Formulating a Question
The goal of any social science is to explain social phenomena; in our case, we are trying to explain those socio-political phenomena things to do with government, the actions of voters or the actions of legislators.
What we are trying to explain are (typically) events of some kind.
For example:
–Why did (x) legislative districts shift towards the Republican party in 2016?
–Why did (x) legislative districts shift towards the Democratic party in 2018?
–Why did legislator (x) disagree with his own party on a particular issue?
The first step is simple but crucialyou have to identify the particular event or events that you are trying to explain.
The process of explanation in general (taken from Jon Elster, Explaining Social Behavior: More Nuts and Bolts for the Social Sciences. Cambridge: Cambridge University Press, 2015.)
(These steps will be explained in more detail below.)
Choose a theoryset of interrelated causal propositions. (e.g. The Identity Crisis theory of the 2016 Presidential Election: President Trumps election in 2016 was caused by his strategic appeal to white identity, which mobilized crucial subsets of the voting public on the basis of Islamophobia and ethno-cultural opposition to immigration.)
Specify a hypothesis, such that the explanandum (that which is to be explained) follows logically from the hypothesis A hypothesis is just an initial guess about what is likely to be the case, in light of the broader theory (e.g. shifts towards the GOP in the 2016 will be concentrated in counties with high levels of white voters who only have a high-school education.)
Identify plausible alternative explanations, and attempt to refute those explanations by specifying testable implications that are NOT observed (e.g. by looking at unemployment rates r median household incomes in pro-Trump counties.)
Exploring the implications of your hypothesis by showing it has testable implications that are in fact observed. (e.g. decline in GOP support amongst Latino voters; increases in GOP support do not occur in counties that had voted for the Democrats in previous two election cycles; counties with high numbers of high-school only whites trend towards Trump in other states, etc. )
THIS IS CRUCIAL TO REMEMBER: YOU ARE NOT BEING JUDGED ON WHETHER OR NOT YOU HAVE PROVEN YOUR HYPOTHESIS; YOU ARE JUDGED ON HOW YOU INVESTIGATE YOUR HYPOTHESIS.
It is also important to be careful to not overstate your conclusions. E.g. the fact that a county or district that voted for President Obama in 2012 and 2016 shifted to the GOP in 2016 does not disprove the Identity Crisis theory, though it should lessen our confidence that the theory is completely accurate.
The Process of Explanation: starting with a theory
Lets say that we are focussing on the first questionthe question of the shift towards the GOP in Presidential voting (not necessarily in terms of the popular vote, but in terms of the specific shifts in voting that led to the GOP victory in the electoral college.)
You might want to frame this question in terms of the various claims presented in the book Identity Crisis.
For instance, the differences between the 2012 election and the 2016 election probably cannot be explained as a result of a particular crisis (e.g. the financial-economic crises that preceded the 1932 and 2008 elections.
The most obvious general explanation for the shift is the peculiar character of the Trump campaigne.g. the way he promoted a populist or nationalist issue agenda.
Exploring a specific hypothesis (an implication of the general theory)
For this paper, we do not want to simply provide a general explanationwe want to explore why political changes have different consequences depending the peculiarities of particular place.
In other words, we want you to analyze the vote by breaking down to its constituent elementswhich will probably be county or congressional district level voting.
In regards to this particular question, this means analyzing the geography of the Trump votee.g. why did some counties or legislative districts shift towards the GOP in the 2016 election? Why was the nationalist or populist message successful in particular places?
To investigate this question, it would make sense to start in a state that is politically significant in Presidential elections we will select a swing state (a closely contested state that) that unexpectedly shifted to the GOP. We will select Michigan.
https://uselectionatlas.org/RESULTS/ (Links to an external site.)
(A map of the 2016 results, with state level results and links to other years.
As a first step, we might want to identify how the vote in Michigan changed between 2012 and 2016.
A simple way to do this is to look at the following map to determine where vote changes were occurring in Michigan.
https://www.cnbc.com/heres-a-map-of-the-us-counties-that-flipped-to-trump-from-democrats/ (Links to an external site.)
