There is a lot of emphasis on the results of these branching expressions returning a value. This is a natural result of Elixir's functional nature.
Like with functions, we can more easily grok and reason about branching code if the code itself is deterministic. If it always returns a result, the branches are tractable. Our mental compilers are able to trace the code and see the result without much extra thought.
The branching examples we have done don't nest either. This arises because it is often unnecessary in functional languages to nest conditions. The syntax often lends itself to allow a single level of branching. If there's more branching, it's in another function, or implicit with pattern matching, or both; we only mentally page the branch depth that we need.
Mental easiness aside, there is also a technical reason for branch determinism in Elixir.
To more easily explain this technical aspect, let's look at another, very common, runtime environment—the Java Virtual Machine...