Bad Math General

That source. Is that suppose to mean something?

File: pop_vs_sample.png (90KB, 2400x1203px) Image search: [Google]

90KB, 2400x1203px

>>7781183
>muh large sample maymay

With just 50 randomly chosen people from the population we get a pretty good fit. Pic related, the distributions of height of 25,000 UC Davis students.

File: nederlandse-vlag.jpg (7KB, 425x283px) Image search: [Google]

7KB, 425x283px

>>7781183
DIT IS VANAF NU EEN NEDERDRAAD

>>7781250
>pretty good

shit non-mathematicians say

>>7781250
what kind of shitty graph is that ? don't you learn to label in like 2nd grade now? every morning in my behavorial statistics class my teacher would show examples of shitty graphs and this would be one of them

>>7781300
>He thinks I'm going to take more than 20 seconds to make a shitty graph for /sci/

kek.

>>7781250
But hur dur it needs to be a census of the population of interest, or it isn't representative because magic. Even though populations change, and censuses take a lot of time and money, and are likely more inaccurate because of population change.

>>7781183
Blijf gewoon in reddit, dat is beter voor iedereen

>>7781320
OP is een fagot.

>>7781250
>pick something normally distributed
>look guys no need for large samples

also nice axis labels, you in humanities?

>>7781305
Kek, this
Autistic Mathfags btfo desu senpai

>>7781291
>let's look at the real world

shit non-mathematicians say

>>7781335
>BTFO'd so nitpicking points that don't matter

It's okay, man, I bet your math degree will help a lot when you are teaching Jamal to add while you flip burgers with him.

>>7781335
It's more than just height distributions, senpai. The full dataset gives heights and weights, so if we preform a regression:
>Population
r value = 0.503
p value = 0.0

>sample
r value = 0.435
p value = 0.001

Obviously a larger sample is preferable, but having a smaller one doesn't instantly mean the study is worth shit.

>>7781250
>pretty good
>fit

translation for anyone who cares about science or math: we grabbed random data and it looks ok i think, we even forced it to fit our preconceived notion of how it should look kinda

you see the board title? Science & Math? Fuck off, illiterate fucking humanities mongoloids

>>7781418
Not that guy but if you have ever actually done any stats you would know that margins of error can be calculated and it is possible to "fail to reject" a claim if the sample mean falls within the margins of error of the expected/desired mean meaning it warrants more study, so these papers might not be conclusive but they are indicative and by taking multiple small samples the population mean can be found (small meaning <100).

[eqn]p-\alpha\sigma \bar{x} \le \mu \le p+\alpha\sigma \bar{x} [/eqn]

Gives this error where p is the sample mean and [math]\sigma \bar{x} [/math] is the standard deviation of the sample.
[math]\alpha[/math] is the desired confidence interval level (so 1.96 for 95% iirc)
and [math]\mu[/math] is the population mean.

>>7781418
thank you

>>7781418
>I have never heard of Lilliefors test before

>>7781446
>>7781448
>complain about humanities faggots fitting shitty samples to support a claim and say they're confirming shit
>some smug poster comes and for no reason explains babby stats 101
????
It's like if I bashed a faggot saying dumb shit about quantum magic and you come and start reciting some basic result in GR

>>7781454
>Doesn't get why he's wrong
>Blames everyone else

Gentlemen, I present: the plebeian.

>>7781454
>OP claimed a sample size of 50 isnt enough for a sample to be representative
>gets butthurt when he realises it is a legitimately sized sample and nobody claimed 1 study does a proven claim make

>>7781394
>Obviously a larger sample is preferable, but having a smaller one doesn't instantly mean the study is worth shit.
No, but it does mean you can't TELL if it's shit or not.
And if you can't tell if it's reliable or not, what use is it?

>>7781475
>I have never studied stats before
if you had taken even 1 stats lesson you would know there are ways to tell if it worth shit, even if it is going to have a huge margin or error like +/-30% that margin can be calculated and if the results fall within that then they at least indicate the researchers are in the right direction, no single sample is conclusive, but they all contribute.

>>7781482
>they indicate that you're in the right direction

when you think everything indicates you're in the right direction if it doesn't prove you're wrong, even though your rigor is shit and your margins are huge, you're not doing science or math.

seriously, why the fuck are you humanities faggots in this thread anyway?

>>7781465
>>7781467

>>7781490
>Be you
>Want an estimate on global height
>"Gotta be rigorous, autism demands it"
>Decided that a sample of 10% of the population would be pretty good.
>700 million people need to be surveyed
>Takes decades and tens of millions of dollars
>Finally have an estimate
>It's now out of date
>Still have large margins because it's only 10% of the population

It's like you've no idea what you're talking about.

>>7781507
>Be you
>Want an estimate on global height
>"Can't be rigorous, anything will do"
>Decide that a sample of 50 people will be pretty good
>Takes 5 minutes
>Have a 100% margin of error
>Look, what we thought fits the estimate! We're right!
>"Headline: Study says, science confirms..."

It's like you've no idea what you're talking about.

>>7781515
>Be me
>Want global estimate on height
>Do research (n=50)
>Check shit is within margin of error
>Publish
>Sometime later
>Some other researcher preforms a meta-analysis
>Very small errors
>Moderately sized sample (n= a few thousand)
>Publish