An informal fallacy is an argument whose stated premises fail to support its proposed conclusion.
This is certainly wrong if taken as a definition, even an informal one, which it surely is. For instance, here is an argument. Joan says, "I believe that my life is good." Then David says, "Some people suffer starvation and are persecuted; this makes life bad. Therefore not all life is good." David's argument is one whose stated premises support his proposed conclusion. But he has committed an informal fallacy; this is a valid and likely sound argument, but it is not a relevant one, and lapses of relevance are informal fallacies.
The listed source is David Kelley's The Art of Reasoning (Kelley is misspelled Kelly). I don't think I have this one on my shelf, but as the fallacies chapter seems to be online, and says nothing of the sort (e.g., he recognizes that it's all the premises, stated and implicit that matter, and that there are informal fallacies in which the stated premises do support the proposed conclusion) this is probably a misreading original to Wikipedia; but looking it up online one sees, here and there, people quoting this in arguments (usually in various forums), so we are seeing a seed crystal that might, if conditions are right, eventually develop into a bit of folklore.
It continues a bit later:
In contrast to a formal fallacy, the error has to do with issues of ratiocination manifest in language used to state the propositions; the range of elements that can be symbolized by language is broader than that which the symbolism of formal logic can represent.
This is obscure at best; it seems to be saying that informal fallacies are problems of language. It is actually quite controversial whether there are things in natural language incapable of being expressed in formal logic; but it doesn't seem to be relevant to informal fallacies. The problem with informal fallacies is not that they are in natural language -- you can commit informal fallacies in the artificial languages of formal systems, too, because informal fallacies typically arise from how you try to apply arguments to the immediate situation; and everyone who accepts that natural language is a larger domain than possible formal symbolism holds that failure to conform to formal symbolism does not automatically make the reasoning fallacious.
There is then a curious distinction between 'deductive' informal fallacies and 'inductive' informal fallacies; I say curious because ignoratio elenchi falls into neither category, despite being quite common, and because this is not a standard way of dividing informal fallacies. The standard ways of dividing informal fallacies are not in general all that great, but this way of doing it is not an obvious improvement. The description of 'informal fallacies of deductive reasoning' seems to be a description of enthymemes, which are not informal fallacies at all. We then have an odd claim that arguments are inductive if their populations are too large to sample completely. 'Induction' is used in lots of ways, but this is nonstandard.
In any case, I notice that it has dead links and that nothing has happened on the Talk page in five years, so it looks like this is a dead article, floating around and misinforming people.