It's frustrating to listen to some of this. Everyone here knows basic biostats right?
Let's stick with science and not philosophical exercises.
If you took obese kids and stratified or otherwise controlled for the usual variables (race, SES, etc), and then looked at 'utensil type', you probably wouldn't find a meaningful association. Your spoon vs. fork vs. spork table would be fairly uninteresting if you accounted for confounders.
Now, look at gun ownership. Honestly, have a look at those papers that were posted. You will find that the researchers accounted for a lot of things, including age, substance/ETOH abuse, and neighborhood (controls were taken from the nearby area in the first NEJM article).
If you think there's a reasonable variable that would account for the incredible rise in risk seen in gun owning homes, then that would be a productive debate.
Having said that, you'd have to come up with a phenomenal confounder to explain away the OR/RR in these studies (CI no lower than 1.6!!). If you look at high quality journals (JAMA, NEJM, etc) you'll find the evidence is pretty convincing. A pubmed search of "spoon versus fork" and "obesity"? Probably not as well supported