Table 1 Details of model fitting for ten different settings for COVID-19 cluster size data [8]. One of three possible models [(i) – (iv), as described in the text] was chosen based on the log-likelihood ratio statistic, the value of which is given for testing a simpler model versus a more complex one [β = 0 for (ii) versus (i), and c = 1 for (iv) versus (iii)]. The p-value of this statistic is also given
From: Comparisons of statistical distributions for cluster sizes in a developing pandemic
Setting | Model chosen | Log-likelihood ratio statistic for comparison with alternative model | p-value for comparison with alternative model |
|---|---|---|---|
Elderly care | Negative binomial | 1.54 | 0.21 |
Food processing plants | Negative binomial | 0.25 | 0.62 |
Household | Poisson | 2.30 | 0.13 |
Large shared accommodation | EPPM | 5.18 | 0.023 |
Meal | EPPM | 7.19 | 0.0073 |
Party | Negative binomial | 0.44 | 0.51 |
Religious | Negative binomial | 0.12 | 0.73 |
School | Negative binomial | 0.53 | 0.47 |
Sport | Negative binomial | 1.15 | 0.28 |
Work | EPPM | 10.00 | 0.0016 |