The 1DollarScan website says they do the OCR for you. They also claim that the result is searchable, which of course it couldn't be if they didn't OCR it.
I may still want to OCR it myself for more control over the process (particularly in identifying and correcting OCR errors). In my experience, there's a difference in expectation of quality between OCR to make a PDF searchable and OCR to generate a standalone text file.
PDF is really a display format, whereas epub/mobi are more HTMLish in that they are not quite so specific in how they want things displayed, and thus can be flowed into different screen sizes pretty easily.