I am not sure I understand the refill example, and I do not have any empirical study in mind, but here is a classical fictitious example drawn from Mas-Collel, Winsthon and Green, Microeconomics, which might be relevant to your question. The example is slightly fetched but I think it's good to get the idea.
Suppose you want to choose a color to paint your bathroom. You go to two shops who have the same colors available but walk you through a different choice process.
Both shops have available colors number 1 and 20 in the following grid.
Here comes the difference.
The seller ask you whether you prefer color 1 to color 20. You are rather a green person so you say $20\succ 1$, meaning you prefer 20 to 1. As a consequence, you chose color 20 for your bathroom.
The seller still has only 1 and 20 available but walks you through a different process to elicit your preferences. He first asks you whether you prefer 1 to 2. Really, the difference is so small that you cannot see it, so you say $1 \sim 2$, meaning that you are indifferent between 1 and 2.
Then he asks you whether you prefer 2 to 3, and again you answer $2 \sim 3$.
As the process keeps unfolding in the same way, you eventually get to the point where your preferences are $1 \sim 2 \sim 3 \sim \dots \sim 19 \sim 20$. So if you are consistent (i.e. your preferences are "transitive"), you should admit that $1\sim 20$, that is you are indifferent between 1 and 20 and you might end up letting the seller chose for you.
Now this argument is usually used to demonstrate that people preferences are not always transitive : $1\sim 2$ and $2\sim 3$ does not always imply $1 \sim 3$. In effect, in the shop example it seem to make no sense to claim that you are really indifferent between 1 and 20. In fact, you really prefer green but you have been tricked by the progressiveness of the preference elicitation process.
Although this is convincing in the shop example, you might consider other situations in which you would be tricked so subtly that you would not change you preference at the end of the choice process. Suppose I come in you bedroom every month and repaint it following the same pattern (from color 20 to 1), without you knowing it. Chances are you will not realize it, get use to the new color and end up claiming that you prefer the final color, although at the beginning of the repainting process you strictly preferred green to yellow.
I guess this is a common strategy in marketing. Not so long ago, Mc Donalds steadily changed the background of their logo from red to green. They could have done this all at once, in a week say. But they did not. They changed the background color of some restaurant first, and others later. My impression is that they purposely spread the repainting process in order to get people used to the new green color. The likely outcome of such spreading is that some people who were initially disgusted by the new green color steadily became to wonder how they ever liked the logo with the red background.