That is exactly the problem. They give the impression they're writing about startups; they may even believe it themselves; but this data is based on a random c

That is exactly the problem. They give the impression they're writing about startups; they may even believe it themselves; but this data is based on a random cross-section of new small businesses. They explicitly mention "coffee shops" as an example.