I wonder of the origins of some of those numbers, especially the first one - the "33% in 1808".

Quick googling brings me unfortunatly only to wikipedia (bear with me for a while) which is full of it like:

According to the Italian linguist Matteo Bartoli, by the end of the Venetian rule, 33% of the Dalmatian population was Venetian-speaking.

or

The Italian speakers constituted (according to the Italian linguist Bartoli) nearly one third of Dalmatians in the second half of the 18th century

or

the Italian (Venetian speaking) population constituted (according to the Italian linguist Bartoli) nearly one third of Dalmatians in the second half of the 18th century.

If i traced the source, it all comes from a single source, Matteo Bartoli and (i assume) his 1919 work Le parlate italiane della Venezia Giulia e della Dalmazia : lettera glottologica di M.G. Bartoli a un collega transalpino. I can't access it nor read Italian but I am very curious what is in it. I would definetly scrutinize this text and look at methods and sources for such an estimate. I would also like to see clarification of some things, like does "Venetian speaking" mean only people who have Italian (Venetian) as their mother tongue, or also ones who use it as their second language? Are we sure if those are ethical italians, and not some slavs who just use Italian as a language? What is the geographical distribution of people who are the source? There is a large difference in asking urban or rural population.

I also suspect strongly, that Mr. Bartoli included population of Istria, and not just Dalmatia in his estimate. The confusion persists till today, with many people not realizing that geographically, historically and politically Istria does not count as Dalmatia. I also assume the sources for numbers like 20% for 1848 also stem for confounding the areas of Dalmatia and Istra.

Another reason I am suspicious is this 1855 map of ethnographical distribution of Austrian Empire according to 1851 census. If you zoom in the lower right corner, you can see the ethnical composition of regions. For Dalmatia specifically it gives numbers of 378,676 Croatians (which strangely seem to omit the Serbs, presumably they are counted as Croats) and 13,701 Italians, with some others brings the total to 393,715. This would bring the percentage of Italians to 3.4%, somewhere in line with this the 1890 number.

Another source for the 1850 census is this PDF where following numbers are given in a footnote:

The census of 1837 revealed that of 373, 479 people there were 296, 700 Catholics and 70,400 Orthodox. According to the census of 1850, there were 393, 715 inhabitants of whom 304, 152 were Croatians; 74,524 Serbs and 14,645 Italians. See also Marmont’s memoirs, Vol. Ill, 24-29; a detailed description of Dalmatia is on pp. 50-70.

which again would place Italians at around 3.7%. The numbers are actually the same, just the map joins together Croats and Serbs, while the PDF joins italians and albanians together

Now, I am not sure if this numbers are correct. However they do indicate that there is reason to delve deeply into the problematic of demographics of Austrian Empire. If we e.g. look at "Jstrien, Gorz, Gradiska, Triest..." region we can see a large number of Italians, who by counting parts of Istria as "Dalmatia" could tip the percentages higher, even though officially it wasn't the case. The problem wouldn't arise if all the censuses had the same data but it is possible the collectors managed to comapre one census of only Dalmatia with one where data for Dalmatia+Istria is provided. I wouldn't want to speculate further, other than to repeat, the numbers provided should be double checked before any kind of analysis would continue