library(poliscidata)Registered S3 method overwritten by 'gdata':
method from
reorder.factor gplotslibrary(dplyr)
Attaching package: 'dplyr'The following objects are masked from 'package:stats':
filter, lagThe following objects are masked from 'package:base':
intersect, setdiff, setequal, unionlibrary(tidyr)
library(stats)
library(ggplot2)##Intro and Background Info
I’ve had more time to think about my final, and I’ve decided to simplify my research topic. Rather than look at ethnic conflict, I’m going to look at how internet usage affects a people’s confidence in their country (institutions and governance). In order to focus on this relationship, I will be using the ‘world’ data set in the ‘poliscidata’ package, and I will be using the following variables: ‘country’, ‘unnetuse’ (internet usage per 100 people), confidence (population’s confidence in their country’s institutions; scaled out of 100), and ‘effectiveness’ (government effectiveness scale).
Below is a general presentation of the data I will be using, filtered down as I described.
library(poliscidata) %>%
data("world")Warning in data(., "world"): data set '.' not foundworld %>%
select(country,regime_type3,unnetuse,confidence,effectiveness) country regime_type3 unnetuse confidence
1 Afghanistan Dictatorship 1.7 NA
2 Albania Parliamentary democ 23.9 49.335926
3 Algeria Dictatorship 11.9 52.055735
4 Angola Dictatorship 3.1 NA
5 Argentina Presidential democ 28.1 7.299325
6 Armenia <NA> 6.2 27.132735
7 Australia Parliamentary democ 70.8 46.838886
8 Austria Parliamentary democ 71.2 49.680190
9 Azerbaijan Dictatorship 28.2 54.168485
10 Bahrain Dictatorship 51.9 NA
11 Bangladesh Parliamentary democ 0.3 71.644389
12 Belarus Dictatorship 32.1 15.645705
13 Belgium Parliamentary democ 68.1 43.315162
14 Benin Presidential democ 1.8 NA
15 Bhutan Dictatorship 6.6 NA
16 Bolivia Presidential democ 10.8 NA
17 Bosnia and Herzegovina Dictatorship 34.7 70.204136
18 Botswana Dictatorship 6.2 NA
19 Brazil Presidential democ 37.5 32.501337
20 Bulgaria Parliamentary democ 34.7 38.976006
21 Burkina Faso Dictatorship 0.9 NA
22 Burma (Myanmar) <NA> 0.2 NA
23 Burundi Dictatorship 0.8 NA
24 Cambodia Dictatorship 0.5 NA
25 Cameroon Dictatorship 3.8 NA
26 Canada Parliamentary democ 75.3 58.203871
27 Cape Verde Parliamentary democ 20.6 NA
28 Central African Republic <NA> 0.4 NA
29 Chad Dictatorship 1.2 NA
30 Chile Presidential democ 32.5 58.368495
31 China Dictatorship 22.5 73.788814
32 Colombia Presidential democ 38.5 NA
33 Comoros Dictatorship 3.6 NA
34 Congo, Democratic Republic of the <NA> NA NA
35 Congo, Republic of the <NA> 4.3 NA
36 Costa Rica Presidential democ 32.3 NA
37 Cote d'Ivoire Presidential democ 3.2 NA
38 Croatia <NA> 50.5 44.175465
39 Cuba Dictatorship 12.9 NA
40 Cyprus Presidential democ 38.8 NA
41 Czech Republic Parliamentary democ 57.8 35.172868
42 Denmark Parliamentary democ 83.3 63.137017
43 Djibouti Dictatorship 2.3 NA
44 Dominican Republic <NA> 21.6 NA
45 Ecuador Dictatorship 28.8 NA
46 Egypt Dictatorship 16.6 59.227665
47 El Salvador Presidential democ 10.6 41.978817
48 Equatorial Guinea Dictatorship 1.8 NA
49 Eritrea Dictatorship 4.1 NA
50 Estonia Parliamentary democ 66.2 39.182308
51 Ethiopia Dictatorship 0.4 NA
52 Fiji Dictatorship 12.2 NA
