So how to look at the first section:
You would state a null hypothesis about rather you think the opinion has changed or not. So understand that, you'd understand that the question of interest is "did opinions change"? So we then start to work off of the assumption that opinions haven't changed. So H0: Population 2010 follows the 2001 distribution, meaning that the overall distribution of opinions don't really differ that much, while H1: The population doesn't follow the 2001 distribution.
Look at the second section:
So now you have to calculate the actual values, which is multiplying each percentage in the 2010 distribution by the number of respondents, which is 1019. So you'd multiply 1019*.37, 1019*.12, etc. and you'd do the same thing for the 2001 respondents, which is multiplying 1019 by the 2001 probability distribution, hence (1019*.38, 1019*.37,.... etc). That would be the list of expected numbers, while the numbers for 2010 would be the Observed numbers.
Look at the 3rd section:
We'll do a chi-squared test, which would be given by the formula Sum(Oi - Ei)^2/Ei), with degrees of freedom (number of categories -1 ). In this case, that would be 3, since we have 4 different types of responses. Then, once we determine the alpha level, that will give us the rejection region, which in this case there should be 2, so we would divide our alpha level by 2. You can choose our alpha level to be .01,.05, .10.
Deontae H.
11/03/20