The mechanism is asymmetric scrutiny. Evidence that supports your position gets accepted at a glance; evidence against it gets interrogated until a flaw appears, and a flaw can always be found.
The motive is usually the open position. A long makes bullish data feel solid and bearish data feel like noise, and the effect grows with size. Size is the single best predictor of how badly your analysis is compromised.
One habit helps more than any other: write the bear case for your long, in full, before you enter. If you cannot write a strong one, you do not understand the trade well enough to size it.
Related: confirmation-bias, belief-perseverance, cherry-picking, narrative-fallacy