Can you answer this stats question?

[Anonymous No.10887561]

I literally have no idea of what test to use.
Apparently it's NOT an ANOVA.

[Anonymous No.10887564]

>>10887561
Did you try ANOVA?

[Anonymous No.10887570]

>>10887564

I really think it IS an ANOVA, but apparently the ANOVA is not being taught until next year!!! amd this question is an example of what will on this years exam. Which It makes no fucking sense to me!

[Anonymous No.10887574]

>>10887570
Just eyeball the numbers and decide if they look different enough it can't be that hard.

[Anonymous No.10887582]

>>10887574

It must be done using SPSS, in APA format, including charts!

[Anonymous No.10887583]

>>10887561
T-test

[Anonymous No.10887587]

>>10887583

But t-tests are for comparing TWO groups only!

[Anonymous No.10887607]

>>10887587
So did you try ANOVA?

[Anonymous No.10887620]

>>10887607
>>10887570

ffs

[Anonymous No.10887628]

>>10887620
>>10887574
Cmon anon

[Anonymous No.10887650]

>>10887628
>>10887582

ffs

[Anonymous No.10887652]

>>10887620
I sat in my room mumbling to myself every day "Pee pee poo poo" those words haunted me, consumed me, decimated me. Looking around my room at my copious shit bottles and piss jugs "NO" I said, "THERE HAS TO BE MORE", getting on my computer I pull up hentai and furiously masturbate my small penis with tears in my eyes. I'm a broken man in a broken world, "They must pay." I say to myself as I pull back my blackout curtains, "They will all pay". I pull out my hunting knife and pierce my piss bottles, I shiver as the stench of 3 year-old piss reaches my nostrils. going outside I perform my part of the beta rebellion, soaking everyone I can in the foul concoction of piss and shit.
Next time bring your homework to reddit.

[Anonymous No.10887669]

>>10887561
Well, first of all, you could just take averages to see if there's a difference.

If there is (which we can see there is)

Then take stdev of no alcohol and see if there are people that on the other groups that statistically are so far away that it has to be the product of alcohol and not chance.

Also, you should add that since non-alcohol could be worse due to placebo which makes it harder. Furtherly, there's. not many data points, so alone from that makes it hard to actually conclude much.

This is ofcourse assuming you're not allowed to use ANOVA or any other form of more advanced techniques.

[Anonymous No.10887686]

>>10887652
As the roasties scream around me in the high density metro area, I smile to myself, knowing that my essence will forever contaminate some part of their body, I will forever be with them. Chads and white knights are already in pursuit as I run down an alley, their masculine voices screaming at me "wtf bro", I'm cornered, but that's no problem, I pull my fedora over my eyes and smile. "You're not going to be leaving here alive" it told them, and turned my back, pulling down my pants, I had just eaten 4 meals of spicy mexican food that day, my diabetes and indigestion amplified the growing bomb in my asshole. Releasing the shit I used my advanced STEM degree to calculate the neccessary arc for maximum impact. They laid on the floor gagging and choking as I did my victory naruto run past them.
Next time bring your homework to reddit.

[Anonymous No.10887704]

>>10887686
Walking into the elevator my trench coat swooshes behind me and I look outside at the devastation around me. The door closes and I'm left alone, I press the button for the 100th floor, the top floor, and hear the ding as it ascends. It's almost time now, I pull out my katana and admire the ornate patterning on the blade, blessed by 1000 betas and forged in the blood of hillary clinton. The door dings and opens to a large dark room. There he is, OP, sitting at the desk of power, his feet on the top of it. I growl "You can't just do this, you're never going to leave here alive" run at him with my katana. Alas, his 4 black 6'4" lovers foil my progress, holding me down. "hmph" OP stutters out in derision "You don't know who you're dealing with".
Next time bring your homework to reddit.

[Anonymous No.10887718]

>>10887561
This:>>10887669
Just calculate the expectation value and standard deviations and see if any error intervals overlap. If not then you can assume with reasonable certainty (which afaik is some stat terminology) that the groups are significantly different.

[Anonymous No.10887723]

>>10887704
OP get up and walks forward, his eyes go wild, his posture clenches, "I have faggotry on levels you would never dream" he laughs menacingly, the lights click on as 70 naked men stand around the room perimeter, fully erect and ready. "NO" I say, my eyes wide in fear, "NO" it becomes a screech, "NOOOOOOOOOOOOO" the deafening hypersonic sound goes all around me. The windows shatter into millions, if not billions of pieces, blood leaking from the ears of every person in the room. Still screaming, I change my pitch, as it reaches the natural frequency of the building the floors begin to crack, the ceiling begins to fall. Total collapse around me, and in mid fall I experience seeming suspended animation, I think to my self, surely, there will never be a greater faggot than OP.
Next time bring your homework to reddit.

