[deltaonefour]

>Going to a non-selective school isn't special, and neither is a selective one. For one, they can have the same difficulty and SELECTIVITY DOESN'T CONFER DIFFICULTY.

It does. Because selectivity effects the curve. If all schools had the same curve and the same curriculum and same everything except selectivity then, selectivity influences the curve. I've said it a thousand times. If you disagree you have to counter the logic of the curve which you haven't.

I'm also well aware that choice of a curve is a random variable. You've stated it multiple times. Therefore if it's a random variable then hold it the same when comparing mathematical models. Assume all random variables are the same and only adjust the relevant variable which in this case is selectivity. In this case selectivity is causal to difficulty.

>YOU pulled the figures like 25% and 50% out of a hat, which is fine because we acknowledged they weren't hard numbers. Don't be a hypocrite.

The context around those numbers make it clear that they're just really rough estimates. Additionally those numbers were not used in calculations, I simply stated that the numbers I found are inn-accurate because of several factors and gave a really rough estimate on what i thought the numbers could be (50% and 25%). The subsequent conclusion and calculation were based on the lower more inaccurate numbers which STILL show that selectivity is causal to difficulty.

99.9% is not a rough estimate. It's clearly a really exact number with a decimal point.

>Personally I think you just have a chip on your shoulder thinking a selective program is generally more difficult than a non-selective program. Elitist arrogance.

Well did I insult you in any way to the degree of calling you Elitest or arrogant? I'm just making a point. You're the one crossing the line with insults. I think you have a chip on your shoulder.

>I said could be, because you decoupled by saying now that it merely COULD be that a selectivity influences difficulty, before backtracking to play your devious and treacherous claim you were never talking about the school from which your anecdotes came.

In all of reality and all of science, especially the social sciences all causal connections are coupled with probable connections. Nothing is connected with pure logic. Logic only exists in math and logic games. If I said all men are stronger than women. Clearly the generality is true, but obviously there are exceptions and it is unnecessary to comment about those exceptions. But nonetheless, nobody talks in terms of probabilities. In common human parlance we communicate using absolute statements in reference to topics that are in truth generalities and we assume that the other party has the intelligence to know about the exceptions to the generalities.

I simply brought up the probability because I explicitly wanted you to know that YOUR school might be an exception. But by probability it is most likely not an exception. That is all.

>Edit: here's a comment from a student at one non-selective school

I'm more interested in the comment from a student who went to both a very selective school and a school that's not selective. Many students in an "easy" school think it's hard only because they don't have a point of comparison.

>This is how it tends to work in non-selective schools. They hold students to the 'selective' school standard freshman/sophomore year until by junior year you're effectively in a cohort that looks like the cohort at the selective school. The difficulty is still there, although it's an independent variable from the selectivity.

Again. The curve is not an independent variable from selectivity. Therefore selectivity is causal to difficulty. This is a literal numeric connection. I've said this multiple times. Anecdotal and qualitative evidence is worth considering but this causal quantitative connection far stronger.

Your best bet is to find someone who did undergrad stuff at both MIT and Georgia Tech. That persons anecdotal experience is far more accurate.

[notch656a]

>If all schools had the same curve and the same curriculum and same everything except selectivity then, selectivity influences the curve. I've said it a thousand times. If you disagree you have to counter the logic of the curve which you haven't.

I already have. Different 'curves' for different school. Or even uniform objective standards so that cohort performance is irrelevant to grade outcome. You can create a 'virtual' objective standard by creating curves that are compensated -- i.e. the 'selective' school has a curve where the median score receives a C while the non-selective is curved such that like-performer parity is present and the median receives an F. In practice this is how the non-selective school manages to give F, D, or incomplete/drop-out to the majority of students in a course.

Again the school doesn't give a shit when people complain about this, there is a suicide by a 3rd world national who fails out almost every few years because the school will allow the parent to spend their last dime sending their kid to college and the kid ends up killing himself out of honor when they find out the non-selective university will actually take anyone and happily fail out most of them. The media doesn't give a shit, the judge doesn't give a shit, and the jury of your peers (in this city, basically working-class midwesterners who very well may relish putting a snobby complaining rich kid in his place) don't give a shit. The jury will probably laugh as they go home and have a toast to their wife that their competition for rental housing are falling after eliminating the competition.

>You're the one crossing the line with insults. I think you have a chip on your shoulder.

How original, you'd make a nice parrot.

>99.9% is not a rough estimate. It's clearly a really exact number with a decimal point.

You indicated that selectivity doesn't always mean that the school is more difficult. You failed to designated a bound other than your school wasn't included amongst those whom your were claiming were more difficult on basis of selectivity (in fact you go so far as to say you SPECIFICALLY did not include your school), so threw out a number that COULD be the case based on your own argument.

>Your best bet is to find someone who did undergrad stuff at both MIT and Georgia Tech. That persons anecdotal experience is far more accurate.

Pretty much agree except I would do UC Berkeley (#3) vs Purdue (#4) since those are the most near peer selective vs non-selective I could find on 2023 engineering top 10 rankings. MIT is #1 vs GIT at tie #7 is a bit wider. As number 1 MIT is probably a class of its own vs 2-10, since they have 'winner-take-all' advantage in anything where only number one will do.

[deltaonefour]

The argument below:

>I already have. Different 'curves' for different school. Or even uniform objective standards so that cohort performance is irrelevant to grade outcome. You can create a 'virtual' objective standard by creating curves that are compensated -- i.e. the 'selective' school has a curve where the median score receives a C while the non-selective is curved such that like-performer parity is present and the median receives an F. In practice this is how the non-selective school manages to give F, D, or incomplete/drop-out to the majority of students in a course.

