There is an insidious thing with non-contiguous catrograms: the size of the regions is calculated based on their original size and you should not compare regions areas to each other.
Take a look at the distribution of younger population the Luninyets region's area looks clearly larger than the Zhabinka region, while both of them have almost the same value (19.4%). That's because the area of Luninyets district (2,669.8 km2) is four times larger than the area of Zhabinka (686.3 km2).
The visualization method may not be very efficient for dataset with a large range of values and outliers.
Like in this case demonstrating population size for Belarusian regions. Since the population of Mensk and the Mensk district is a quarter of the country's population, it is very difficult to see the details for the remaining regions.
It can also be challenging to use this method with areas that are concave or contain holes due to possible overlap.
In particular, to depict Belarusian districts non-contiguous cartogram method seems not an ideal solution. At this administrative level in Belarus, together with districts, there are a number of cities of regional subordination (marked red here), which are taken into account in the reports separately.
For this specific case I've aggregated all values to the districts level to build cartograms without issues. The practical value of such visualization is questionable, as is visualization without aggregation.
