Gain measurements smaller than simulation 1.5dB

Hi, I have a question about HFSS patch antenna array simulation.

For a normal 16 elements patch array I simulate in HFSS, the gain is about 1.5dB lower than that in a anechoic chamber measurement. What kinds of factors would possibly drop the gain? For example, is it due to feed loss, chip resistors loss in the feeding network (Wilkinson divider), connector loss.. and so on?

Thanks.

Is the simulated and measured Return Loss vs Frequency accurate ?