"The Harvard/Northeastern study is based on a survey of nearly 4,000 Americans conducted online in 2015 by a market research company, GfK, with a nationally representative panel of opt-in participants who are compensated to complete surveys on a variety of issues."
So, total garbage, then? Most gun owners I know keep the fact that they own guns private. They will not tell the truth when telephone pollsters call ("Oh, no... no guns here!") and they certainly aren't going to be participating in some online "opt-in" survey.
Change to ""50,000 (carefully) randomly selected people in the US, for whom you have strong evidence that they won't lie about their gun ownership" and you're going to get nearer the truth. I guess that falls under "carefully", but it needs to be made explicit.
Basically, any study or poll on a controversial topic should be considered of little value until positively proven otherwise. If one point of view (or whatever) has been widely demonized (e.g., gun owners, Trump voters), you're almost certain to get garbage results (e.g., "Hillary Clinton has a 90% probability of winning the election").
So, total garbage, then? Most gun owners I know keep the fact that they own guns private. They will not tell the truth when telephone pollsters call ("Oh, no... no guns here!") and they certainly aren't going to be participating in some online "opt-in" survey.