If we assume 16K/27K is Trump approval, then nearly 60% approval is not bad! But surveying representative populations is a bit more complex than 'are they being truthful'. Whether or not the sampled group is representative of actual voting population is the key question. Ideally, all Yahoo users would = actual 2020 voters for this poll to be accurate. But you also have 750K+ views... what are they, undecided? Independent? Nonvoters?

This is how pollsters work, they take numbers from 1000 random people. Only here we got 27000 and anonymous. So people didn't suppress their views.