Official STAT270 thread (2 Viewers)

[MaryJane]

I dont know how many people are doing STAT270, but I know that at least redrum, clairegirl and myself are...

Basically, I have issues with the homework, question 3 (b). I think I'm too Englishy for my own good, and pick apart questions too much.

If you read the question, they have given us output from a vehicle's (so, a sample from a population) MPG. Fair enough. But then I'm confused with part (b) because it asks for "a 95% CI for µ, the mean MPG for this vehicle".

You see, I dont get it. µ is a symbol used for a population, right? But then they are asking "for this vehicle". I can get it both ways:

Possibility 1: they are asking for the possible population CI, based on this one sample (which is a pretty unrealistic way to sample a population but hey...) So we use the population standard deviation.

Possibility 2: how can they ask for a population measure from this one sample? And I get the impression that its this one sample because a) they say "for this vehicle" and b) we have no other output. So we use the sample standard deviation, not the population standard deviation.

I think I'm reading into it waaay too much, and possibility one is correct (thus using the µ standard deviation etc), but then its worded crappily and its annoying me!

Kill me now.

 

[redruM]

samples are used to estimate the population.

the population in this case is the MSG of this vehicle, not of all the vehicles in the world (i think this is the point of confusion?). and the sample is trying to estimate it.

i should get this done aswell. want it out of the way. anyways, hope that makes some sense^. i've just come from work and brain isnt in stat mode.

 

[Skittled]

Yeah... samples are used to estimate the population...

The question MJ is really asking is whether to do a Z or T* test. I've just finished stat270, and I know that they're statisticians, not english teachers, so they're not always overly clear in how they ask their questions.

In the end, they've given you the population parameter, theta (or whatever the population SD is), and you need to use that somewhere. That's just my observation, anyway..

Regardless, they ask for "a 95% CI for µ, the mean MPG for this vehicle". Yes, that could be for this, physically, one vehicle. In that case, you'd use T*: you don't know the true SD for this vehicle, only have an estimate of it from the samples given (the population parameter is irrelevant, in this case, because we're not trying to find out anything regarding the population, just this one car, which then becomes our target population, as redrum pointed out.)... and thus, you do a CI using the T tables.

However, if they ask for "a 95% CI for µ, the mean MPG for this vehicle", and then you interpret "this vehicle" to mean "this type of vehicle"... all of a sudden, the population parameter is relevant, and you can now do a CI using the Z tables.

I'm leaning both ways, (I know that shouldn't be possible but I am), so forgive the rambling but I'm trying to talk myself through it!

 

[redruM]

firstly, i'd say that with stat you dont need to worry about being tricked by tricky wording.

i see the problem is in the wording of the question. that problem of what the population is. it has actually come up a couple of times. once in lectures slides. maybe that is something we may need to keep a lookout for in this subject.

my decison would be to use sigma = 2.9 in a z test. you could make a comment mentioning that you are assuming that it is referring to population as the mpg of this vehicle.

 

[Skittled]

my decison would be to use sigma = 2.9 in a z test. you could make a comment mentioning that you are assuming that it is referring to population as the mpg of this vehicle.

Good call. Why didn't I think of that?

 

[redruM]

are you doing stat270 this sem skittled, or did it last semester?

 

[Tabris]

This might be abit off topic but anyway

I was sitting on the white metal seats outside library waiting for my lecture to start (15 min to go)

Out of nowhere, this african American (can tell coz he had stron US accent)

came out of nowhere and started asking people if they played chess....

I said i use to long time ago and he gave me a problem sheet....on it were lots of problem scenarios and u had to give the solution

I didnt finish all of it (needed more time) i handed it back and asked what was it for

He said its for Psychology and he was a transfer student.... and he needed the survey and the statistics for his work....

Anyone else did the survey?


BTW: i heard pysch is pretty statistics heavy i like the theory but stats is scary, never was good at maths....u arts degree people dont know how lucky u are compared to the ECFS people....

