Table 1 Estimated number of people killed according to each candidate log-linear model and overall, based on Bayesian model averaging (Dec 2018-Feb 2019)
From: Deaths, injuries and detentions during civil demonstrations in Sudan: a secondary data analysis
Model (interaction terms included) | Degrees of freedom | Bayesian Information Criterion | Estimate of deaths not on any list or m000 (95% CI) | Posterior probability |
|---|---|---|---|---|
1. No interactions | 3 | 0.30 | 2 (1–4) | 0.036 |
2. [Independent Movement x “Lest We Forget”] | 2 | −5.07 | 5 (2–12) | 0.531 |
3. [Independent Movement x Sudan Doctors’ Union] | 2 | 4.53 | 2 (1–5) | 0.004 |
4. [“Lest We Forget” x Sudan Doctors’ Union] | 2 | 0.30 | 1 (0–3) | 0.036 |
5. [Independent Movement x “Lest We Forget”] + [Independent Movement x Sudan Doctors’ Union] | 1 | −2.87 | 16 (3–127) | 0.177 |
6. [Independent Movement x “Lest We Forget”] + [“Lest We Forget” x Sudan Doctors’ Union] | 1 | −2.77 | 2 (0–9) | 0.168 |
7. [Independent Movement x Sudan Doctors’ Union] + [“Lest We Forget” x Sudan Doctors’ Union] | 1 | 4.45 | 1 (0–3) | 0.005 |
8. [Independent Movement x “Lest We Forget”] + [Independent Movement x Sudan Doctors’ Union] + [“Lest We Forget” x Sudan Doctors’ Union] (saturated) | 0 | 0 | 7 (1–81) | 0.042 |
Bayesian model average | 6 (2–31) | 1.000 |