A change in color is a pretty big change.
Let’s consider two questions:
“Are you happy with your life?”
“Are you unhappy with your life?”
Functionally, this is the same question. Practically, these questions yield extremely different answers - even when posed to the same person. (This is a well-studied phenomenon).
This may be the point where I realize I'm the problem. You should know that my nickname among certain groups is "Spock", and I see both of those questions as objectively different and further, that they are sort of like asking me, "How do you feel?" I know the answer is supposed to be, "I feel fine." and the phrase really means the same as "Hello" but with a different required response, but it's a rather meaningless question.
We can remove the assumption that the test takers in your well-studied phenomenon are irrational if we assume that test takers intuitively or consciously do not divide life into two states, and therefore do not use binary logic. It is quite reasonable to suggest that there is a state of being "Happy" and a state of being "Unhappy", and that there are states of being which are neither markedly "Happy" or "Unhappy" but which are neither. If we answer the questions according to trianary logic, we no longer have to assume the actors or irrational because the two questions are very different:
"Are you happy with your life" => Answer in the affirmative if you are either happy, but not if are neutral or unhappy.
"Are you unhappy with your life" => Answer in the affirmative if you are unhappy, but not if you are neutral or happy.
As such, we might really expect some people to answer "No" to both questions, even if they are asked in the same circumstances. They are experiencing no state of being that they associate with either happiness or unhappiness. These people will invariably taint the results of any simple polling, even before we get in to other very real problems with that sort of methodology. (The problem with polls is that the answer you get depends entirely on how you construct them, so you can get pretty much answer you want.)
Human language is not very precise, and in order to understand it, you really have to question people regarding what they were thinking when they employed it. Polls don't do that, which is why they are so regularly inaccurate and conclusions drawn from them spurious.
Because the “color” can change how a player feels about what happens, even when what happens is functionally the same thing (lose 8 HP/spend 8 HP), the perspective change is significant.
Again, the problem here may be that I'm too Spock-like to see these two things as shift in perspective. Functionally, I have less 8 hit points and there was nothing I could have done to avoid it either way.
I won’t argue what constitutes a “real” change because that seems like a matter of opinion.
As I am using the word "real" here, I mean a change that doesn't seem like a matter of opinion. "the perspective change is significant" seems like a matter of opinion. When I talk about something being real, I mean there is an objective difference in the procedures of play - particularly one that produces outcomes unlikely or impossible in the compared procedure of play.
But I will say that the expectation/attention shift that the player narrate how they deal with the damage affects more than just color - it engages their creative impulse (the use of the imagination, in this case) and their Expression (how they see their character/avatar and how that character exists in the fictional world). That seems significant to me, even if the practical and functional operations are identical.
If it causes a real change in process, then it is real and worth describing. So for example, if Laurefindel's players more regularly narrate their actions or the results of their actions rather than just reporting numbers ("16... hit... take 11 damage") then that's a real change in process that could be measured. The two games, one without narration, and one with narration are different. Likewise if it means a shift in narrative burden from a GM to a player, and we can quantify that, then that is also real. The thing is, I'm not convinced that this shift actually causes a measurable shift in the process of play more than simply prompting a player to engage in that way, or a player assuming the right to narrate their own actions as part of their role-play. That isn't so much subjective as it is particular to an individual, and I'm not sure this is a strong enough process of play to override individual preferences because it doesn't incentivize the player for departing from their preferences of narration or not narration.