In skeptics fashion, here's a peer-reviewed scientific literature article.
Type size, as reported in Table 8, was found to be a significant main effect. (F (2,104) = 6.582, p = 0.002). This result leads to the rejection of null hypothesis H3 and forces a conclusion that size does matter.
Please avoid microscopic fonts (e.g. ≤12px)
There should be no font with less than 12px on screen as a default. It's too small as defined by the study above, for example.
Examine the following copy (for example: "votes", "answer", "views", "active", "featured" and many more):
Notice the font size (e.g. "Inspect element" on Chrome)
All copy text should have a size of at least 13px.