I wish i could choose HIST and AHST units.....but my degree wont let me and besides... u must be very very good at it to make a living out of it.....usually find myself sitting in the library 3rd level going trough Roman history and medieval stuff... (yeah i know alot about this stuff) and Freud is pretty freakky stuff to

 

[redruM]

was this guy a bit older looking? because if he is, he can really sing well (mq idol)

how are you guys going with homework4? i got all but the conclusion for q3 done in the first half of the lecture, then i went home

 

[MaryJane]

You're v. stuious redruM, I havent even looked at it! Thats my job for today/tonight.

I havent seen a guy coming around randomly with surveys. If it was for psych he should have given you a consent form to show that you participated voluntarily and you dont mind having your data released. Psych isnt that statistics heavy, in fact I've only used my statistics knowledge in first year (PSY105) where you have to write reports on data they give you. You do all the calculations and take it from there. I'm guessing it will get more use in third year for this assessments core unit thing. Personally, I just think its in there as a bit of a token/history thing-psych and stat have always been linked and so we have to learn it. It would help the honours/post-grad students the most, not just boring undergrads like me who are doing the degree and running!

I hate statistics. But at least the lecturer is nice.

 

[redruM]

lol, how can you hate statistics?

btw, are you doing this as an elective or is it core for your degree?

 

[MaryJane]

Ahahahahaa... Before I went to uni, I hadnt done maths since year 10. I am not a mathsy person. Its a core unit for psych, thank god its my last year of statistics. I can think of many other units I would have done if I had an elective!

 

[redruM]

we have exam next week? it is in the lecture, yeah.

(shameless stat270 thread bump)

 

[MaryJane]

lol

Yeah, next lecture in the first half (only goes for an hour), and then the lecture resumes as per normal, and we're given another assignment (but a proper assignment, not homework).

Thank god for the ability to write your own formulas sheet!

 

[redruM]

omg! test tmw!!!

i made some notes on assumptions for a friend. so i thought i would upload it for the sake of procastination.

 

[MaryJane]

Hmmm.. didn't Sue said that we always assume independence?

Either way, the notes are good! I've written my cheat sheet, but its hand-written for the test, so I cant upload it. I just have the assumptions, different tests, type I and II error and the CI's... I think thats all we need, right? We havent really covered anything else?

I practised error on Q.10 from the misc. exercises and got them right (!!!) so hopefully the test will be similar! I still dont totally understand the error concepts, but know how to do it, and where to shade in under the bell-curve, so thats good enough for me!

 

[redruM]

i must not have paid attention when she mentioned that we always assume independence .

i havent made my cheat sheet yet. and i wasnt at the powers lecture, so i have to learn that stuff. but my tutor was good enough to teach me a bit in the tutorial! should be ok i think. i remember that same thing, mj, i did a type I error question, and said, "yeah i got that answer, but got no idea what it means"

i guess we are lucky that stat doesnt need too much explaining.

 

[MaryJane]

Did you do 170, or 171? I find that the little bit of base laid down by 170 really helps in understanding 270. I mean, we've only learnt three new things (unpooled t-test, error and power), but you kinda implicitly know what you have to do in a test (like p's and assumptions etc).

I wouldnt go as far as to say that I'm enjoying stat, but its good. I really like Sue, she's so nice and approachable and down-to-earth, and doesnt talk up the stat stuff (like making a t-test seem really complicated when its one of the simplest things you can do) the way that some of my other lecturer/tutors have.

 

[redruM]

you can say i did 171, i failed it and didnt learn anything. then i did 170 the following semester and really enjoyed it. what really helped my understanding of stat was doin econ141 semester 1 this year. i didnt go crash hot in it, but understood that 'thinking' needed statistics. makes for efficient learning (read: cramming)

 

[redruM]

how did you go?
i found that 2 hrs sleep the night before didnt help

haha, but i had a colourful cheat sheet

(take note of the staple job )

 

[MaryJane]

Ohh, it is v. pretty and colourful - mine was just in boring black ink!

I thought the test was fine. I freaked out though when I got a negative CI (I had the blue paper) and then I started second-guessing myself re: whether I was right to use unpooled, but I spoke to a few other blue-sheeters afterwards and they got the same result which was a relief! I didnt even attempt to answer those ones that wanted the prediction without working it out! I started working it out on my cheat sheet, but then Sue said we only had 10 mins left, so I finished the typeII error question dead on time, so I'm glad I didnt bugger about with it.

Glad its over: only two more tests next week for psych