The first two files are what I need to do. The directions and data. The last three are slides from class
Directions Dear PLS3100 student Please upload your final to Pilot’s dropbox (Final Exam Dropbox). Do not email me your answers. Late work will not be accepted. I do not need to see any SPSS output. From this point on, I am no longer able to answer any of your questions. You are welcome to use your notes, refer to the book or rewatch whatever video lectures you think will help you. Good Luck and have a great holiday. SPSS directions To get OLS results on SPSS, start by loading the data set as usual. Once the data is loaded, click on Analyze and then click on Regression. Select linear. Select your dependent and your independent variables. Click on Statistics and select estimates and Confidence Intervals. Click on Continue and click OK . 1 The Covid-19 Pandemic The COVID-19 pandemic is about to turn 2 years old, yet it continues to disrupt our lives and shows little signs of slowing down. The orders to social distance, wear masks, the closures of business and other related issues have been very controversial. The Delta and Omicron variances created even more fear and confusion. Despite all of this, lots of people still refuse to take the vaccine. In an effort to understand why people are unwilling to take a vaccine that can save their lives, we consulted with several experts on the topic. They believe that unwillingness to take the vaccine is caused by [1] degree of conservatism, [2] fear of a government sponsored vaccine and [3] mistrust in science. To test these claims, you collected a data set with several variables [Vaccine]. The data set contains one variable measuring the unwillingness to take the vaccine [UnwillingnessVaccine], two variables for level of conservatism[ConservLvl1 and ConservLvl2], two variables for mistrust in science [MistrustScience1 and MistrustScience2] and one variable for fear of government [FearGov]. Please use the data set to answer the following questions: 1. Which concepts are the drivers of unwillingness to take the vaccine? 2. What is the relationship direction for each of the drivers ? 3. How many models do you intend to run and why ? 4. Of all models you ran, which one did you pick and why ? 5. List the null and working hypotheses that are necessary to test the claim 6. Is the impact of all independent variables real ? 7. When it comes to the push, which variable has the weakest impact? 8. Which variable has the strongest pull ? 9. Which variable has the overall greatest impact ? 10. Please compute the expected level of unwillingness to take the vaccine when mistrust in science is at 9, fear of government is at 5 and conservatism is at 2 PLS3100: Quantitative Methods In Political Science PLS3100: Quantitative Methods In Political Science Carlos Costa 1/19 Part 2a Random Error Avoid OVB SPSS Previous lectures tests and tests OLS and its many features ◦ direction, push (magnitude),pull (strength) DV continuous and unbounded relationship must be linear PLS3100: Quantitative Methods In Political Science Carlos Costa 2/19 Part 2a Random Error Avoid OVB SPSS What do we need to run an OLS test last requirement: random error error must be random what is the error? ◦ difference between the estimated and observed values if error not random: OVB Omitted Variable Bias ◦ population β is not reliable PLS3100: Quantitative Methods In Political Science Carlos Costa 3/19 Part 2a Random Error Avoid OVB SPSS Error the part we cannot explain consider that we have some data ◦ bunch of x values ◦ 1,2,3,4,. . . ,10 ◦ bunch of corresponding y values ◦ 6,9,12,15,. . . ,33 plot them for the sake of illustration PLS3100: Quantitative Methods In Political Science Carlos Costa 4/19 Part 2a Random Error Avoid OVB SPSS Graphing the points getting a sense of the data PLS3100: Quantitative Methods In Political Science Carlos Costa 5/19 Part 2a Random Error Avoid OVB SPSS Error the part we cannot explain run an OLS and get following form: Y=3 + 3X β = [+] 3 gets the following line PLS3100: Quantitative Methods In Political Science Carlos Costa 6/19 Part 2a Random Error Avoid OVB SPSS Graphing the points the line calculated by OLS PLS3100: Quantitative Methods In Political Science Carlos Costa 7/19 Part 2a Random Error Avoid OVB SPSS Error not exactly perfect notice that the line is really close but not exactly perfect if X is 1 ◦ Y=3 + 3X ◦ Y=3 + 3*1 = 6 [expected Y] actual Y = 7.24 error: variation between the expected value and the observed value PLS3100: Quantitative Methods In Political Science Carlos Costa 8/19 Part 2a Random Error Avoid OVB SPSS Graphing the points Difference between observed and expected PLS3100: Quantitative Methods In Political Science Carlos Costa 9/19 Part 2a Random Error Avoid OVB SPSS Error not exactly perfect expected Y = 6 actual Y = 7.24 error = Y[exp] − Y[obs] ◦ 7.24-6 = 1.24 calculate this difference for every data point end up with measure of error for every data point This error/variance must be RANDOM ◦ cannot be explained by any other IV ◦ these differences just happen to happen PLS3100: Quantitative Methods In Political Science Carlos Costa 10/19 Part 2a Random Error Avoid OVB SPSS Random Error navigating the claim Immigration reform has been a hot topic in the last decade. Pundits argue that support for comprehensive reform is widespread amongst law abiding citizens. claim: concern for rule of law is related to support for immigration reform DV: support for reform, IV concern for rule of law relationship is positive/direct PLS3100: Quantitative Methods In