53 Finland Parliamentary democ 82.0 46.875924
54 France <NA> 67.9 55.116509
55 Gabon Dictatorship 6.2 NA
56 Gambia, The <NA> 6.9 NA
57 Georgia Dictatorship 23.8 38.627903
58 Germany Parliamentary democ 75.5 52.728927
59 Ghana Presidential democ 4.3 NA
60 Greece Parliamentary democ 43.1 NA
61 Guatemala Presidential democ 14.3 NA
62 Guinea Dictatorship 0.9 NA
63 Guinea-Bissau Presidential democ 2.4 NA
64 Guyana Presidential democ 26.9 NA
65 Haiti <NA> 10.1 NA
66 Honduras Presidential democ 13.1 NA
67 Hong Kong <NA> 67.0 NA
68 Hungary Parliamentary democ 58.5 45.543249
69 Iceland <NA> 59.0 63.628568
70 India Parliamentary democ 4.5 62.205316
71 Indonesia Presidential democ 7.9 50.531385
72 Iran Dictatorship 32.0 57.685353
73 Iraq Parliamentary democ 1.0 NA
74 Ireland Parliamentary democ 62.7 68.497642
75 Israel Parliamentary democ 47.9 NA
76 Italy Parliamentary democ 41.8 37.016570
77 Jamaica Parliamentary democ 57.3 NA
78 Japan Parliamentary democ 75.2 37.033760
79 Jordan Dictatorship 27.0 73.628692
80 Kazakhstan Dictatorship 10.9 NA
81 Kenya Presidential democ 8.7 NA
82 Korea, North <NA> 0.0 NA
83 Korea, South <NA> 75.8 NA
84 Kuwait Dictatorship 36.7 NA
85 Kyrgyzstan Dictatorship 16.1 NA
86 Laos Dictatorship 8.5 NA
87 Latvia Parliamentary democ 60.4 39.220066
88 Lebanon Dictatorship 22.5 NA
89 Lesotho Parliamentary democ 3.6 NA
90 Liberia Dictatorship 0.5 NA
91 Libya Dictatorship 5.1 NA
92 Lithuania <NA> 54.4 48.126086
93 Luxembourg Parliamentary democ 79.2 NA
94 Macedonia Parliamentary democ 41.5 12.330052
95 Madagascar <NA> 1.7 NA
96 Malawi Presidential democ 2.1 NA
97 Malaysia Dictatorship 55.8 NA
98 Mali <NA> 1.6 NA
99 Malta Parliamentary democ 48.3 NA
100 Mauritania Dictatorship 1.9 NA
101 Mauritius Parliamentary democ 22.2 NA
102 Mexico Presidential democ 22.2 26.040388
103 Moldova <NA> 23.4 42.191709
104 Mongolia <NA> 12.5 NA
105 Montenegro <NA> 47.2 NA
106 Morocco Dictatorship 33.0 NA
107 Mozambique Dictatorship 1.6 NA
108 Namibia Presidential democ 5.3 NA
109 Nepal Parliamentary democ 1.7 NA
110 Netherlands Parliamentary democ 87.0 58.481013
111 New Zealand Parliamentary democ 71.4 38.105963
112 Nicaragua Presidential democ 3.3 NA
113 Niger <NA> 0.5 NA
114 Nigeria Presidential democ 15.9 59.358170
115 Norway Parliamentary democ 82.5 66.453811
116 Oman Dictatorship 20.0 NA
117 Pakistan Dictatorship 11.1 51.238722
118 Palestine <NA> 9.0 NA
119 Panama Presidential democ 27.5 NA
120 Papua New Guinea Parliamentary democ 1.8 NA
121 Paraguay Dictatorship 14.3 NA
122 Peru Dictatorship 24.7 6.494946
123 Philippines Presidential democ 6.2 63.305240
124 Poland <NA> 49.0 65.605741
125 Portugal <NA> 42.1 37.364882
126 Qatar Dictatorship 34.0 NA
127 Romania <NA> 28.8 26.884683
128 Russia <NA> 31.9 41.145354
129 Rwanda Dictatorship 3.1 NA
130 Saudi Arabia Dictatorship 31.5 NA
131 Senegal <NA> 8.4 NA
132 Serbia <NA> 44.9 NA
133 Sierra Leone Presidential democ 0.3 NA
134 Singapore Dictatorship 69.6 NA
135 Slovakia Parliamentary democ 66.0 25.269169
136 Slovenia Parliamentary democ 55.7 39.855957
137 South Africa Parliamentary democ 8.6 66.400869
138 Spain Parliamentary democ 55.4 42.274258
139 Sri Lanka <NA> 5.8 NA
140 Sudan Dictatorship 10.2 NA
141 Suriname Presidential democ 9.7 NA
142 Swaziland Dictatorship 6.9 NA
