**HL Maths P2 Before/After

ehshup

in the 2010 paper, Q9A (c) there's a question: Conduct a hypothesis test at the 5% level of significance to decide whether the machine's setting has become innacurate. You should start by clearly stating the null hypothesis and the alternative hypothesis, and finish by clearly stating what you conclude about the machine.

The marking scheme has
Null Hypothesis : The machine is accurate
Alternative Hypothesis: The machine isn't accurate

My question: How do you know which is the null hypothesis? I thought because the statement being tested is that the machine is inaccurate hence the null and alternative would be the opposite way to the marking scheme? Does it matter?

RedTexan

ehshup wrote: »

in the 2010 paper, Q9A (c) there's a question: Conduct a hypothesis test at the 5% level of significance to decide whether the machine's setting has become innacurate. You should start by clearly stating the null hypothesis and the alternative hypothesis, and finish by clearly stating what you conclude about the machine.

The marking scheme has
Null Hypothesis : The machine is accurate
Alternative Hypothesis: The machine isn't accurate

My question: How do you know which is the null hypothesis? I thought because the statement being tested is that the machine is inaccurate hence the null and alternative would be the opposite way to the marking scheme? Does it matter?

The 5% level of significance isn't on the paper any more, or so my edco papers suggest

Canard

ehshup wrote: »

in the 2010 paper, Q9A (c) there's a question: Conduct a hypothesis test at the 5% level of significance to decide whether the machine's setting has become innacurate. You should start by clearly stating the null hypothesis and the alternative hypothesis, and finish by clearly stating what you conclude about the machine.

The marking scheme has
Null Hypothesis : The machine is accurate
Alternative Hypothesis: The machine isn't accurate

My question: How do you know which is the null hypothesis? I thought because the statement being tested is that the machine is inaccurate hence the null and alternative would be the opposite way to the marking scheme? Does it matter?

I assume its up to you from the bolded bit

RedTexan wrote: »

The 5% level of significance isn't on the paper any more, or so my edco papers suggest

It is...where are people getting the idea that it isnt, its in all the books and its in the sample papers

ehshup

Patchy~ wrote: »

I assume its up to you from the bolded bit


It is...where are people getting the idea that it isnt, its in all the books and its in the sample papers

It is according to http://www.projectmaths.com/index.php/2012/05/project-maths-statistics-some-clarifications/

Ok that's ok so, I hate the way the marking schemes don't include for that, there's a good chance a corrector would see them the other way around in the marking scheme and put a big fat X on my page, I don't think project maths could have been organised in a worse way, the masterminds behind this should be ashamed:pac:

cocopopsxx

Patchy~ wrote: »

I assume its up to you from the bolded bit

It is...where are people getting the idea that it isnt, its in all the books and its in the sample papers

If you have the edco exam papers, it says at the top of this question that "material has been deferred for three years". Afaik, we don't have to conduct a hypothesis test. :S

ehshup

cocopopsxx wrote: »

If you have the edco exam papers, it says at the top of this question that "material has been deferred for three years". Afaik, we don't have to conduct a hypothesis test. :S

we definitely do have to be able to as far as i know! that's what margin of error is used for!

Helloxoxo

ehshup wrote: »

cocopopsxx wrote: »

If you have the edco exam papers, it says at the top of this question that "material has been deferred for three years". Afaik, we don't have to conduct a hypothesis test. :S

we definitely do have to be able to as far as i know! that's what margin of error is used for!

It is a complex version that is not on our course but the courses when other strands are brought in

cocopopsxx

ehshup wrote: »

we definitely do have to be able to as far as i know! that's what margin of error is used for!

But we were told we have to do it at 95% confidence interval and not 'conduct a hypothesis' at 5% level of significance? Just the margin of error thing, not this conduct a hypothesis thing? the question you just asked, we were told it's not on this year and my exam papers say so too? I'mconfused....