Political Science Carlos Costa 10/19 Part 2a Random Error Avoid OVB SPSS Random Error navigating the claim Immigration reform has been a hot topic in the last decade. Pundits argue that support for comprehensive reform is widespread amongst law abiding citizens. claim: concern for rule of law is related to support for immigration reform DV: support for reform, IV concern for rule of law relationship is positive/direct PLS3100: Quantitative Methods In Political Science Carlos Costa 11/19 Part 2a Random Error Avoid OVB SPSS Random Error navigating the claim Imagine that we run an OLS ◦ β is positive, high t value, reject the null are we satisfied ? ◦ i.o.w. are we confident that we have explained all there is to know about support for immigration ? NO ◦ diversity in the respondent’s home town ◦ crime level in the respondent’s home town ◦ proximity to the border ◦ respondent’s race ◦ respondent’s religion ◦ is respondent from sanctuary city PLS3100: Quantitative Methods In Political Science Carlos Costa 11/19 Part 2a Random Error Avoid OVB SPSS Random Error navigating the claim Imagine that we run an OLS ◦ β is positive, high t value, reject the null are we satisfied ? ◦ i.o.w. are we confident that we have explained all there is to know about support for immigration ? NO ◦ diversity in the respondent’s home town ◦ crime level in the respondent’s home town ◦ proximity to the border ◦ respondent’s race ◦ respondent’s religion ◦ is respondent from sanctuary city PLS3100: Quantitative Methods In Political Science Carlos Costa 12/19 Part 2a Random Error Avoid OVB SPSS Random Error The β from our OLS captures the systematic portion of the relationship that relates to concern for rule of law however, there are other systematic parts of the relationship that have not been explained ◦ systematic part related to above variables There is still variance that can be explained If it can be explained, it is not random PLS3100: Quantitative Methods In Political Science Carlos Costa 13/19 Part 2a Random Error Avoid OVB SPSS Random Error in our OLS, the part of the variance that is not explained is assumed to be random ◦ we know it isn’t OLS VIOLATION ◦ variables that should be included are omitted ◦ there is omitted variable bias ◦ OVB because of OVB, calculate β is unreliable PLS3100: Quantitative Methods In Political Science Carlos Costa 14/19 Part 2a Random Error Avoid OVB SPSS Avoiding OVB navigating the claim to avoid OVB we must ensure that all concepts that are systematically associated with the DV are present in the OLS regression Pundits argue that support for comprehensive reform is widespread amongst law abiding citizens. Diversity also counts, with greater diversity positively associated with support for reform Crime levels also count, with greater crime negatively associated with support for reform (. . . ) blah blah blah for every possible relationship Y = α + β(RofL)X(RofL) + β(Div)X(Div) + β(crm)X(crm) PLS3100: Quantitative Methods In Political Science Carlos Costa 14/19 Part 2a Random Error Avoid OVB SPSS Avoiding OVB navigating the claim to avoid OVB we must ensure that all concepts that are systematically associated with the DV are present in the OLS regression Pundits argue that support for comprehensive reform is widespread amongst law abiding citizens. Diversity also counts, with greater diversity positively associated with support for reform Crime levels also count, with greater crime negatively associated with support for reform (. . . ) blah blah blah for every possible relationship Y = α + β(RofL)X(RofL) + β(Div)X(Div) + β(crm)X(crm) PLS3100: Quantitative Methods In Political Science Carlos Costa 15/19 Part 2a Random Error Avoid OVB SPSS estimation process what to do in SPSS come up with a list of all IV variable to include estimate a single OLS using multiple IVs conduct separate Hypothesis test for each variable PLS3100: Quantitative Methods In Political Science Carlos Costa 16/19 Part 2a Random Error Avoid OVB SPSS Ensuring Random Error navigating the claim The Federal government has designated extra funds for school districts whose students score well in Science. Interested in securing these extra funds, Beavercreek school district asked several educators what they thought are the key factors impacting students’s science score. Math teachers argued that science scores are a product of how well students do in math. Literature teachers argued that it is a product of how well students can read. Social Studies teachers argued it is related to student’s scores in Social Studies. Last but not least, the Phys Ed teacher argued that boys just score better in Science than girls. Please analyze these claims. PLS3100: Quantitative Methods In Political Science Carlos Costa 17/19 Part 2a Random Error Avoid OVB SPSS Ensuring Random Error navigating the claim DV: science score IV: 4 ◦ [1] math score, positive ◦ [2] reading score, positive ◦ [3] social studies score, positive ◦ [4] girl, negative Y = α + β(math)X(math) + β(read)X(read) + β(soc)X(soc) + β(grl)X(grl) PLS3100: Quantitative Methods In Political Science Carlos Costa 18/19 Part 2a Random Error Avoid OVB SPSS Graphing the points Difference between observed and expected PLS3100: Quantitative Methods In Political Science Carlos Costa 19/19 Part 2a Random Error Avoid OVB SPSS Analyzing Error individual hypothesis testing math score: β = .389, t = 5.25 reject the null girl : β = 0 accept the null social studies :β = 0 accept the null reading score:β = .335, t = 4.6 reject the null We reject the claim that social studies scores or being a girl are related to