I don’t have FB, so I can’t see the real source behind the source linked in the OP. Nevertheless, the data coincides in the top 5 cases with the numbers provided
in this article (June 2022). This latter article indicates that the data comes from Triple A, and bears a link to the data source that is no longer valid (ends up
here - the WayBackMachine does not help in this case).
Another article from 2021 coincides in the value of some of the countries listed (those being African).
My feeling is that the info does indeed come from Triple A, but is showing data that is now updated on their website, in relation to the data in the graph/link from the OP. Their
current data takes Nigerian crypto holders, for example, to the 22,3M mark (13M on the graph shown in the OP).
If we look
at their methodology (at the bottom of the page, and pretty long):
For the latest TripleA Global Crypto Ownership data, TripleA utilized the following metrics: (i) Country Weighted Scoring, (ii) Global Weighted Scoring, (iii) Outlier Research and (iv) Primary Data Collection to obtain the most encompassing and accurate set of statistics in conjunction with the various data sources that we evaluate.
i. Country Weighted Scoring
Our ownership data is derived based on the report “The Chainalysis 2021 Geography of Cryptocurrency Report” in which each country is given a score based on 3 factors:
(1) the country on-chain cryptocurrency value received,
(2) the country on-chain retail value received, and
(3) the Peer-to-Peer exchange trade volume
Our number of users per country estimations are based on Canada’s score and a recent Central Bank of Canada report, which estimates that 5% of the Canadian population own cryptocurrencies. In order to estimated for all the countries, we calculated the correlation between the Chainalysis score (=0.196) and the Central Bank ownership = 5% and then applied the same rationale to other countries’ scores.
<…>
I’d need to understand better the rationale behind the methodology, but I’ve got a strong feeling that the extrapolation method is leading to results that are probably way off...