Which is better for kinematic at 1 second recording?

A multiple set on 6 second occupations and reoccupations or a long 20 second occupation.

I realize that multiple occupations give a sort of check and personally I prefer the multiple occupations with single occupations at more "dangerous" points(like busy intersections, where 6 seconds is about 3 seconds too long).

Not sure I understand the Q because 20 sec sounds stop&go rather than kinematic to me, also if the concern is about accuracy or otherwise. Either way, here’s something to chew on:

I take a lot (like in hundreds, even thousands) of 20 seconds stop&go shots on points with known co-ordinates. I have yet to come across a case where a vector which passes the QA test yields a bum co-ordinate. This is for 2D, very little investigation into 3D but the little evidence I have seems similar to 2D.

Verdict: vector OK, co-ord spot on.

I use the term interchangaly since my Locus days.

I mean Stop & Go kinematic