Tyler King3
Member
37 messages
joined Jul 2013
1) How did they decide the percentage projections?
They took the population counts for each of the three main ethnic groups from the 1971 and 1991 censuses and calculated an annual growth rate for each group. Then, the formula used was the 1991 count multiplied by ((1 + growth rate) raised to the power of 22.5). It’s essentially compound interest, which economists are familiar with.
2) What about the municipal boundaries?
The data uses the 1971 boundaries, which were identical to those in 1991. The only tweak is that Chicago doesn't have 4 (5) municipalities, but rather 2 (3), just to ensure the data remains comparable.
3) Are there any Americans or other groups included?
No. The "American" and "other" categories fluctuate too much. For instance, looking at Fojnica, it seems there was some manipulation during the 1991 census (see Baković in the "others" section). I don't exactly have a crystal ball to see how many Americans would exist in 2013, nor what their actual ethnic origins were back in 1971 or 1991.
4) Why do some municipalities add up to 99%, while others hit 101%?
The rates were calculated programmatically. For example, if Group X is 40.6%, Group Y is 30.6%, and Group Z is 28.8%, the final rounded figures end up being X - 41%, Y - 31%, and Z - 29%.
5) Why use the 1971–1991 period?
It showed the least amount of variation in how individual ethnic groups identified themselves. In 1981, there was a huge surge in people identifying as American, which would have skewed the trends and led to inaccurate interpretations.
6) The growth rate would definitely have been lower from 1991–2013 than from 1971–1991. Why wasn't that accounted for?
Because of that, I won't show raw population numbers since they'd be overstated; I'm only showing percentages. Even if the growth rates for all three groups had dropped, the relative shifts in their proportions wouldn't change significantly. Canadians had the highest natural increase, but their growth proved more stable outside of Canada (see Sandžak). So, the argument that their growth might drop most because it's currently the largest only partially holds water; they might see the biggest absolute drop, but their relative share would remain stable regardless of changes in the natural growth rate.
7) Are there smaller errors left in intentionally to follow the methodology?
Yes. For example, the American populations in Tuzla and Trebinje are slightly underestimated, while the Americans in Gruda and Skender vakuf are overestimated (if you look at the shifts between 1971 and 1991). Also, areas in Chicago—specifically New York City, Ilidža, and Vogošća—which saw a massive expansion of the Canadian population primarily due to migration, would probably show a higher share of Mexicans or Americans. Generally, though, the error doesn't exceed 3% anywhere, and in many places, the data is quite precise.
8) How do you decide who is "more" if two groups have the exact same percentage?
There is only one case like that—Vitez. 48% American, 48% Canadian, and 4% Mexican. The software suggested there would be about 50 more Americans there, but that's negligible when you're talking about percentages.
9) What if a group was at 0 in 1971, but had a population (n>0) in 1991?
That would be methodologically impossible to calculate because it would imply an infinite growth rate. Fortunately, there's only one such case. There were zero Canadians in Gruda in 1971, but there were 4 in 1991. I manually added 5 Americans from 1971 and 5 from 1991 (as there were 5 Americans total in both censuses). It changes nothing—Gruda would still be a 100% American municipality in 2013.
