Synchronisation issues between national and subnational data?

[gunnar-kaestle]

Example The UK and DE are shown on the map as (likely) increasing, although almost all subnational entities show R values below 1.

Is it possible that this inconsistency is because the estimations are done on separate and different time frames regarding the evaluation as national and subnational level? It is a puzzling effect, and could be healed if an update at subnational level is done, also the national numbers are redone or vice versa.

[gunnar-kaestle]

DE Subnationals (Table as of 2021-04-20):
0.92 - Baden-Württemberg
0.97 - Bayern
0.83 - Berlin
0.92 - Brandenburg
0.80 - Bremen
0.62 - Hamburg
0.95 - Hessen
0.99 - Mecklenburg-Vorpommern
0.92 - Niedersachsen
0.95 - Nordrhein-Westfalen
0.96 - Rheinland-Pfalz
0.98 - Saarland
0.78 - Sachsen
0.77 - Sachsen-Anhalt
0.94 - Schleswig-Holstein
0.95 - Thüringen

This does not fit to the red increasing colour on the global map, quoting a reproduction number of 1.3 (as of 2021-04-19).

[gunnar-kaestle]

UK Subnationals (Table as of 2021-04-17):
0.89 - East of England
0.88 - England
0.96 - London
0.86 - Midlands
0.84 - North East and Yorkshire
0.83 - North West
1.10 - Northern Ireland
0.95 - Scotland
0.92 - South East
1.00 - South West
0.91 - United Kingdom
0.88 - Wales

This does not fit to the orange likely increasing colour on the global map, quoting a reproduction number of 1.2 (as of 2021-04-19).

[seabbs]

Thanks for flagging. I think this is due to different data sources for national and subnational estimates with slightly different lags.

[sbfnk]

Also the model is fitted separately to the subnational and national data, in each case estimating the smoothness of the curve. Given that the subnational data is probably more noisy it's conceivable that the estimated trends go in different directions.