EDIT:- nevermind, I'm just confusing myself and others...sorry

Dapics

cocopopsxx wrote: »

If you have the edco exam papers, it says at the top of this question that "material has been deferred for three years". Afaik, we don't have to conduct a hypothesis test. :S

I was fairly sure it was only two tailed hypothesis tests that have been deferred.

As far as i know Hypothesis testing is may still come up.

As for null hypothesis, its like probability.

Let H0 = Probability hypothesis is correct
Let H1 = Probability hypothesis is incorrect.

When you do your 95 percent confidence interval, if the value/variable they have given and with which you assigned to H0 and H1 is within the confidence interval, then the statement is true.
I.e- H0 within interval

Same goes for if its not in Confidence interval, the statement is untrue.

I.e - H1 not within interval.

Also Probability can be summised like this:

If you are given a coin/lotto/cards question on probs then the first thing you should attempt to do is to answer it using bernouill's theorem as these types of events are independent of each other.

cocopopsxx

Dapics wrote: »

I was fairly sure it was only two tailed hypothesis tests that have been deferred.

As far as i know Hypothesis testing is may still come up.

As for null hypothesis, its like probability.

Let H0 = Probability hypothesis is correct
Let H1 = Probability hypothesis is incorrect.

When you do your 95 percent confidence interval, if the value/variable they have given and with which you assigned to H0 and H1 is within the confidence interval, then the statement is true.
I.e- H0 within interval

Same goes for if its not in Confidence interval, the statement is untrue.

I.e - H1 not within interval.

Also Probability can be summised like this:

If you are given a coin/lotto/cards question on probs then the first thing you should attempt to do is to answer it using bernouill's theorem as these types of events are independent of each other.

Thanks

So the 2010 SEC sample Q9A (c) is still on the course? Oops....

RedTexan

ehshup wrote: »

we definitely do have to be able to as far as i know! that's what margin of error is used for!

that's 95% level of significance which is still on the course

RedTexan

You do have to be able to that with the null hypothesis and the margin of error, but the 5% level of significance isn't still there it's 95% don't ask the difference because I don't know!

RedTexan

Wait are the just the same thing?

cocopopsxx

RedTexan wrote: »

You do have to be able to that with the null hypothesis and the margin of error, but the 5% level of significance isn't still there it's 95% don't ask the difference because I don't know!

Yeah this is exactly what I wanted to say. We've only done that null hypothesis and margin of error with the 95%, not the 5% level of significance....

Dapics

RedTexan wrote: »

You do have to be able to that with the null hypothesis and the margin of error, but the 5% level of significance isn't still there it's 95% don't ask the difference because I don't know!

Null Hypothesis is easy and yes i would think it will come up.
Statistics should be a lovely question, i just learned off loads of definitions/methods of sampling and how to apply them to questions.

Should be a lovely paper for me.

Im hoping the trig questions are horrible so I can out-perform everyone else in my year and hopefully get an A1 in Paper 2 :cool:
I know that damn question inside out, I reckon it will either be very hard or very easy, here's hoping its hard.

Apart from trig, the paper should be lovely. The Circle may get pretty disgusting if they ask you to prove a tangent at the point of intersection of two circle's goes through the centre of one of the circles... lovely question for me but would be hard for lots of other people.

I think the paper 2 course is far broader than what the textbooks do, particularly in regards to the circle and trigs.... I only learned recently the true depth of what they could ask us...I mean there's a question I came across which depended on a rule i had never seen before, but was actually in the course.... it was based on the CAST circle.

ehshup

RedTexan wrote: »

Wait are the just the same thing?

yes they are. 95% confidence means we can be 95% sure the value will fall within the boundaries. 5% significance means there is a 5% chance it will be outside. Which is pretty much the same thing

RedTexan

Dapics wrote: »

Null Hypothesis is easy and yes i would think it will come up.
Statistics should be a lovely question, i just learned off loads of definitions/methods of sampling and how to apply them to questions.

Should be a lovely paper for me.

Im hoping the trig questions are horrible so I can out-perform everyone else in my year and hopefully get an A1 in Paper 2 :cool:
I know that damn question inside out, I reckon it will either be very hard or very easy, here's hoping its hard.