[Anonymous No.10887727]

What's the variance and mean for all groups, nigger, do you think I will be calculating them myself? ANOVA has assumptions, see if they are satisfied here.
>apparently the ANOVA is not being taught until next year
So use the method that you have been taught, genius.

[Anonymous No.10887742]

>>10887727

What method would that be then??

[Anonymous No.10887747]

>>10887718

Can you put that in simpler terms, and stop showing off!

[Anonymous No.10887783]

>>10887742
how can we fucking know you nigger what have you been taught

[Anonymous No.10887793]

>>10887747
Just take the average and the =STDEV() in excel.

Either that or you recode the data to be like, what is the probability of being within X range of obstacles hit if you don't drink. Then if someone is above the range in the other two, you mark with 1 if not 0.

Then you perform a simple binomial test (look on youtube) for how likely it was that X amount of people have the value of 1 in N observations.

[Anonymous No.10887948]

>>10887793

I told you it must be in SPSS

ffs

[Anonymous No.10887949]

>>10887561
3 two-proportion z tests

the real answer is anova but if it's first year then this is all you can do.

[Anonymous No.10887955]

>>10887587
do three t-tests you dummy
>placebo vs low
>placebo vs high
>low vs high
I see no reason to do all three at once

[Anonymous No.10887956]

>>10887949
oh wait these aren't proprotions. um, just test for statistical difference in mean; test each drinking group against placebo, then test the two drinking groups to see if they're significantly different.

>>10887955
ya do this.

[Anonymous No.10887967]

>>10887955
>>10887956

[Anonymous No.10887980]

>>10887967
interested in which statistical reasons this is
even if it happens to be beyond OPs use case

[Anonymous No.10887982]

>>10887561
Try to make a bell curve from each three group.
When you get a curve which doesn't look like a bell that's the fake data.

[Anonymous No.10887986]

>>10887982

Jesus fucking Christ

[Anonymous No.10887987]

>>10887980
>>10887967
so I guess it is because you increase your chances of making a type I-error?

[Anonymous No.10887998]

>>10887987

A type 1 error, is believing something to be true, when it isn't.

A type 2 error, is believing something isn't true, when in fact it is.

[Anonymous No.10888007]

>>10887998
t-thanks for unsolicited explanation while disregarding the actual comment
(though you did not specify what is true, like whether it is the null hypothesis or the alternative one)

[Anonymous No.10888016]

>>10887561
Number 6 drives better while drunk

[Anonymous No.10888018]

>>10887967
see
>>10887949

"illegal moves" are also horseshit

[Anonymous No.10888019]

>>10888007

I'm just feeding your ego now, for teh lulz (1990's terminology, you will never over-stand).

We always assume the null is correct! We try and disprove the null with an alternative hypothesis. This alt-hypothesis (ha) can be one or two tailed. Do you know anything about a 2-tailed alternative research hypothesis?

[Anonymous No.10888022]

The question isn't even asking you to apply any method. It's just asking why you would pick a method. You don't even have to pick the correct method, it's just asking you what you would use and why you would use it.

You are all thinking way too hard about this, really.

[Anonymous No.10888024]

>>10888022

3a. State what the independent variable is.

3b. Compute the global mean, median, and standard deviation (i.e., for all participants regardless of group).

3b. Compute the median, mean, and standard deviation for each group.

3c. What statistical approach would you use to identify which groups are statistically different from each other and why?

3d. Carry out the analysis you described in 3c.

3e. Present the results graphically using Microsoft Excel.

3f. State what conclusions you would draw from your analysis.

3f. Suggest a title for a report of this experiment.

[Anonymous No.10888025]

>>10887561
Plot these all in an Excel spreadsheet, y axis is collisions, x axis has 3 values: sober, buzzed, drunk
Compute an average line that is the average y value of all participants for each x
Then plot each participant as a different colored line. The more they point away from the average line the more abnormal they are.
If the line has a negative slope, that participant drives better while drunk and it is a public safety hazard for them to drive sober

[Anonymous No.10888028]

>>10888024
>3e. Present the results graphically using Microsoft Excel.
what the fuck sort of course is this

[Anonymous No.10888030]

>>10888024

None of these questions suggest that you have to pick the correct method for solving the problem.

[Anonymous No.10888032]

>>10888025

It MUST be done in SPSS.

Only Graphs/Charts can be produced using Excel mate

[Anonymous No.10888033]

>>10888030

Are you retarded Anon?

[Anonymous No.10888037]

>>10888030
>3d. Carry out the analysis you described in 3c.

[Anonymous No.10888040]

>>10888033

It's an intentionally ambiguous question which forces you to think a little outside the box. You are the one that needs to chill out.

[Anonymous No.10888042]

>>10888037

You can make your analysis anything you want as long as you explain why you are using it. Think more scientifically and less textbooky.