To get our investigation started, we will begin by selecting two counties where votes shifted by the most significant amountsMonroe County and Shiawassee county.
I want to get more information on the details of the vote in these two counties, so I download the county presidential election returns from the MIT election lab site to an excel file.
https://electionlab.mit.edu/data (Links to an external site.)
I arrange the data in the excel file by state by arranging the data by state (by clicking on the sort and filter button in excel. I then copy and paste all of the county level returns from Michigan from 2012 and 2016, and then copy and paste the returns from Monroe and Shiawassee from 2012 and 2016.
You should then have an excel sheet that looks something like this.
Year
County
Candidate
Votes for Candidate
Total Votes
“2012”
Monroe
Barack Obama
36310
72853
“2012”
Monroe
Mitt Romney
35593
72853
“2012”
Monroe
Other
950
72853
“2016”
Monroe
Hillary Clinton
26863
74218
“2016”
Monroe
Donald Trump
43261
74218
“2016”
Monroe
Other
4094
74218
“2012”
Shiawassee
Barack Obama
17197
33679
“2012”
Shiawassee
Mitt Romney
15962
33679
“2012”
Shiawassee
Other
520
33679
“2016”
Shiawassee
Hillary Clinton
12546
34111
“2016”
Shiawassee
Donald Trump
19230
34111
“2016”
Shiawassee
Other
2335
34111
What are we to make of these numbers? Not much, at least not yet.
We might notice that the shift towards Trump was not accompanied by a decline in overall vote totalsthere is a modest uptick in voting in both counties. How significant is this change?
We might want to see if we can determine the change in population in those counties by looking at the U.S. census bureau website:
https://www.census.gov/quickfacts/fact/table/monroecountymichigan,MI/PST045218 (Links to an external site.)
Keep that link in mind, as it will provide us with some other useful information on these and other counties.
What kind of information do we want to acquire about these counties?
We might want to start with some basic economic information, such as unemployment rates.
How do you get unemployment rates for counties?
Go to the U.S. Department of Agriculture website.
Click on the unemployment and median family income link.
Scroll down the next site and click on Michigan.
. And now you have the unemployment rates by county.
https://data.ers.usda.gov/reports.aspx?ID=17828 (Links to an external site.)
What are some other data that we might want to get, in addition to unemployment rates and median household income? What about some basic demographic data?
If we go back to the U.S. census site, you can get some basic data on demographics.
Exploring the Implications of your initial findings.
You have some basic data on these counties. How can you develop your investigation?
Compare the demographic and economic data of the two counties with each other, with the averages of Michigan and the United States using the U.S. Census bureau. To what extent, and in what way, are these counties outliers in either the USA or Michigan.
There are two main ways to extend the scope of your analysis: extend it in time, or extend it in space. Lets stick with time firstis it possible to learn anything by looking at how these counties voted prior to the 2012 Presidential election? This will allow us to consider whether the shift to the GOP was relatively abrupt, or whether there was a longer-term trend in the GOP direction. It may be the case that the peculiarities of the 2016 election (particularly regarding the two main Presidential candidates) were less significant than longer term trends. (You can look at the trends going back to 2000 by using the MIT election lab data.
Compare the two GOP-trending counties with two other Michigan counties where the shift was not as large (or perhaps in the other direction.) It would make sense to start by looking at the counties that are adjacent to Monroe and Shiawasee. Are there relatively obvious explanations for the differences between the counties that can explain the differences?
Similarly, you could extend your analysis to other GOP-trending counties in Michigan to see if you can identify a pattern. Do these counties all share roughly similar demographic-economic characteristics, or can they be classified in separate categories?
Extend your analysis to other swing states (or states that almost swung from the Democrats to the Republicans) such as Pennsylvania, Ohio, Florida, Wisconsin, Minnesota etc. or you could extend the analysis by looking at the relationship between voting patterns and demographic-economic factors in solidly Democratic or solidly Republican states.
There are literally thousands of different variations on this general approach.
