@badp (continuing discussion here)

"That's your argument and the onus of the proof is on you."

Unfortunately (fortunately?) that's incorrect. I'm not making any claim which needs substantiation; I'm reviewing your methodology and pointing out any potential mistakes in it. This is called the peer review process.

Your conclusion could still be the correct one, but that is not supported by your argument, because of the potential confounding factors.

However, it could also be incorrect, because (due to the confounding factors) your study does not correctly support it.

At this point you either tell us why you think your methodology already takes into account those confounding factors, or you (or someone else) needs to do a different study which does take them into account.

A much better study would be to look at the distribution of cards in peoples' backpacks; that would eliminate all of the confounding factors (as long as the site we get the information from includes ALL backpacks, not just people who have joined it)

and then do a chi-square test on that data

I've been looking for a site that lists that information, but can't find one; backpack.tf would be the normal place to go, but they don't seem to list general steam-inventory, only tf2-inventory

(Even if the site only lists the inventories of users who have joined the site, that information would still be much, much better than looking at the number of listings on the market - there are many fewer, much less serious confounding factors. The only one I can think of would be the selection bias caused by looking at only the type of users who join that type of site)

ugh tf2items.com would work too, but again it's only tf2 items -_-

@blue if you're going to make an argument I also expect you to bring supporting data. I don't need your permission to publish a paper, unfortunately. If popularity counts for that much, other factors that depend on it for a standing are worth a fraction of that; so are your arguments in your latest comment.

backpacks are a parsing nightmare that suffers from pretty much all objections raised and more - what about private backpacks, people are not equally likely to have private backpacks, you cannot get a snapshot in time of all backpacks because you can only look at one at any one time and that duplicates/ignores cards traded between players. et caetera.

All of those are more or less equally worthless, drops are assumed not depending on the player, trading under the hypothesis will be of evenly random cards, etc., so what remains is a matter of practicality.

Selection bias is always there, be it "people that uses the market", "people that uses backpack.tf", "people that have public steam backpacks", "people around you who like skyrim and waste a ton of money in their entertainment"

it doesn't matter because all those groups of people, without proof to the contrary, all get served by the same drop system

If cards are even the cards in the market will be even

It's just a matter of picking the most extensive and easy data source and the steam market is it by a light year

It's public, it's a time-consistent snapshot for sets of 10 or less cards, it's not meaningfully biased

and if it is it is only in ways that make a starting uneven distribution more uneven, but cannot create the unevenness in the first place

really the best objection I can think of is people hoarding 50 of a card to take a screenshot of them and post it on reddit. Too bad it's still a popularity-based effect and thus only part of that very small correlation effect

When the total correlation of an effect is that low you can never hope to explain a 26% difference from there

It has to be a different factor altogether that doesn't tie in with essentially any kind of cool factor

such as - you guessed it - there being 25% or so more FM cards than skyrim cards to begin with

May someone point a me thread, or give a simple explanation (if it is possible to do that) of how to use the "user feed" lower left corner orange button? I know SE users use it to follow other users, but I have never did that before and I am clueless about to start using this. Thank's.

correction: lower right corner orange button

@badp what you've just said is all complete nonsense

@BlueRaja-DannyPflughoeft We'll have to agree to disagree.

Of course the steam market is meaningfully biased, it's a market. The incredible number of hidden variables basically guarantees it will be meaningfully biased.

That's why economics is such a complicated subject

And @whuber verifying your mathematics argument does not your argument is correct. I agree that you applied the chi-square test correct, but you are mis-interpreting its results - you cannot jump from "there are an uneven number of goods being sold" to "there is an uneven amount of goods available to begin with"

does not mean*

There are, again, just too many confounding factors to allow us to make that jump.

That would be like going to a (or multiple) grocery store and noticing there are more apples than corn, and coming to the false conclusion that more apples are grown than corn.

(or it could be true, who knows - but the point is, you can't come to that conclusion using that data set, no matter what math you apply, because of all the possible confounding variables)