Homework answers / question archive /
Econ B2000, Statistics and Introduction to Econometrics
The first questions do not require any work in R (although you might find it convenient, I’m not stopping you) – I will provide some summary data and you can construct hypothesis tests on your own
Econ B2000, Statistics and Introduction to Econometrics
The first questions do not require any work in R (although you might find it convenient, I’m not stopping you) – I will provide some summary data and you can construct hypothesis tests on your own
Economics
Share With
Econ B2000, Statistics and Introduction to Econometrics
The first questions do not require any work in R (although you might find it convenient, I’m not stopping you) – I will provide some summary data and you can construct hypothesis tests on your own. Subsequent questions require R coding.
- (20 points) I’ve run crosstabs on a subset of the data (so you cannot replicate, just use these data as provided). These give verbose summary of vaxx choice by educational qualification and region. Form a hypothesis test of the form, “people with various educational qualifications in Region have different fraction vaxxed compared with other Region.” I expect that you will choose different ways to operationalize educational qualification (compare above some level with below that level, but what level?) and different regions (Census provides 4 – Northeast, Midwest, South, West, you might combine them). You can choose how to deal with NA responses to vaxx – perhaps count them as ‘no’? I expect that different people may choose different levels of significance. Please provide estimate, standard error, t-stat and a p-value for the hypothesis test and a confidence interval. Write a short paragraph explaining the test (carefully noting what is the null hypothesis) and explaining the results of that test.
> xtabs(~ EEDUC + RECVDVACC + REGION)
, , REGION = Northeast
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 0 25 15
some hs 1 95 17
HS diploma 25 924 174
some coll 20 1407 201
assoc deg 15 834 104
bach deg 34 2603 147
adv deg 19 2680 88
, , REGION = South
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 0 92 45
some hs 4 192 101
HS diploma 31 1844 585
some coll 57 3498 747
assoc deg 28 1704 360
bach deg 59 5193 498
adv deg 62 5024 284
, , REGION = Midwest
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 3 38 17
some hs 5 95 49
HS diploma 25 1217 329
some coll 32 2093 454
assoc deg 26 1227 245
bach deg 56 3228 311
adv deg 39 2680 143
, , REGION = West
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 2 107 28
some hs 4 204 75
HS diploma 24 1532 386
some coll 55 3896 716
assoc deg 29 1870 313
bach deg 68 5397 467
adv deg 50 4676 218
- (20 points) I’ve run crosstabs again, this time on vaxx choice by educational qualification and gender identification. Form a hypothesis test of the form, “people with various educational qualifications who are one or more gender ID have different fraction vaxxed compared with another gender ID.” I expect that you will choose different ways to operationalize educational qualification (as noted in Question 1) and different genders (including the NA response, perhaps it makes sense to combine some). Choose a level of significance. Please provide estimate, standard error, t-stat and a p-value for the hypothesis test and a confidence interval. Write a short paragraph explaining the test (carefully noting what is the null hypothesis) and explaining the results of that test.
> xtabs(~EEDUC + RECVDVACC + GENID_DESCRIBE)
, , GENID_DESCRIBE = NA
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 1 9 2
some hs 9 9 2
HS diploma 66 63 14
some coll 84 84 20
assoc deg 62 43 7
bach deg 148 126 20
adv deg 123 120 16
, , GENID_DESCRIBE = male
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 3 106 29
some hs 1 248 91
HS diploma 14 1992 557
some coll 21 4238 783
assoc deg 8 1902 297
bach deg 26 6773 504
adv deg 20 6271 263
, , GENID_DESCRIBE = female
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 0 131 57
some hs 3 309 144
HS diploma 25 3372 876
some coll 54 6411 1292
assoc deg 28 3613 703
bach deg 39 9354 873
adv deg 22 8524 413
, , GENID_DESCRIBE = transgender
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 1 3 6
some hs 0 6 1
HS diploma 0 21 4
some coll 0 42 5
assoc deg 0 12 3
bach deg 1 42 2
adv deg 0 28 10
, , GENID_DESCRIBE = other
RECVDVACC
EEDUC NA yes got vaxx no did not get vaxx
less than hs 0 13 11
some hs 1 14 4
HS diploma 0 69 23
some coll 5 119 18
assoc deg 0 65 12
bach deg 3 126 24
adv deg 5 117 31
- (80 points) Now do your own analysis using “Household_Pulse_data.RData”. Choose an interesting topic to explore, different from previous questions. The data includes information on housing (rent or own; whether behind on rent or mortgage), food shortage, whether work remote or in person, whether kids are in school in person or remote or homeschooled, how anxious or worried, vaxx status and plans, along with demographics like race/ethnicity, gender, marital status, and household income (as a factor).
- Choose a subgroup of the sample to consider and provide summary statistics of that subgroup. Explain why this subgroup is interesting.
- Form a hypothesis test about an interesting variable, explore whether your chosen subgroup differs from the rest of sample. Please provide both a p-value for the hypothesis test and a confidence interval. Write a short paragraph explaining the test (carefully noting what is the null hypothesis) and explaining the results of that test.
- Using a k-nn classifier, can you find relevant information to predict an interesting outcome? How good is the classifier? Discuss.
- Can you explain some other interesting information about this data? Some interesting crosstabs? Maybe regressions? Impress me.
