Homework1
Jordan Glendrange
February 6, 2021
Dataset
The data set I chose is titled “How Unpopular is Donald Trump?” The link can be found here: https://projects.fivethirtyeight.com/trump-approval-ratings/
The dataset is hosted on my github account.
approval_poll <- read.csv("https://raw.githubusercontent.com/jglendrange/DATA607/main/approval_polllist.csv", TRUE, ",")
head(approval_poll)## president subgroup modeldate startdate enddate pollster grade
## 1 Donald Trump All polls 1/20/2021 1/20/2017 1/22/2017 Morning Consult B/C
## 2 Donald Trump All polls 1/20/2021 1/20/2017 1/22/2017 Gallup B
## 3 Donald Trump All polls 1/20/2021 1/20/2017 1/24/2017 Ipsos B-
## 4 Donald Trump All polls 1/20/2021 1/21/2017 1/23/2017 Gallup B
## 5 Donald Trump All polls 1/20/2021 1/22/2017 1/24/2017 Gallup B
## 6 Donald Trump All polls 1/20/2021 1/21/2017 1/25/2017 Ipsos B-
## samplesize population weight influence approve disapprove adjusted_approve
## 1 1992 rv 0.6800286 0 46.0 37.0 45.68678
## 2 1500 a 0.2623230 0 45.0 45.0 45.86144
## 3 1632 a 0.1534812 0 42.1 45.2 43.45156
## 4 1500 a 0.2428446 0 45.0 46.0 45.86144
## 5 1500 a 0.2273795 0 46.0 45.0 46.86144
## 6 1651 a 0.1415310 0 42.3 45.8 43.65156
## adjusted_disapprove multiversions tracking
## 1 38.05580 NA
## 2 43.53919 TRUE
## 3 43.78039 TRUE
## 4 44.53919 TRUE
## 5 43.53919 TRUE
## 6 44.38039 TRUE
## url
## 1 http://static.politico.com/9b/13/82a3baf542ae9018e5b6e1008379/170103-topline-politico-v3-kd.pdf
## 2 http://www.gallup.com/poll/201617/gallup-daily-trump-job-approval.aspx
## 3 http://polling.reuters.com/#poll/CP3_2/
## 4 http://www.gallup.com/poll/201617/gallup-daily-trump-job-approval.aspx
## 5 http://www.gallup.com/poll/201617/gallup-daily-trump-job-approval.aspx
## 6 http://polling.reuters.com/#poll/CP3_2/
## poll_id question_id createddate timestamp
## 1 49249 77261 1/23/2017 11:47:59 20 Jan 2021
## 2 49253 77265 1/23/2017 11:47:59 20 Jan 2021
## 3 49426 77599 3/1/2017 11:47:59 20 Jan 2021
## 4 49262 77274 1/24/2017 11:47:59 20 Jan 2021
## 5 49236 77248 1/25/2017 11:47:59 20 Jan 2021
## 6 49425 77598 3/1/2017 11:47:59 20 Jan 2021The column names are pretty clear and straighforward. I decided to drop some columns due to data redundancy and mostly null values.
wanted_columns <- c("pollster","subgroup","startdate","enddate","grade","samplesize","population","adjusted_approve","adjusted_disapprove","poll_id","question_id","createddate")
approval_poll <- approval_poll[wanted_columns]
head(approval_poll)## pollster subgroup startdate enddate grade samplesize population
## 1 Morning Consult All polls 1/20/2017 1/22/2017 B/C 1992 rv
## 2 Gallup All polls 1/20/2017 1/22/2017 B 1500 a
## 3 Ipsos All polls 1/20/2017 1/24/2017 B- 1632 a
## 4 Gallup All polls 1/21/2017 1/23/2017 B 1500 a
## 5 Gallup All polls 1/22/2017 1/24/2017 B 1500 a
## 6 Ipsos All polls 1/21/2017 1/25/2017 B- 1651 a
## adjusted_approve adjusted_disapprove poll_id question_id createddate
## 1 45.68678 38.05580 49249 77261 1/23/2017
## 2 45.86144 43.53919 49253 77265 1/23/2017
## 3 43.45156 43.78039 49426 77599 3/1/2017
## 4 45.86144 44.53919 49262 77274 1/24/2017
## 5 46.86144 43.53919 49236 77248 1/25/2017
## 6 43.65156 44.38039 49425 77598 3/1/2017I decided to take a look at the approval and disapproval rattings via a historgram
hist(approval_poll$adjusted_approve)
hist(approval_poll$adjusted_disapprove)
You can see easily see that the disapproval histogram skews further to the right.
For fun I decided to plot the dissapproval rating against the approval rating. I expected to see a 1 to 1 relationship, but was suprised to see a few outliers. It would be an interesting follow up to see which pollsters reported those numbers.
plot(approval_poll$adjusted_disapprove, approval_poll$adjusted_approve)
Conclusion
After reviewing the data set in the article Donald Trump held an average approval rating of ~42%. The only point Trump had an approval rating higher than his disapproval rating was at the very begining of his presidenciy. Furthermore at the end of his presidency there was a 20% gap between his approval and disapproval rating.
In terms of follow ups I would like to investigate - It would be interesting to dig into what each pollster is reporting to try and tease out any biases.