Quantitative Methods III (QM3 for IB/FE)

at Maastricht University

Join course
++++++ Ask your QM3 question here and an experienced Success Formula tutor will answer it as soon as possible ++++++ PLEASE ask your questions as a SEPARATE POST and NOT as a COMMENT here! Any problems or questions while studying/preparing for QM3? Don't worry, we got you covered! Just post any of your questions in the QM3 course on Studydrive and our professional tutors will help you!!! We wish you lots of success! Your Success Formula Team
how to use Eviews??
What is the best way to study QMIII?
I'm rewatching the lectures rn and they're a really nice refresher. Don't do the cases, I think just doing a ton of past exams is the only way
Could anyone tell me the exact date and time of the resit?
11 april 2019 09:00-12:00
Are you sure about this one? Why not c?
it is c)
Otherwise the effect of tangibles directly on the loyalty is not represented
why is it A?
which question?
can someone explain this?
why D is wrong?
Can you guys explain those 3 because for me it makes no sense
17) the definition of model c ensures that the coefficients of clr will be equal to b1, b2 etc. (for clr 2: 2*0.05=0.010 etc). And model c is only a specific version of the more general model d where y0 = 0 and y1=0.005. So, c is the correct answer 18) you can only use the partial f-test if one of your models is the reduced version of a complete model. This the case for every option except a) 19) basically its just plugging in the values for a partial f-test ((129.562-73.908)/(6-1))/(73.908/(12-6) = 0.9 which is between 0.75 and 1
No area was marked for this question
fyi: k) iii and o) iii are tested on a two-sided alternative here. Forgot to calculate the values for a one-sided alternative. might be that the p-values are significant then
Where do you get those number from? Do you maybe mean 1.016?
there is a one missing but the result is still correct
whoops yes you're right. typo
Can someone explain me answer C please ?
The combination of cross section and time series is called "panel data"!
Could someone explain what k and g are again please?
k: number of coefficients in the complete model g: number of coefficients in the reduced model - what counts as a coefficient? The intercept (constant) + all explanatory variables on the right hand side of the regression model. So a model [y = b0 + b1*x + b2*x ] would have k = 3 coefficients. If you now wish to compare this model to a reduced model using a partial F-test, you could omit one of the predictor variables, say, b2, and the model reduces to [y = b0 + b1*x] with g = 2 coefficients. Hope this comment helps.
No area was marked for this question
Yes we can!
can anyone explain more in depth?
As you see from the question "L) iii)" our significance level is 10%. Here we are testing the variables "Break" and "BreakMarket_Interact", so we oberserve significance levels of 0.035 and 0.107. So for 0.035 we can safely reject the null as it is below our 10% however for the other one the p-value is 10.7% which just above 10% we cannot reject the null at the 10% significance level even if it is very close.
Could anyone explain this ?
You are testing if the intercept is equal to 0 in other words if the intercept has any influence on the model. You can get all the data which she used from the SPSS output. The formula to calculate the value (point estimate-hypothesised value)/SE. The corresponding t value is higher than 1.653 which means we can reject Ho at alpha=0.05
How did you manage to create the dummy variable? I keep ending up with an error...
View 2 more comments
I believe it is "1 and 2" and not "0 and 1"
no dummies always take zero as a "no" and 1 as a "yes"
i made a mistake here, 1986-1998 are 13 years instead of 12 :D and there are 24037 firm observations whcih means we have 24037/13=1849 firms in our data set, 4069/13= 313 of them adopted the SAP module. This mistake does not affect any other task, just change these numbers :)
No area was marked for this question
quick question; how did you calculated the t-statistics in I) ii? thanks in advance
read it off the outputs
No area was marked for this question
How did you create the 2 graphs?
googled for something. I don't know how you can create those in spss
There is a free programme called Geogebra
Can someone explain this one ?? Thanks
No area was marked for this question
Question g ii) I was wondering whether the claim that the constant alpha appears to be zero should rather be non-zero, since we can clearly reject our null hypothesis that alpha january is equal to zero
No area was marked for this question
Isn't the rejection of null hypothesis the other way around for d)2) ?
yes it should be
How do you create the LAG formula in spss?
compute variable and then you have to klick "all" in the formula box, then you can choose "lag(1)" in the next box.
