The Student Room Group
Uni students - Complete our survey >>
Uni students - Complete our survey >>
Uni students - Complete our survey >>

Help with Goodness of Fit Test (X^2) Economics/Statistics

I'm currently completing coursework as part of my Economics degree, The task is to do a hypothesis test (goodness of fit test - using the X^2 distribution) on the census. I'm not sure if the working out I've done is right, Could someone please have a look at the excel sheet I've attached and critique the working out I've done. I'm not 100% sure what I've done is right. Thanks

https://drive.google.com/file/d/1OUuDNp2z5AQT1hrLNvrMBtPp5I9kZQs8/view?usp=sharing
(edited 3 years ago)
Reply 1
Avatar for vc94
vc94
12
Chi-squared calculations look sound, big values though - is this usual in economics?!
For Region and Economic Activity, you probably mean "Reject H0".
Similarly for Health and Occupation, you reject H0, since your chi-squared stats are much more extreme than the critical values.
Couldn't you test at the 1% or 0.1% level (instead of just 5%) to show extremely significant results?
Reply 2
Avatar for roshrr
roshrr
OP
3
Original post by vc94
Chi-squared calculations look sound, big values though - is this usual in economics?!
For Region and Economic Activity, you probably mean "Reject H0".
Similarly for Health and Occupation, you reject H0, since your chi-squared stats are much more extreme than the critical values.
Couldn't you test at the 1% or 0.1% level (instead of just 5%) to show extremely significant results?

Thanks for replying so quickly, I also thought the values were too high so that's why I posted it here to see if it was unusual. The critical values at the 1% level arent that much greater than what it is currently. Am I missing something here?I think it may be incorrect.
Reply 3
Avatar for vc94
vc94
12
If you look at your section "Observed value of the test statistic" you can identify which cells have comparatively large contributions to the chi squared statistic, i.e. zone in on areas where the association between the factors are most significant.
Reply 4
Avatar for roshrr
roshrr
OP
3
Original post by vc94
Chi-squared calculations look sound, big values though - is this usual in economics?!
For Region and Economic Activity, you probably mean "Reject H0".
Similarly for Health and Occupation, you reject H0, since your chi-squared stats are much more extreme than the critical values.
Couldn't you test at the 1% or 0.1% level (instead of just 5%) to show extremely significant results?

Ive completed it using the excel method however now the chi squared statistic is now showing as 0 despite the observed and expected values being different. I've updated the link with a new excel sheet, Could you please have a look to see whats wrong?

https://drive.google.com/file/d/1OUuDNp2z5AQT1hrLNvrMBtPp5I9kZQs8/view?usp=sharing
Reply 5
Avatar for vc94
vc94
12
Did you figure out why it's zero?!
Reply 6
Avatar for roshrr
roshrr
OP
3
Original post by vc94
Did you figure out why it's zero?!

Nah I'm struggling to understand unfortunately the values are different so the chi squared should surely have some sort of value
Reply 7
Avatar for vc94
vc94
12
I copied and pasted one set of data into:
https://www.icalcu.com/stat/chisqtest.html
To get a chi square of 14644.52322408896 (like you had before) and a p-value of zero!
Reply 8
Avatar for roshrr
roshrr
OP
3
Thank you so much!!! I manged to figure it out now, was going round in circles because i was confusing the critical values and the p values for some reason. Thanks so much for your help

Quick Reply

Related discussions

Latest

Trending

Trending

Articles for you