You are on page 1of 5

Efe Edevbie

POLSCI 311
Recreation of Mason’s Figure 1

Overview:

For the assignment, I chose to update Figure 1 from Lilliana Mason’s 2015 article “‘I

Disrespectfully Agree’: The Differential Effects of Partisan Sorting on Social and Issue

Polarization.” The following paper will include a data methods section detailing the various steps

taken to replicate the graph, a discussion portion explaining the new data’s relevance to the

original graph and its overall place within the article, and the updated graph, featuring data from

the 2008, 2012, and 2016 U.S. elections.

Data Methods:

Mason used the American National Election Studies (ANES) Time Series Cumulative

Data File throughout her paper and in the creation of Figure 1. With the Cumulative File being a

continuous record book for all of ANES’ Time Series studies, the Cumulative File I downloaded

from the ANES website (https://electionstudies.org/data-center/) contained all of the data that

was made available to Mason during the writing of her article, in addition to data from the three

Time Series studies (2008, 2012, and 2016) that were not included in Mason’s paper. With the

Cumulative File in tow, I used the 15th edition of Stata (a data science and statistic software

package) to upload the DTA file.

With Mason using a very specific line of coding in the creation of Figure 1 (reverse

coding to calculate the sorting score used in the graph, for example) it was vitally important in

my recreation of Figure 1 to replicate her very steps. As noted in “I Disrespectfully Agree,”


Mason’s replication files are stored in the American Journal of Political Science (AJPS)

Dataverse. Once in the Dataverse (https://dataverse.harvard.edu/dataverse/ajps), I was able to

search Mason’s article and find its corresponding replication files. Stored in the Dataverse are 1)

Mason’s Stata coding “do” file for the ANES Cumulative File (labeled “Coding cumulative

ANES.do”), 2) her “do” file for the tables and figures used in her paper (labeled “Mason

Cumulative ANES tables and figures.do”), and 3) Mason’s post-coding State file for the ANES

Cumulative File (labeled “NES cumulative coded by author_1.dta”).

After downloading each of three files listed above, the next step towards replicating

Mason’s figure was to implement her coding “do” file into the ANES Cumulative File. Opening

the file in Word Pad, I copied and pasted each command from the file into the Stata document I

had used to implement the ANES Cumulative file and hit enter. One coding command, labeled

“omscore,” had to be downloaded into Stata’s software. To do this, I typed out the command

“findit omscore” into Stata, and from there I was able to download the “omscore” command

from Stata’s database.

With all of Mason’s coding commands implemented, the dataset was ready to be used in

my graph replication. Now opening the “do” file for her tables and figures in Word Pad, I was

able to find the specific commands Mason had used in her creation of Figure 1. With the

variables “pidstr” and “idcomplexity” being highlighted, I copied and pasted the Figure 1

demands into the Stata document. For the pidstr commands, I updated them to reflect the new

years of Time Series studies needed (typing in “tab pidstr if year==2012,” instead of typing “tab

pidstr if year==2004,” for example). This was not needed for the idcomplexity variable.
With the pidstr variable, the percentage corresponding with 0 represented the percentage

of pure independents shown in Figure 1, and the percentage corresponding with 1 represented the

percentage of strong partisans. With the idcomplexity variable, the percentage corresponding

with the mean represented the sorting score shown in Figure 1. I copied and pasted each of these

data points from the years 2008, 2012, and 2016, and inserted them into a Microsoft Excel

worksheet. No further use of Stata or the “do” files was needed after this.

With the data points inputted into Excel, I was able to create a 3-line graph close to that

of the one Mason made. The individual years served as the X-axis, with the 0-1 scale serving as

the Y-axis.

Discussion:

One of Mason’s main overarching points in “‘I Disrespectfully Agree’: The Differential

Effects of Partisan Sorting on Social and Issue Polarization,” is that there’s a significant

difference between social polarization and issue position polarization, and that while Americans

may agree on many political topics (lower issue position polarization), our country has grown

more and more apart because of a higher level of social polarization — in which Americans have

continued to sort themselves into political parties that are increasingly becoming more aligned

with distinct ideologies, creating for stronger, more in-line political identities (Mason 128).

Figure 1 mainly works to commentate on this concept of social polarization and how it

has changed over time (with Figure 2 tackling issue position polarization). As shown In Mason’s

version of Figure 1, the percentage of strong partisans has noticeably increased over time (a more

than 10 point increase from 1972 to 2004), along with the sorting percentage (a near 10 point
increase in the same time period). Conversely, the percentage of pure independents has seen a

dip — a near 10 point decline from the mid 70s to the start of the 1990s, with fluctuating peaks

and troughs all the way through 2004.

In my version of Figure 1, the percentage of strong partisans has increased from its 2004

position, climbing from roughly 35% to roughly 38% in 2016. Sorting has also increased along

with it, with a 2004 score of roughly 28% increasing to roughly 38% in 2016. This remains

relatively consistent with Mason’s findings.

The percentage of pure independents, however, has also seen an increase in recent years,

leaping from around 7% in 2004 to just under 15% in 2016. While this seems like a bit of a buck

to Mason’s overall findings — the percentage of pure independents in Mason’s figure had been

shrinking overall over time — it potentially adds an interesting new point to Mason’s 2015

article, while still giving way to social polarization. With the amount of strong partisans and pure

independents both increasing over the past twelve years of data, this must mean that weak

partisans and/or “leaners” are decreasing in amount. Thus, those within the Democratic and

Republican parties are increasingly moving towards the more polarized ends of their respective

sides of the political scale. Further research within this particular area of social polarization

could be a great addition to the political science field if the trend continues.
Graph:

Works Referenced:

Mason, Lilliana. 2015. “’I Disrespectfully Agree’: The Differential Effects of Partisan

Sorting on Social and Issue Polarization.” American Journal of Political Science.

59(Jan):128–145.

You might also like