Refer to the
following post on performing R&R studies on push/pull testers. This may be the best approach.
Your situation (using a proof load to pass/fail the product) is really an attribute R&R. Unless I am misunderstanding you. To me applying a proof load means to apply a static load equal to the requirement and evaluating whether the headrest remains in position (pass) or drops (fails). Because the headrest mechanism is relatively complex and deals with highly variable factors such as surface finish and the resulting coefficient of friction, you are experiencing large variation in the within-product performance. This will muck up the R&R results.
The Instron white paper to which I directed you, removes the highly variable product and replaces it with a less variable product (i.e., a spring with a similar spring rate).