Why 9 ??
everyone who checked the same value for the two questions is probably a joker. Therefore, you go along the diagonal line of 1/1, 2/2 etc. that's 5+0+0+31+1+3. disregard the 31 because we cannot be as sure about them and you get a total of 9.
What do you have to type as a formula in spss to get the LAG ?
compute variable and then you have to klick "all" in the formula box, then you can choose "lag(1)" in the next box. I hope you understand what I mean haha
also works if you just type LAG( and then put the variable in the brackets, if you can't find it in the formula box
which method do you use to solve the last question?
How do you rescale this?
Transform > Compute Variable: “loyal3R” = 8 – “loyal3
No area was marked for this question
In part h it mentions youll need two new variables but you only compute and use one variable. I don't understand what the other variable would be
we've put them together in one, the first variable is "0.02*(clr1wk+clr2wk.....)" and the second variable is "0.01*(1*clr1wk+2*clr2wk+...6*clr6wk)".
No area was marked for this question
Where does the 4.7% chance come from, in question f
P-value from e) is the chance of a type 1 error
Shouldn't it be =/> 75% ?
it is an arbitrary number. all alphas are way above 70 anyways ;)
it's some number kerckhoffs told us
isn't it divided by 3 since there are 3 variables? same for the "-8"?
yes sorry i made a typo !
no problem :)
Did u deselect “Include constant in equation" in the linear regression analysis tools?
nope, I have the constant included
my bad
CASE 4: How do you solve task d), e) and f) ??
Could someone upload case 4? I have something wrong in the beginning as i keep ending up with positive autocorrelation.. but i can't find what.
Mistake here. Value was 0.022 so we can reject at 5% that the observed values are equal to the expected values. This means that the sample is actually unrepresentative of the population.
Here's a better explanation for c). We can with a certainty of 0.693 say that the variances of the Dutch and the German are equal. In this case we have to look at the Sig. (2-sided) value of that row, which is 0.000. This means that we can with high certainty say that the true means of the two population groups are not the same (Reject the H0 hypothesis). Hope its clearer
could anyone explain this ?
I made a mistake here, just figured it out myself. Its actually the opposite. With all significance values (except Q9_6) below the 5% threshold, we can assume that the true means differ from the benchmark mean value of 3. For Q9_6 on the other hand, we can with certainty say that the true mean is 3 for the population
No area was marked for this question
Why use Q4bin instead of normal variables (Q4) in question f)? Thanks
same question... i think the manual doesn't make it clear which we have to use, Q4 or Q4bin
doesn't make a difference in the results. you just have smaller outputs and results with the binary variable.
Why do we have to substract 1 from the birth month and day ??
The whole idea is kind of that we want to be able to compare people birthdays, and do that by calculating how far away from the beginning of the respective year their birthday is. If you're birthday is on the 15th of April then that's 4-1=3 fully completed months and 15-1=14 fully completed days away from the beginning of the year. So your birthday is after 3 full months and 14 full days, meaning during the 4th month and 15th day. Hope that makes any sense
Where do we see that they are negatively correlated ?
in the big correlation matrix from f) ii)
What does that mean ?
Instead of writing the same thing 20 times I just marked the correlations in the matrix that have a significantly high correlation and then sorted them whether they make sense to be positive or negative.
No area was marked for this question
for e test value should be 3
No area was marked for this question
Thanks a lot for this! Probably saved my life
a) i) q9 and 10 are not nominal but ordinal
Does someone still have the data sets for last years cases?
No area was marked for this question
Q18 asks "which option is not correct" in version SZ the answer should be C (looking at the grid) but the next sentence in the explanation says "The correctness of a) and c) should be obvious"????? can anyone explain?
Because they talk only about version TS in which b) is the right (wrong) answer
oh... reached a new level of braindead I guess, thank you!
No area was marked for this question
Can anyone explain Q13 step by step? I just don't get how you're supposed to get to the model solution by substitution. If I substitute y2= 1-y1 in the model, there's no more assurance there, how am I supposed to substract it then? And even if I get the first step, y1 and -1 equal themselves out don't they?
Shouldn't it rather be Ordinal data, here ?
Load more