Trying to figure out how Assault works

[Mowgli]

Just to make things clearer, here is how I would assume the situation works out according to the manual:
--------------------------


SITUATION
Attacker = 10 Americans, Quality A, M16 assault rifles, no negative statuses, 100% effectiveness
Defender = 10 Cubans, Quality D, AK 47 assault rifles, no negative statuses, 100% effectiveness
combat takes place in open terrain, no cover, no height difference etc
all optional rules deactivated

US COMBAT POWER:
base: 10 men * 10 (infantry assault value) * 1 (100% effectiveness)
+weapons: 10 M16s * 18 (M16 assault value)
= 280

Casualty threholds for defender: 0.12% - 1.2%
casualties -> 0.336 - 3.36; expected losses for Cubans: 0-4 men (average 2)
effectiveness loss -> 2.688 - 26.88

CUBAN COMBAT POWER
base: 10 men * 10 (infantry assault value) * 1 (100% effectiveness)
+weapons: 10 AK 47s * 19 (AK assault value)
= 290

Casualty thresholds for attacker: 0.24% - 2.4%
casualties -> 0.69 - 6.96; expected losses for US: 0-7 men (average 4)
effectiveness loss -> 3.48 - 34.8

-----------
FINAL LOSSES:
Cubans: 0-4 men (average 2) and 3-27 effectiveness
Americans: 0-7 men (average 4) and 3-34 effectiveness
------------
Regarding the condition that a defending unit needs to suffer "losses" in order to test for retreat: This might include the loss of effectiveness?

[Xerxes77]

Just to make things clearer, here is how I would assume the situation works out according to the manual:
--------------------------


SITUATION
Attacker = 10 Americans, Quality A, M16 assault rifles, no negative statuses, 100% effectiveness
Defender = 10 Cubans, Quality D, AK 47 assault rifles, no negative statuses, 100% effectiveness
combat takes place in open terrain, no cover, no height difference etc
all optional rules deactivated

US COMBAT POWER:
base: 10 men * 10 (infantry assault value) * 1 (100% effectiveness)
+weapons: 10 M16s * 18 (M16 assault value)
= 280

Casualty threholds for defender: 0.12% - 1.2%
casualties -> 0.336 - 3.36; expected losses for Cubans: 0-4 men (average 2)
effectiveness loss -> 2.688 - 26.88

CUBAN COMBAT POWER
base: 10 men * 10 (infantry assault value) * 1 (100% effectiveness)
+weapons: 10 AK 47s * 19 (AK assault value)
= 290

Casualty thresholds for attacker: 0.24% - 2.4%
casualties -> 0.69 - 6.96; expected losses for US: 0-7 men (average 4)
effectiveness loss -> 3.48 - 34.8

-----------
FINAL LOSSES:
Cubans: 0-4 men (average 2) and 3-27 effectiveness
Americans: 0-7 men (average 4) and 3-34 effectiveness
------------
Regarding the condition that a defending unit needs to suffer "losses" in order to test for retreat: This might include the loss of effectiveness?

Your calculations seem correct as far as casualties are concerned. Not sure about effectiveness though; I haven't tracked losses of effectiveness in my tests, but I think you may be using the standard calculations for fire combat? Losses to effectiveness are doubled in assaults.

Regarding the condition that a defending unit needs to suffer "losses" in order to test for retreat: This might include the loss of effectiveness?

That's an interesting thought, however that would mean that ALL defenders should always test for retreat as all attacks from a standard squad produce a non-zero minimum loss of effectiveness.