> That is better that almost any other heating source

... if you don't charge generation and transmission inefficiencies against it. High efficiency natural gas heating systems are 98% efficient; compare to natural gas combined cycle plants that are ~60% efficient, plus a few percent more lost to transmission and distribution.

> with exception of heat pumps

Sure, that's the actually-efficient source of heat from electricity.

> For light bulb, i guess most energy is dissipated by IR radiation and only small part by convection.

For incandescent bulbs, this is unrealistic. The bulb is quite hot and heats the air around it and the fixture.

Moreover, of the IR, a whole lot also strikes the fixture and is conducted into the ceiling.

... if you don't charge generation and transmission inefficiencies against it.

Well yes. Statement 'efficiency of X' generally means 'useful work of X / relevant energy input into X on X boundary', not including efficiency of whole supply chain from primary energy source.

I have no idea what the percentages are but based on a lifetime of anecdotal experience esp with incandescent, I'd bet you're wrong. Being across the room from a light bulb will warm you a bit, but nothing like proximity to the bulb, especially when you're above it.

You're also being hit with more IR when you're closer. That being said, getting the same amount of light from an LED seems to only take 10% as much power, but they're also probably putting out much less IR, but LEDs still heat up a little, ...

A bare lightbulb is almost an anisotropic point source. You can put a reflector around it to target the IR in some direction. Its energy efficiency at the bulb is the same, i.e. nearly perfect. The energy delivered to target is different.

> Resistive heat has efficiency ~1

It does if you only consider the efficiency of the electricity -> heat conversion, but electricity generation itself is often not that efficient. It can be a lot more efficient to go fuel -> heat than fuel -> electricity -> heat.