Apart from trig, the paper should be lovely. The Circle may get pretty disgusting if they ask you to prove a tangent at the point of intersection of two circle's goes through the centre of one of the circles... lovely question for me but would be hard for lots of other people.

I think the paper 2 course is far broader than what the textbooks do, particularly in regards to the circle and trigs.... I only learned recently the true depth of what they could ask us...I mean there's a question I came across which depended on a rule i had never seen before, but was actually in the course.... it was based on the CAST circle.

The part (a) of the trig question is almost always based on the unit circle? It would only make sense that they made it reasonable to support introducing it in the first place, but who knows.

Spattersonox

I HATE PROBABILITY .. unless it's Bernoulli, I haven't a bauldy notion

Dapics

RedTexan wrote: »

The part (a) of the trig question is almost always based on the unit circle? It would only make sense that they made it reasonable to support introducing it in the first place, but who knows.

What i mean is that they may throw you a question from which you have to proove something. This something is based on the fact that in a CAST circle, one angle plus another angle =90. ( Due to quadrants)
They may give you a question which is based on that twist.

ehshup

Dapics wrote: »

Null Hypothesis is easy

Dapics wrote: »

Statistics should be a lovely question

Dapics wrote: »

Im hoping the trig questions are horrible so I can out-perform everyone else in my year and hopefully get an A1 in Paper 2

Dapics wrote: »

here's hoping its hard.

that's the spirit:pac:

Dapics

Spattersonox wrote: »

I HATE PROBABILITY .. unless it's Bernoulli, I haven't a bauldy notion

In case you are wondering when you can use bernoullis theorem... I have always thought that if :
If you are given a coin/lotto/cards question on probs then the first thing you should attempt to do is to answer it using bernouill's theorem as these types of events are independent of each other

Am I right? I think so. Anyone care to reaffirm this suggestion?

I too am not great at probability as I am usually unable to understand which method I should use to solve the question.

ehshup

Dapics wrote: »

In case you are wondering when you can use bernoullis theorem... I have always thought that if :
If you are given a coin/lotto/cards question on probs then the first thing you should attempt to do is to answer it using bernouill's theorem as these types of events are independent of each other

Am I right? I think so. Anyone care to reaffirm this suggestion?

I too am not great at probability as I am usually unable to understand which method I should use to solve the question.

it depends on the question, just see if you can apply it and if you can, then do.
i.e. if it asks about the probability that a certain number of independent events happen out of a total number of events. But be careful to read it like if it's on lotto it could ask the probability of winning, that's not gonna be bernoulli. But i think you get the concept yep!

RedTexan

Dapics wrote: »

In case you are wondering when you can use bernoullis theorem... I have always thought that if :
If you are given a coin/lotto/cards question on probs then the first thing you should attempt to do is to answer it using bernouill's theorem as these types of events are independent of each other

Am I right? I think so. Anyone care to reaffirm this suggestion?

I too am not great at probability as I am usually unable to understand which method I should use to solve the question.

It depends, in the case of the lotto/cards if balls/cards are being removed without replacement bernoulli trials won't work because the probability is changing

Spattersonox

Dapics wrote: »

In case you are wondering when you can use bernoullis theorem... I have always thought that if :
If you are given a coin/lotto/cards question on probs then the first thing you should attempt to do is to answer it using bernouill's theorem as these types of events are independent of each other

Am I right? I think so. Anyone care to reaffirm this suggestion?

I too am not great at probability as I am usually unable to understand which
method I should use to solve the question.

I think you are yeah
You can use bernoulli when something can either happen or it can't.
So if you have a 0.75 chance you will have a flight on time and a 0.25 chance it'll be late.. then you can use it. S, yeah, completely independent events
I'm pretty sure of this anyways. As I said, i hate probability and it is most definitely my worst section

Spattersonox

I just do not get probability! It confuses the hell out of me.

