The 95% confidence range indicates that we can be 95% confident that the true figure is between 200 and 240 fatalities. The central estimate for 2020 is lower than the final figure for 2019, but the decrease is not statistically significant. This represents about 15% of all deaths in reported road collisions in 2020. The final central estimate of the number of deaths in collisions with at least one driver over the alcohol limit for 2020 is 220. Otherwise, we say the change is ‘not statistically significant’ – this does not necessarily mean that there is not a genuine change, but that we are not able to determine one exists from the data available, at the chosen level of confidence.
When we compare figures for the latest year with the previous year, we say that a change is ‘statistically significant’ if we are sufficiently confident (at the chosen level – here, 95%), based on the data available, that the difference between the two figures is the result of a genuine change, rather than being a product of chance - broadly, this is when the two sets of bars on chart 1 don’t overlap. Technically, it indicates that in 100 years with the same risk of fatalities (or injury), 95 of those years will result in a number of fatalities (or injuries) between a given range. The bars on chart 1 are ranges of values for an estimate which we are 95% confident that the ‘true’ value falls in. The 95% confidence level is the standard against which statistics are typically tested. This means that it is impossible to be sure of the precise number of fatalities, so ranges and confidence intervals are used for fatalities throughout the publication. These statistics, especially the number of fatalities, are subject to considerable uncertainty (see sampling uncertainty below).
0 kommentar(er)
