Rate my election analysis. Note, I have been out of college for over 8 years and I only took a couple pstat classes. I am asking for peer review.
Compare changes in voter turnout between 2016 - 2020.
Step 1: Get the vote totals for every county in a state for 2016 and 2020 general elections. You will have four columns of data. Hillary 2016, Trump 2016. Biden 2020, Trump 2020
Step 2: Subtract the 2016 numbers per candidate from the 2020 numbers for the corresponding candidate. This will give you your 5th and 6th columns. Column 5 is difference between Hillary turnout and Biden turnout. Column 6 is difference between Trump turnout between 2016 and 2020.
Step 3. Divide column 5 (difference between 2016 and 2020) by column 1 (Hillary turnout). Repeat for Trump % difference and Trump turnout 2016 This will give you a percentage, per county, difference between election turnouts.
Step 4: Calculate average of % change voter turnout and standard deviation of % change. You will have 4 values, average change for Biden, average change for trump, standard deviation of biden change, and standard deviation of trump change.
Step 5: Repeat this process for all 50 states. Take your 50 Trump standard deviations and 50 Biden standard deviations, and calculate the standard deviation of both of these data sets.
You now have a national average standard deviation, using every county's vote totals, which you can compare to the standard deviations of percentage changes in contested states and use statistical analysis to determine how likely it is that any particular county went so much harder for Trump or Biden compared to national averages. This could help rule out fraud or point to specific counties where an investigation is warranted.
Thoughts?
Compare changes in voter turnout between 2016 - 2020.
Step 1: Get the vote totals for every county in a state for 2016 and 2020 general elections. You will have four columns of data. Hillary 2016, Trump 2016. Biden 2020, Trump 2020
Step 2: Subtract the 2016 numbers per candidate from the 2020 numbers for the corresponding candidate. This will give you your 5th and 6th columns. Column 5 is difference between Hillary turnout and Biden turnout. Column 6 is difference between Trump turnout between 2016 and 2020.
Step 3. Divide column 5 (difference between 2016 and 2020) by column 1 (Hillary turnout). Repeat for Trump % difference and Trump turnout 2016 This will give you a percentage, per county, difference between election turnouts.
Step 4: Calculate average of % change voter turnout and standard deviation of % change. You will have 4 values, average change for Biden, average change for trump, standard deviation of biden change, and standard deviation of trump change.
Step 5: Repeat this process for all 50 states. Take your 50 Trump standard deviations and 50 Biden standard deviations, and calculate the standard deviation of both of these data sets.
You now have a national average standard deviation, using every county's vote totals, which you can compare to the standard deviations of percentage changes in contested states and use statistical analysis to determine how likely it is that any particular county went so much harder for Trump or Biden compared to national averages. This could help rule out fraud or point to specific counties where an investigation is warranted.
Thoughts?
