Does this line look like a good fit?
The data points are the attractiveness ratings of the voices of women aged 18-55. The data is supposed to show that women's voices are at their most attractive at about 26 but the data points are all over the place. We also don't know how voices under 18 would be rated and this could change the best fit line a lot.
>>7899232
>The data is supposed to show
no the data isn't supposed to show anything.
you're trying to make it show something, and you can always transform the data to make it show what you want.
I really hope you never publish anything
>>7899232
It looks reasonable, you could do Graphical Residual Analysis to check, but I think that only works for linear fits, if that's the case would it matter much if you did a semi-log or log-log plot?
what about you make the diagram start at 18 instead of 10
>>7899246
why?
>>7899241
>no the data isn't supposed to show anything.
I'm saying that the researchers who carried out the study claim that the data shows a peak at 26.
>>7899247
because there's no data below 18...
>>7899248
it clearly doesn't. Looks more like 18
>>7899248
Here's the study.
https://www.researchgate.net/publication/235922464_Women's_faces_and_voices_signal_fecundity_in_industrial_and_forager_populations
>>7899241
This.
OP, I could just as easily draw a straight line with a negative slope through those points that'd look every bit as good of a fit.
>>7899254
With 52 data points i got the 60 year old peaking at 4.5 attractivness *bleaches mind*
Looking for honest answers here... Do you guys think I'm cool?
>>7899232
>the data points are all over the place
I wonder what that means? It's almost as if your study found no statistical significance between age of voice and attraction. This is what you get when you study clinical psychology and pretend to be a scientist.
>>7899721
it's obviously going down with age after a certain point faggot
for 1, never say things like "the data is supposed to look like this", no no no. Either your measurements are bad or your theory is wrong, the data does w/e the fuck it wants. Since it is an attractiveness rating I would just simply state you can see a bumb around the 20 to 38 range but that no conclusion can be drawn from the data. The fit line is pretty meaningless as well desu, a fit is supposed to represent the data, which is not the case here. but anyways im no expert.
>>7899232
>Does this line look like a good fit?
As good as any.
By quickly glancing over this chart I can tell you would need a much, much higher density of data points for the best curve to "present itself" if you will.
>>7899676
heck yeah!
>>7899248
The peak is at 35-37. Are you blind?
>>7899724
Keep telling yourself that sweet heart
>>7899232
>extrapolating the curve past 18 with no data there and more data above the line
Trash desu
What's the Pearson r for your data
I'd say go for a straight line, negative grad, from 20 down to 50
Pretty meh correlation though, get more data
kind of related question guys, I'll use OP's pic as an example.
assuming someone really wanted to fit parabolas on that data, is it acceptable to fit two parabolas instead to form an envelope and give BOTH parabolas as representation of the data, as well as some correlation coefficient(s)?
Is this used anywhere in science/does it have a name?
>>7899232
>this line
>degree-two polynomial curve
fgt pls
>>7899778
Thats not the peak of the trend line, its just the maximum of the entire data set
>spline fit
into the trash
>>7899838
It's not acceptable to fit a parabola on those data in the first place
This might be a better curve.
>people are actually getting grants to do these researchs and publish them
fucking statistics
>>7900153
jesus fuck please answer the question or don't say anything, it's just an example.
And you can choose to fit anything if you want as long as you give the corresponding correlation coefficient.
Anyone to actually answer my question?
>>7899232
Why are people so gung ho to fit curves to data. The purpose of fitting a curve to data is to generate something with some minimum predicative value.
There don't appear to be enough data points in this plot to make anything curve reliable or useful.
At minimum, this plot should include plots with input parameters +-one standard deviation to show possible variance to illustrate possible ranges.
>>7900299
i have a stats degree and I agree of this fit
>>7900273
hey guys!
I made a cool diagram to determine the most attractive random number and my fit shows, that the best numbers are between 30 and 35!
So if you want to create a sexy number, I suggest you take something in that interval
>>7900331
what kind of idiot would put an extra curve that's worst than the best curve?
>>7899232
Publish or Perish 3: The Last Stand
>>7899232
You know theres ways to analyze the fit.
Also your sample size looks criminally tiny.
>>7900299
Kek
>>7900411
>>7900338
It could generate clicks as plain click bait and thus generate more interest which turns into massive funding and scientific success at last.
>>7900273
>collapse at 38
In some warmer countries women age brutally in their 30's. This could be a sciencey way of proving this (ahem) that the voice warns men to stay away no matter how much makeup is applied with trovels. Clearly our survival depends on this.
>>7899838
>is it acceptable to fit two parabolas instead
If you can relate this to distinct regions, then yes.
In this case you have 3 regions: childhood, fertile women and infertile women.
Repeat the data recording and relate this to the fertility state of each woman including to what phase in her period she is in to get something meaningful.
Nature provides us with an enormous range of indicators to determine if another person is a suitable mate for having our offspring. At the same time women are controlled by instincts that make them want to appear attractive at any age, which explains why cosmetics is such a gigantic money maker. This simply is a natter between detection and not wanting to be detected.
