I have both DAW's set side by side with 16th note snapping enabled in each. My cursor is 1/16th note before measure 353. 
 
In both Sonar and Reaper, this position is given as 33786000 samples. 
 
But if I switch the units to M:B:T, Sonar shows 352:04:720 while Reaper shows 352:04:75
 
I know it's a very small difference, but what is the reason for it? I see that Sonar shows the "T" part to 3 digits while Reaper only has 2 (is there a way to change this in Reaper?) but even so, there is no reason why Reaper couldn't show 352:04:72 to match Sonar here. Or is Sonar wrong? 

.

It looks like the ticks portion of play cursor position display is 1000ths in Reaper, giving you 0, 250, 500 and 750 for the beginning of each 16th note.  My ppq is set to 960 and changing 480 doesn't change the play cursor position display.  I can't find a setting anywhere that changes this for the transport play cursor position.
 
There is however, an option for the MIDI editor ruler:

 
I'm guessing your SONAR project is set to 960 ppq and that SONAR's positon ticks are based on ppq.  This will give 0, 240, 480 and 720 for the start of each 16th note.
 
I wouldn't say either is right or wrong.  To me, the song position ticks matching MIDI ppq is meaningless in mixed audio/midi projects.  But 1000 ticks per 1/4 doesn't make sense for midi either.
 
 

.

mister happy
^ You see where you can choose to view?:
 
Measures.beats
Measures.beats.100ths
Measures.beats.Ticks
 
100/100ths = 1 beat.
75/100ths = 0.75 of a beat
 
You can select the number of Ticks per beat in Preferences. The default is 960.
 
720/960 = 0.75 of a beat ^

Yes, I see.  I posted it
 
Note that this option is specific to the MIDI editor ruler, NOT the Transport Play Cursor Position

.

I think we're saying the same thing

It does not really matter that much. For example Studio One's internal midi resolution is way higher than both of these programs mentioned here.
 
There are only two things you need to be concerned about:
 
1    That the program you are using will quantise material perfectly if its needed.
2    That the program, when quantising is switched off in fact records very faithfully the timing you put in at the time. Either against the click or not and that includes all subtle human timing. (Studio One excels at this BTW) I am also referring to a very good player here, not an average player.
 
If your program can do both of those things you are basically set.
 
 
 
 

It matters if you want to understand the difference(s) between tools you're using

.