content
. Author manuscript; available in PMC: 2020 Sep 17.
Published in final edited form as: Stat Interface. 2020;13(4):533–549. doi: 10.4310/sii.2020.v13.n4.a10

Table 4.

Estimable covariates for the underlying full model by study and by mapping matrix pattern

Study N Covariate position in the full model Mapping matrix pattern
0 1 2 3 4 5 6 7 8 9 10 11 12
1+ 348 X X X X X - X X X - X - X 1
2 230 X X X X X X - X - X - - -
8a 1486 X X X X X X - X - X - - - 2
8b 2155 X X X X X X - X - X - - -
8c 600 X X X X X X - X - X - - -
9 302 X X X - X X X X X X X X X 3
10.1 348 X X X - - - X X - - X - - 4
11 383 X X X - - X - X X X - X - 5
12 167 X X X X X - X X X - X - X 6
13/14 109 X X X X - X X X X X X - X 7
18 215 X X X X X X - X X X - X - 8
20 928 X X X X - - X X - - X - - 9
21 216 X X X X X X X X X X X X X 10
22 509 X X X - X - X X - - X - - 11

Notes. “X” indicates estimable parameters whereas “−” indicates inestimable parameters. Covariate in the full model are (0) Intercept; (1) Man (= 1 vs. woman = 0); (2) White (= 1 vs. non-white = 0); (3) First-year (= 1 vs. other = 0); (4) PBS (Estimated latent trait (θ) scores for utilizing protective behavioral strategies) at Baseline; (5) PF (stand-alone personalized feedback intervention) (= 1 vs. control = 0); (6) MI + PF (in-person motivational intervention with personalized normative feedback profile) (= 1 vs. control = 0); (7) LS (Linear slope of time in month); (8) QS (Quadratic slope of time in month); (9) LS × PF (vs. control); (10) LS × (MI + PF) (vs. control); (11) QS × PF (vs. control); (12) QS × (MI + PF) (vs. control). N represents sample size at baseline. += Study 1 did not have a control group, thus, PF served as a comparison group.