Table 2.
Estimated regression coefficients describing the differences in the probability of an observation event being a stationary (i.e. point) count among years, in April. Coefficients are from the single logistic regression model fitted to data from each of the four political units separately. These models were parameterized so that the April 2020 proportion of stationary counts is the intercept/reference value, and the coefficients for all earlier years represent deviations from the 2020 proportions. Thus, negative values for regression coefficients for the years prior to 2020 indicate that the proportion of stationary counts was lower in these years than in 2020. Coefficients and standard errors in bold font indicate coefficients for which the 95% confidence intervals did not overlap with zero, indicating statistically reliable estimates. The final column, presenting the estimated proportions of observation events that were from stationary counts, was calculated calculating based only on the fixed effect in the model (i.e. setting the random effect coefficient to zero).
| Region | Predictor | Coefficient | SE | Proportion stationary |
|---|---|---|---|---|
| Spain | Intercept (2020) | 1.59 | 0.06 | 0.831 |
| Year (2017) | −1.97 | 0.04 | 0.408 | |
| Year (2018) | −2.13 | 0.03 | 0.368 | |
| Year (2019) | −2.34 | 0.03 | 0.322 | |
| Portugal | Intercept (2020) | 0.12 | 0.08 | 0.529 |
| Year (2016) | −1.13 | 0.07 | 0.267 | |
| Year (2017) | −0.89 | 0.06 | 0.316 | |
| Year (2018) | −0.32 | 0.05 | 0.450 | |
| Year (2019) | −0.58 | 0.05 | 0.386 | |
| California | Intercept (2020) | −0.39 | 0.05 | 0.404 |
| Year (2016) | −0.40 | 0.02 | 0.313 | |
| Year (2017) | −0.36 | 0.02 | 0.322 | |
| Year (2018) | −0.45 | 0.02 | 0.303 | |
| Year (2019) | −0.44 | 0.02 | 0.303 | |
| New York | Intercept (2020) | −0.19 | 0.07 | 0.453 |
| Year (2016) | 0.13 | 0.02 | 0.486 | |
| Year (2017) | 0.10 | 0.02 | 0.478 | |
| Year (2018) | 0.0002 | 0.02 | 0.453 | |
| Year (2019) | −0.001 | 0.02 | 0.453 |