Dividend adjustment wouldn't really matter in this study. The price discount at dividend ex date should match the price increase post earnings announcements, so it's netted and disappears once you compute >=quarterly returns.
The price increase at earnings announcements is (to a first order approximation) indicative of the extent to which the company's earnings exceeded the market's expectations. It's as likely to be positive as negative. Otherwise you could buy the company's stock the day before the earnings announcement and get yourself some free money.
