A handful of people have such short attention spans when reading that they would prefer to watch a video clip of the same information. But most people who have a short attention span would prefer to watch a video clip of a dog on a skateboard wearing sunglasses, or some such stupid thing.
If it's done really, really well, a video clip can really enhance the content on a site. How often do you reckon it's done really, really well? 10%? 5%? No way. I reckon less than 1% of sites with video on are actually benefiting from it.
Hardly anyone who is actually interested in your content will want to watch a video, unless the video conveys information that can't readily be transcribed in text. If it's just a video of someone reading a script, maybe with fancy-pants backgrounds, that's no good. Why? Because it usually takes a lot longer to watch the video, the visual and audio reproduction is usually pretty poor making it harder to understand, you can't as easily skip ahead or go back over something, you can't copy and paste it into an email or a research paper ... the list of downsides to online video is looooong.
It was true when Nielsen said it in 2005, and it's still true now: Talking head video is boring.