Just to clarify, that specific tool is not measuring latency to connect to the actual Droplets running on DO, just the response times of the DNS infra used, which I would assume that in your case you are using DO’s Domains for the specific cases you are mentioning.
In any case, the response time will vary according to the place where the request is coming from and the average numbers just mix them all up, which is not ideal from a methodology perspective.
DNS requests will have a better response time if there are DNS servers or proxies close the points of origin of the test, and we need to improve on that aspect especially for regions on the eastern hemisphere.
However, in many cases this is a relatively small issue on regular web performance, since the responses from the DNS are cached and the majority of the requests will skip the DNS resolution. At this point, latency to the actual Droplets become critical, and that depends on the proximity to the client machine.
For the latency to the Droplets, I would recommend using our speed tests, so you can have a good measurement for deployments on different regions: http://speedtest-nyc1.digitalocean.com/
I hope I have cleared your question.