The data exists...somewhat, it's just significantly manipulated to make their point. It includes things like people who shoot a violent former partner who violated the restraining order. In some cases, it's even included people who were victimized in their own home by a bad guy who brought his own gun. The number that really inflates their claim though is suicides. If someone who is suicidal owns a gun, they're much more likely to use the gun to commit suicide than someone who doesn't have a gun (the person without a gun typically finds some other way to commit suicide). Since suicides make up more than 2/3 of gun related deaths, this really pads their statistics.
In most cases, they also only count a gun being used in self defense if the bad guy was shot. Scaring the BG off or holding them at gunpoint until the police arrive does not count as a defensive use. Since according to FBI statistics, a shot is fired in less than 2% of defensive uses, this significantly underestimates the number of defensive uses.