In short, an NBA coach who has never been an NBA player will tend to have a worse W-L record than a coach who has played in the NBA. And NBA allstars who become coaches have the best coaching records. From the paper: "For the typical team, the difference between having a coach who never played NBA basketball, and one who himself played many years of allstar basketball, is approximately six extra places up the league table."
Which is interesting, but yes, let's be careful when drawing conclusions and generalizations. To that end the paper said "our findings may be relevant to a range of high-performance workplaces where the employees are experts." The researchers merely found some data that suggested a coach's past performance as a player was positively correlated to his W-L record as a head coach. The paper did not conclude that interpersonal skills were irrelevant to a coach's record, nor did it conclude that interpersonal skills mattered less than playing skills. The researchers did not interview any people or even define terms like leadership, interpersonal skills or expert knowledge. They did milk 8 seasons' worth of Sporting News though, their primary source.