[Anonymous No.10888056]

>>10887747
Ah sorry Ill try to give you a straighter answer. Figured if you ask this kind of question you would have seen the basic stuff already.
Expectation or mean value is just the average of all your data points, so [math]\mu = \frac{\sum_{i=1}^N n_i}{N}[/math] where [math]n_i[/math] are your values for one group, so 2, 10, 6, 9 and so on and [math]N[/math] is the total number of values you have, so in your case 18 if I didnt miscount.
Standard deviation is the statistical error you can expect from your data. This is important in all kinds of ways, but in any case it gives you an interval around your average value, within which you can expect the real physical value to be. The formula for that is a bit more annoying to work with if you dont have access to something like excel or matlab, but just as straight ahead. Its defined as [math]\sigma = \sqrt{\frac{1}{N} \sum_{i=1}^N (n_i - \mu)}[/math].
Now write [math]\mu \pm \sigma[/math] for all your three columns and check if the intervals overlap. If they dont overlap at all, you have strong reason to believe that alcohol in fact does impact driving skill. If they do overlap you have to admit that either alcohol doesnt affect driving skill or that your study wasnt extensive enough, and with a bigger sample size (leading to a smaller standard deviation) the intervals wouldnt overlap, which looking at the problem is probably the case, since the values are so all over the place that the error intervals must be huge...

[Anonymous No.10888062]

>>10887582
>SPSS
>APA
take a real stats class next time

[Anonymous No.10888063]

>>10887587
so do all the pairwise comparisons with a bonferroni correction

[Anonymous No.10888105]

>>10888056

[Anonymous No.10888132]

>>10888105
And if you don't understand the very simple explanation, you're just dumb.

[Anonymous No.10888134]

>>10887948
Then just fucking do it in SPSS you dimwit, it takes 2 seconds to fucking google this shit. You absolute mongrel.

[Anonymous No.10889792]

>>10888134

Yes, but what statistical test would I use to compare 3 groups without using an ANOVA you fucking moron!

[Anonymous No.10889853]

>>10889792
>>10888063

[Anonymous No.10889874]

>>10887982

[Anonymous No.10889878]

>>10888028
Lol right? The course should be using R or Stata for this.

[Anonymous No.10889882]

>>10887561
add all the numbers together
if the sums are different, that's your answer

e z p z

[Anonymous No.10889884]

>>10889878

Guess what course I'm doing

[Anonymous No.10889889]

>>10888105
Soooo does that mean that I understand what Im talking about or do you still not get it? Are you op? Cause I really thought someone who manages to be smart enough to operate a computer and find this place and then ask a question about statistics would at the very least be able to read and understand what I wrote.

[Anonymous No.10889892]

>>10888105
>fake quote Internet meme

[Anonymous No.10889897]

>>10889889

Look. I need to run a statistical significance TEST IN spss. WHAT STATISTICAL TEST SHOULD I USE.

[Anonymous No.10889898]

>>10889897
i told you like two times already

[Anonymous No.10889903]

>>10889897
Lol Ive never heard of spss before looking it up just now? What kind of course is this where they teach you some shitty stats program but not the mathematical background to understand what youre doing?
Are you some kind of code monkey or some shit? Why are you on /sci/ in the first place? Go to /g/ if you have problems with programs or some other cs shit or whatever the fuck all this is.

I, and many other people in this thread, have told you how to solve the problem. If you do what people here tell you, to take the averages and standard deviations and compare them, and they dont give you full score, change your fucking major cause thats some stupid shit.

[Anonymous No.10889906]

>>10889898

What's it called.

Goto. >Analyze > Compare Means > ???

[Anonymous No.10889909]

>>10889898
like here i even did it in a few minutes

bonferroni-corrected p-value threshold of 0.0167 and a welch's two-sample t-test, the placebo group is not significantly different from the low group, but both the placebo group and the low group are significantly different from the high group (p-values respectively of 0.814, 0.012, and 0.007).

[Anonymous No.10889920]

>>10889909

Are you suggesting I use an ANOVA with Bonferroni Correction (Bonferroni Post Hoc Test) in SPSS

[Anonymous No.10889921]

>>10889920
no i'm suggesting you do serial pairwise t-tests with a bonferroni correction

[Anonymous No.10889930]

>>10889921
>serial pairwise t-test

So a independent samples t-test, for group 1 and group 2. Group 1 and group 3, and group 2 and group 3?

[Anonymous No.10889934]

>>10887967

What's your thoughts on this then?

[Anonymous No.10889937]

>>10889930
yep. it's a hacky, wrong way to do it, but if your course forbids you from doing the right way to do it, it should be fine

the bonferroni correction is just dividing your p-value threshold by how many comparisons you're making. it's a shitty correction because it's often too restrictive (too many type 2 errors) but if you have a limited problem it can be fine.

[Anonymous No.10889949]

>>10889937

Overstood. Thanks fren