There are some problems with how signal strength is calculated and displayed. Often the signal strength is 1 higher visually or 1 lower in reality. Near the end of the redstone line the signal strength can visually jump from 2 to 0, despite the signal strength being differently mechanically.

This only happens when the redstone line runs past both the block the power source is powering and directly next to the power source itself, and the power source is powering the block from the side. Note that this is very directional and can only decrease the power level by one.

Steps to reproduce:

1. Attach a lever to the side of a solid block and power it.
2. Run redstone dust underneath the block and the lever.
3. Place a comparator next to the dust underneath the solid block
4. Note that the signal strength from the comparator is 14, despite that dust visually having a singal strength of 15 (note: I used a resource pack for this).
5. Run a redstone line from that dust and note that most of the signal strengths (visually) are one higher than they should be, although it jumps from 2 to 0 at the end.
6. Run comparators from that redstone line. You'll find the signal strengths from those comparators go something like this: [14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 1, 0]
   

You could also use the test world provided (Issues with signal strength.mcworld). It's probably easier.

Expected results:

I would expect the signal strength to match visually and mechanically, as well as actually being correct.

Actual results:

The signal strength is 1 higher visually and 1 lower mechanically (although if you extend the redstone line the other way that half works fine).

Visually: 15, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 0
Mechanically: 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 1, 0

Notes:

This seems to be related to MCPE-81987

Also, I would highly recommend using a resource pack which numbers the redstone dust based on signal strength while investigating this issue (at least when fixing it visually). There's one in my test world (hopefully).