Table 1 Mixed-effects models for treatment and time fixed effects

From: Mobilising climate action with moral appeals in a smartphone-based 8-week field experiment

Outcome variable

Best model

LRT

FE (treatment)

FE (time– date)

FE (interaction)

Log carbon footprint overall

Mixed effect model (random intercept) with interaction between time and treatment fixed effect

base:

490.83, p < 0.001

with time FE only:

23.709, p < 0.001

0.145 (−0.12, 0.42)

0.018 (0.016, 0.02)

−0.006 (−0.009, −0.004)

Heating carbon footprint

Mixed effect model (random intercept, treatment random slope), with interaction between time and treatment fixed effect

base:

728.19, p < 0.001

no interaction:

11.07, p = 0.001

no random slope:

25.11, p < 0.001

−4.148 (−13.9, 6.77)

0.444 (0.40, 0.49)

−0.098 (−0.15, −0.04)

Food carbon footprint

Mixed effect model (random intercept, treatment random slope), with interaction between time and treatment fixed effect

base:

32.568, p < 0.001

no interaction:

6.4963, p = 0.039a,

no random slope:

25.884, p < 0.001

1.578 (−0.89, 3.82)

0.029 (0.006, 0.05)

−0.031 (−0.06, 0.004)

Consumption carbon footprint

Mixed effect model (random intercept, treatment random slope), treatment & time fixed effects

base:

20.9, p = 0.0003

no time FE:

9.28, p = 0.002

−1.04 (−3.08, 1.07)

0.063 (0.02, 0.10)

NA

Electricity carbon footprint

Mixed effect model (random intercept), time fixed effect

base: 120, p < 0.001

NA

0.007 (0.006, 0.01)

NA

Carbon footprint from flying

Base model (random intercept)

NA

NA

NA

NA

Car carbon footprint

Mixed effect Model (random intercept, treatment random slope), treatment fixed effect

base:

68.484, p < 0.001

no random slope:

64.135, p < 0.001

4.670 (0.55, 9.10)

NA

NA

Civic climate positivity score

Zero-inflated Poisson mixed effect model (random intercept), treatment & time fixed effects, time fixed effect for excess zeros

base:

67.65, p < 0.001

with random slope

no time FE:

56.36, p < 0.001

0.293 (0.04, 0.55)

0.008

(0.006, 0.01)

zero-part

coefficient: 0.003

NA

  1. Note that analyses with domain-specific outcomes are explorative. Full details of all models in Supplementary Tables 2.1, 2.2, 2.3, 2.4, 2.5, 2.6, 2.7, 2.8.
  2. LRT likelihood ratio test, against the base model and, where applicable, against the next best model(s) that were better than the base model.
  3. base: base mixed effect model with just random intercept.
  4. FE fixed effect.
  5. 95 % Confidence interval in brackets.
  6. Bonferroni-corrected alpha value for confirmatory (pre-registered) tests is 0.002 and 0.008 for exploratory (non-pre-registered) tests. a Models that do not pass the Bonferroni-corrected alpha value.
  7. N = 7615 (156 participants, 56 days).