Dapics

ehshup wrote: »

it depends on the question, just see if you can apply it and if you can, then do.
i.e. if it asks about the probability that a certain number of independent events happen out of a total number of events. But be careful to read it like if it's on lotto it could ask the probability of winning, that's not gonna be bernoulli. But i think you get the concept yep!

If its the prob of winning then would that just be 1 divided by the combination of the amount of no's and the no. of numbers necessary to win?

If so then I think I understand that.

If the chance a person is going to school on a tuesday is 2 over 7 and the probability of a person going to school on a wednesday is 3 over 7 and they asked you to find the prob they go both days, would that just be multiplying the prob's out?

Another wee thing Im stuck on is a question like this:

There are 8 boys and 2 girls in a class, 7 boys do french and x girls do german.... how many girls do german?

I take it you just make it 8 - x and divide by the combination of boys who do french?

These events are therfore not independent of each other?

Any help would be appreciated.

ehshup

If you are told that 330/1100 people like Yummy Cola, and a month later you conduct a survey of your own and find that 95/250 people like it. Can you say with 95% confidence that more people like Yummy Cola now?

Now with a question like that (one of which came up in our mock), the margin of error was only used once in the marking scheme as follows:
population proportion = 330/1110 = .3 or 30%
sample proportion p = 95/250 = .38 or 38 %
Margin of error = 1/√250 = .063
Confidence interval = .317 < p < .443
Population proportion, .3, is not within confidence interval, hence one can conclude that more people do like Yummy cola.

Now I have a problem with this solution, because it didn't take into account the possible error in the first test:
Margin of error for first = 1/√1100 = .03
hence the confidence interval for the first is .27 < p < .33
and so it is possible that the amount of people who like Yummy cola is the same, as the confidence interval for one overlaps the confidence interval for the other.
While I made this question up off the top of my head, the question in our mock was almost the same, however I got no marks for it because I calculated margin of error for the first. Can anyone confirm what the right thing to do is in this situation?

ehshup

Dapics wrote: »

If its the prob of winning then would that just be 1 divided by the combination of the amount of no's and the no. of numbers necessary to win?

Yes

Dapics wrote: »

If the chance a person is going to school on a tuesday is 2 over 7 and the probability of a person going to school on a wednesday is 3 over 7 and they asked you to find the prob they go both days, would that just be multiplying the prob's out?

Yes

Dapics wrote: »

There are 8 boys and 2 girls in a class, 7 boys do french and x girls do german.... how many girls do german?

I take it you just make it 8 - x and divide by the combination of boys who do french?

These events are therfore not independent of each other?

Well you can't work out how many girls who do german because we have no info about people doing german at all? You can only conclude that events are independent if P(A|B) = P(A) or in other words P(A ∩ = P(A).P(B) So in this case we need more info to know

hamal

What are ye doin for q6? learnin theorems 11,12 and 13 and hopin for the best?

Also are ye all learning the derivations of the trig formulae? awful lot of work for something that has never been in an SEC paper/sample paper!:rolleyes:

RedTexan

ehshup wrote: »

If you are told that 330/1100 people like Yummy Cola, and a month later you conduct a survey of your own and find that 95/250 people like it. Can you say with 95% confidence that more people like Yummy Cola now?

Now with a question like that (one of which came up in our mock), the margin of error was only used once in the marking scheme as follows:
population proportion = 330/1110 = .3 or 30%
sample proportion p = 95/250 = .38 or 38 %
Margin of error = 1/√250 = .063
Confidence interval = .317 < p < .443
Population proportion, .3, is not within confidence interval, hence one can conclude that more people do like Yummy cola.

Now I have a problem with this solution, because it didn't take into account the possible error in the first test:
Margin of error for first = 1/√1100 = .03
hence the confidence interval for the first is .27 < p < .33
and so it is possible that the amount of people who like Yummy cola is the same, as the confidence interval for one overlaps the confidence interval for the other.
While I made this question up off the top of my head, the question in our mock was almost the same, however I got no marks for it because I calculated margin of error for the first. Can anyone confirm what the right thing to do is in this situation?

You would most likely be told a population proportion rather than a sample at the beginning, badly worded question.