Official STAT270 thread (1 Viewer)

[CrashOveride]

this stat270 stuff was all covered in 171 =p

 

[MaryJane]

Really? Do you do ANOVA and two-way ANOVA and regression? STAT170 was so basic. But I still hate maths.

 

[redruM]

i'm pretty sure they do all of our course in 171, but maybe not much detail with regression. (maybe?)

 

[redruM]

got results back today. 17/20. should have got 18.5/20 (i know its petty, but i am annoyed at the really really stupid mistakes i made)

3/3 for both homework things.

others? (others.....there is only maryjane, no other 270'ers)

oh and for the assignment. i overheard some guy say that you have to transform the data, log it, to make it normal. i dunno, havent looked at it yet.

 

[flyin']

Really? Do you do ANOVA and two-way ANOVA and regression? STAT170 was so basic. But I still hate maths.

no, they don't do anova in 171.

 

[redruM]

omg! test tmw!!!

*runs*

 

[MaryJane]

Ugh, I know, how crap is it?! I'm not sure to how much she can test us on simple regression, because we've just learnt the t-test, but regression ANOVA isnt included in the exam... Should be interesting.

Probably more p-value finding things, a lot of ANOVA. I hope she doesnt get us to do error and power again (I hate it!!)

 

[redruM]

pretty good test i thought. just ran out of time to do the last part properly. another 5 mins would have done it.

 

[MaryJane]

Yeah, I ran out of time in that question where you had to decide whether the CI was reasonable for the plot. Sue said "one minute to go" and I thought that would be fine, but 10 seconds later she says "put your pens down!". That is false adversiting.... of a sort..

Oh well, on to do the assignment. Fun, fun times.

 

[redruM]

for the assignment are you doing;
1) H_o: u_Broadband = u_Dial-Up = u_Other
2) Regression between time spent watchin TV and reading newspaper

did you have the format in assignment 1 correct? the tutor said that mine wasnt in perfect accordance with the Short Guide to report Writing for Stat170 students.

also, do you want to compare the modified data (after getting rid of the students that dont have internet access at home or those that watched tv). i dont know how to work access, so i have to manually delete each 'record' and copy n paste in excel

edit: it wont let me upload *.xls, so had to zip them.

 

[MaryJane]

I didnt get any marks which said "dont do that here" or anything like that from memory. But I followed the stat170 guide pretty closely.... And I also put in a bit of my psych knowledge of reports (which may have lost me marks in the end).

Yep, those are the hypotheses I'm doing too... You just write the regression as "Ho: B1=0" or "H1: B1 not-equal-to 0", right? Cant be bothered going into word and doing subscripts like you!

I do the same thing re: excel and access. I totally dont know how to work access in the slightest! Did you just delete those people who answered funnily to questions? I saw your question to Sue on the board, and she said we could just delete them, so thats what I did, in case it got messy.

Also, with ANOVA, how do you test normality? Is it just using Bartlett's on minitab? or do you look at normal prob. plots too? I tried the prob. plots in minitab, but they look ugly because the answers for type of internet are categorical, not continuous. Although they both came back with p> 0.1. So I'm just saying that all the assumptions are met... is that what you got?

And with the regression question, are you transforming the data? I did a regression plot or something on them the other day when I was just mucking about with the data, and it looked completely off the fitted model, so I'm guessing we've gotta log it or something... I cant imagine Sue just giving us lovely, perfect data.

I've lost the original data in excel, dont know where it has disappeared too... so I'll attach my descriptive statistics, so we can check that way to see if we have the same!

 

[redruM]

with anova, check normality assumption by probability plot of residuals. not bartlett, that tests equality of variance.

yeah i think we have to transform the regression question. transformation works good (actually correctly!) for regression. you just antilog the predicted y. but you cant do that with something like a 2 sample t test. like in our last assignment, our conclusions were invalid for both questions.

i'll compare our descriptive stats after a while. just got back from work.

 

[Tabris]

Hye just wandering, how difficult is stat270? I did 170 and ECON141 seems to be a giant leap forward in extending its concepts. So how does 270 compare to 170 and ECON141??

 

[MaryJane]

I did 170 and I think that the stuff between 170 and 270 is pretty complimentary. I mean, the stat270 unit starts off revising 170 (t-tests and what not) and just builds on t-tests. Then it focuses on ANOVA and regression for the rest of the semester. Its pretty well-structured, and having 170 has given me a good understanding in basics. I'm doing a lot better in 270 than I did in 170, but I think thats all due to 170 topics.

Redrum: when you say transform, is there an actual transform command in minitab? I just logged it in excel and it came out normal...

 

[MaryJane]

Just a question:

With question 2, we can log the TV data, but not the newspaper (as it contains zero). But I dont think it would be statistically correct or viable to test logged data against natural data, as its two totally different types of data.. does that make sense?

So, there is no linearity in the two sets (unless you log TV), so how can we get the normality in order to do the test? Or do we just do the test (but mention the crappiness of the normality), because in the end it looks like there is a non-significant relationship anyway.

But then, if we do the test, do we use the logged Tv v. the natural newspaper? Or just leave them both?

I'm so confused! Stupid stat!! Help, please?

 

[redruM]

hey mj, how is the study going for stat?

i've finished my notes? hxxp://s20.yousendit.com/d.aspx?id=33VGCA4RN0ZN32PB0EML02OH2J (had to upload for a friend, so might aswell share)

pretty much get a few things onto the cheat sheet.

 

[MaryJane]

I havent really started studying as such.. just written up my sheet which surprisingly took a long time - nearly 4 hours by the time I went through the slides and that green book at the front to make sure I have everything. I'll take some pics and upload them, just need to steal Skittled's camera and laptop.

Wont be a tick - and ta for your notes and your email

 

[redruM]

I havent really started studying as such.. just written up my sheet which surprisingly took a long time - nearly 4 hours by the time I went through the slides and that green book at the front to make sure I have everything. I'll take some pics and upload them, just need to steal Skittled's camera and laptop.

Wont be a tick - and ta for your notes and your email

lol. cool.

i know i had seen the name yvette being thrown around here.

 

[MaryJane]

Yeah, its a pretty unique name so it was pretty safe you assuming you were talking to the "right" person

So here are my notes that I'm taking into the exam. Hope they are readable. Not as detailed as your notes obviously because its only the formula sheet, but just to check we have the same kinda stuff going on in our stat revision minds.

Oh, and I might have crapped on a bit with error and power. I totally dont get the whole conceptual part of it, I'm not a visual learner, and so I'm desperatly trying to understand it in figures so I can draw their ugly normal distributions to keep them happy.

 

[redruM]

yeah that power stuff goes over your head. i'd get it one day. and if i try to remember it later, i'd have no clue.

i am not looking forward to making the cheat sheet. it will be packed. lol. look at all your white spaces!