Accessibility in data visualisation is only recently getting the attention it deserves. But it is such a very broad topic, it deserves its own separate training. Below are some specific aspects of accessibility to take into account when designing and publishing data visualisations. All of them are easy to check.
Apart from the concepts and guidelines regarding typography in the Typography and the design of text elements, the use of tyopography can be optimised for people suffering from visual impairments or dyslexia.
A first basic rule is to avoid any unusual or artistic font, like handwritten fonts. They are not designed for legibility and can easily generate reading difficulties.
Fonts should also have more than one character case, so avoid uppercase only fonts for example. Uppercase text is harder to read than normal cased text, and text decoration like italics also reduce legibility.
Some fonts (especially sans serif ones) use basically indistinguishable characters for the number 1, normal cased l and capital I. They do this for stylistical reasons, but this obviously creates a lower legibility. Using a font that has distinctively different characters for these characters is advised.
Source: Google Fonts
Young children, but also some adults, mirror the letter combinations d-b and p-q. Using a font that does not use mirroring characters for these combinations can help them reading these correctly.
Roboto uses mirrored characters, while Roboto Serif does not. Source: Google Fonts
Other problematic letter combinations for people with visual impairments are o, c, e and a. These can be hard to distinguish in fonts that have small openings (called apertures) of the c character, which can appear to be closed and ressemble an o, while the e and the a can also appear to be closed and ressemble the number 8.
Source Sans Pro has wider appertures than Oswald. People with visual impairments might have difficulties with the o, c, e and a in Oswald. Source: Google Fonts
Reducing letter spacing is a technique to fit more text into a small space, but comes at a legibility cost and should therefore be used carefully. But fonts have their own characteristic letter spacing, with some using a wider spacing then others.
For print, a font size of 14 points is a good default. Browsers have a default font size of 16 pixels, which is a good starting value for online text. Less important text can be smaller, but should not be smaller than 12 pixels.
Mix upper and lower case in a senible way. All lowercase or all uppercase are less readable than mixed case.
Use underlined text only for links.
The modules Colour: the basics and Colour: use in data visualisation cover colour blindness and colour contrast exentsively. The baseline here is to always guarantee sufficient contrast between text and its background, and to avoid encoding of data into colours that are not distinguishable from each other by people suffering from any kind of daltonism. Many tools for testing contrast and identifying issues with colourblindness are available.
One very specific aspect of data visualsastion is that it almost always mixes graphic elements with text. In many cases the graphics are published as bitmap images (JPG or PNG, see the File formats, dimensions and units module), and the text is “baked into the graphic”: the text characters are part of the image, and are not accessible.