Looking for ideas for building a little #Rstats function.

Let's say I asked a bunch of people if they felt older, younger, or the same age as half a dozen randomly-selected ages.

Let's say that because that's what I did.

Example: Jane is asked this question about ages 17, 19, 25, 38, 41, and 52.

Each person answers younger/same/older to each of the ages (each person also gets a different selection of ages to answer).

Ultimately, I want to use those answers to get an estimate of the age each person feels. First, however, I want to see how numerically consistent each person's answers are.

Examples: if Jane said
17: older
19: older
25: older
38: same age
41: younger
52: younger

That's numerically consistent. However, if she said

17: older
19: younger
25: older
38: same age
41: older
52: younger

that's not consistent (she can't be both younger than 19 and older than 41).

It's also consistent if she said she felt younger than all the ages or older than all of them.

Maybe because it's late or maybe because I'm slow, I'm having a hard time approaching this. I'll leave it on the back burner for a couple of days since it's late where I am and I won't have time for it tomorrow.

If any of y'all (and many of you are ridiculously more quantitatively skilled than I am) have ideas of how to approach a function for this, or -- even better -- if a solution already exists as a formula or something, lay it on me. I'll be in your debt*.

*up to but not surpassing verbal compliments and possibly $2 USD