Was already countered with this:

>I'm also well aware that choice of a curve is a random variable. You've stated it multiple times. Therefore if it's a random variable then hold it the same when comparing mathematical models. Assume all random variables are the same and only adjust the relevant variable which in this case is selectivity. In this case selectivity is causal to difficulty.

Please respond to that rather than regurgitate an argument I already addressed.

>You indicated that selectivity doesn't always mean that the school is more difficult.

The curve argument addresses this. You'll need to address my counter argument to your argument against the curve.

>Pretty much agree except I would do UC Berkeley (#3) vs Purdue (#4) since those are the most near peer selective vs non-selective I could find on 2023 engineering top 10 rankings. MIT is #1 vs GIT at tie #7 is a bit wider. As number 1 MIT is probably a class of its own vs 2-10, since they have 'winner-take-all' advantage in anything where only number one will do.

Sure. Find someone. Idc if it's MIT vs GIT or UCB versus Purdue. Also another caveat to keep in mind... rankings aren't exactly a good indicator for difficulty as we aren't even sure about the criteria used to determine the ranking.

To really strengthen your side, multiple people from multiple schools should be used. But one person is enough for me to at least speculate on an alternative conclusion.

Until then, selectivity on average is causal to difficulty.

[notch656a]

The choice of a curve isn't a random variable. The choice of a curve is a human selected, non-random variable. Quite probably, and in practice most definitely, I have seen it adjusted so that in a 'top-performer' class the median is a 'C' while in a 'anyone-with-a-pulse' the median is more like 'D/F/dropout/incomplete'.

This isn't random. With higher selectivity, in practice, it seems likely and at very least not impossible that the variable is adjusted to make grades elevated vs median cohort member so that difficulty of passing is constant across selective vs non-selective. IDK how you could possibly assert the curve bias is random. And you've still completely ignored objective grading systems, which I have indeed seen used in core engineering classes to ensure cohort performance is completely irrelevant.

>Until then, selectivity on average is causal to difficulty.

You haven't proven this. It's a totally unsupported claim. There's zero evidence to indicate mere selectivity confers difficulty.

[deltaonefour]

>The choice of a curve isn't a random variable. The choice of a curve is a human selected, non-random variable.

You're misunderstanding statistics. If you have 5000 humans and let them each choose a choice for a curve. Then you select a random human to see what choice he or he/she chose, that choice might as well be a random variable. If you try to find the mode of which curving methodology the person chose you'll find one that's the majority. That's a good value to freeze the choice around.

https://www.wikiwand.com/en/Probability_distribution

For the purposes of this argument ^^ the above is the definition you should center your research around. Deeper meanings into "random variable" become too pedantic.

>IDK how you could possibly assert the curve bias is random.

Random variables don't technically exist unless you study quantum mechanics. It's not even a provable notion whether they actually do exist. But typically in statistics you can treat all measurements as taking a sample of a random variable. When you shuffle a deck of cards, when you draw from a raffle... or when a random school chooses what curve methodology it uses. The language I'm using here "random variable" might have thrown you off, but it's common parlance in science and statistics. Try re-reading my previous response with this knowledge in mind.

>You haven't proven this. It's a totally unsupported claim. There's zero evidence to indicate mere selectivity confers difficulty.

Again, please understand the concept of a random variable, then re-read the response again. Then you will see, the evidence is quite strong.

I'll restate it here for clarity. Read carefully. We can test if selectivity influences difficulty in a mathematical model of schools. First treat all parameters that are part of a model of a school as random. Because those parameters are random we can just freeze all those values at some average and make it the same for everything, because it's not something we're trying to measure... and likely these parameters cluster around some average anyway.

SO, as a result, we have a bunch of mathematical models of schools that are all EXACTLY the same. Each school model takes in an ass load of parameters as inputs, and outputs some number that measures "difficulty" as an output. Since all models are identical all input parameters yield the same Output.

To test "selectivity" we just take two identical models and fiddle with the "selectivity" input parameter. If the "difficulty" output changes when you adjust "selectivity" then you have proven selectivity is causal to difficulty for that model.

I don't have to know exactly what this mathematical model is. Just pieces of it. I know grading policy influences difficulty, I know the curve influences grading policy, and I know the amount of smart people in the school influences the curve, and I know selectivity influences the amount of smart people in the school.

Thus I know when you fiddle with that selectivity parameter, difficulty increases.

Your argument is saying that you can't ignore the other input parameters that the choice of the curve matters. I agree. It matters, but we're not trying to measure that. We want to throw it out of the measurement equation and see if selectivity influences difficulty. So we freeze the curve methodology at some arbitrary option and ignore it. We measure selectivity so that's the only parameter we fiddle with.

Think of it like this. You have 20 switches labeled with numbers 1 - 20 and a single lightbulb. Each switch may or may not be connected to that lightbulb. I don't know, but I only care about switch 5 and I want to know if switch 5 turns on that light bulb. I don't care about switch 3, 4, 6, or 7. Only switch 5.

I test switch 5 by freezing all other switches at ON then flipping switch 5 on and off to test if it has a causal effect on the light bulb. Switch 5 in this analogy would be "selectivity" and switch 7 can be "the choice for the curve methodology"

This is the exact SAME proof I'm using to show that selectivity influences difficulty. Same concept but instead switches are the input parameters and the output parameter is the lightbulb.