143 Sweden Parliamentary democ 87.7 56.535110
144 Switzerland Presidential democ 75.9 56.553995
145 Syria Dictatorship 17.3 NA
146 Taiwan <NA> NA 49.197820
147 Tajikistan Dictatorship 8.8 NA
148 Tanzania Dictatorship 1.2 78.852623
149 Thailand Parliamentary democ 23.9 NA
150 Timor-Leste <NA> NA NA
151 Togo Dictatorship 5.4 NA
152 Trinidad and Tobago <NA> 17.0 NA
153 Tunisia Dictatorship 27.1 NA
154 Turkey Parliamentary democ 34.4 57.070574
155 Turkmenistan Dictatorship 1.5 NA
156 Uganda Dictatorship 7.9 70.747317
157 Ukraine <NA> 10.5 36.022627
158 United Arab Emirates <NA> 65.2 NA
159 United Kingdom Parliamentary democ 76.0 56.467760
160 United States Presidential democ 75.9 61.785208
161 Uruguay Presidential democ 40.2 43.417722
162 Uzbekistan Dictatorship 9.0 NA
163 Venezuela Presidential democ 25.7 21.375093
164 Vietnam Dictatorship 24.2 99.862411
165 Yemen Dictatorship 1.6 NA
166 Zambia Presidential democ 5.5 NA
167 Zimbabwe Dictatorship 11.4 60.019029
effectiveness
1 13.711584
2 35.460993
3 32.624114
4 19.148937
5 34.988179
6 36.643026
7 90.070923
8 88.888888
9 23.877069
10 65.011820
11 34.042554
12 22.222223
13 90.307329
14 31.914894
15 68.557920
16 34.042554
17 25.295509
18 67.139480
19 41.371158
20 45.153664
21 30.260047
22 NA
23 12.056737
24 33.333333
25 31.914894
26 91.016548
27 41.843972
28 12.765959
29 28.841608
30 74.704493
31 50.827423
32 37.352246
33 26.713949
34 NA
35 NA
36 55.319149
37 25.531915
38 51.063830
39 40.425532
40 70.212766
41 63.120567
42 93.617022
43 25.768322
44 NA
45 23.877069
46 39.007092
47 34.042554
48 14.184397
49 36.170213
50 65.011820
51 25.531915
52 47.990544
53 94.089834
54 86.052008
55 35.933806
56 NA
57 28.368795
58 88.179669
59 46.808511
60 65.248227
61 32.151300
62 28.132388
63 14.657210
64 39.007092
65 9.692673
66 29.314420
67 NA
68 65.011820
69 93.380615
70 43.498818
71 33.333333
72 35.697399
73 7.801419
74 84.869976
75 70.685579
76 68.085107
77 44.917258
78 71.867614
79 55.082743
80 27.659574
81 26.477541
82 NA
83 NA
84 50.354610
85 27.423168
86 27.659574
87 62.411348
88 36.879433
89 40.425532
90 10.874705
91 26.004728
92 60.992908
93 96.926717
94 37.352246
95 37.588653
96 30.496454
97 68.321513
98 26.713949
99 73.995271
100 42.789598
101 59.101654
102 50.118203
103 31.678487
104 42.316785
105 NA
106 48.226950
107 36.879433
108 50.827423
109 34.515367
110 97.163123
111 93.144209
112 26.004728
113 27.895981
114 20.094563
115 90.070923
116 62.884161
117 34.751773
118 NA
119 43.262411
120 28.132388
121 16.075651
122 35.460993
123 45.153664
124 60.992908
125 70.921985
126 62.884161
127 38.770685
128 37.115839
129 27.186761
130 45.390071
131 42.316785
132 NA
133 10.165486
134 100.000000
135 56.028369
136 65.957447
137 58.865248
138 82.742316
139 47.281324
140 20.330969
141 42.789598
142 36.170213
143 90.070923
144 100.000000
145 33.096927
146 70.212766
147 17.494089
148 34.515367
149 53.191489
150 NA
151 18.912531
152 NA
153 61.938534
154 41.843972
155 11.820330
156 36.879433
157 29.078014
158 NA
159 94.562647
160 86.761230
161 58.628841
162 20.567375
163 19.621750
164 40.189125
165 26.004728
166 24.586288
167 27.659574The internet and media consumption play a massive role in politics now; at this point, it’s pretty much expected to find out what is happening in the world on the internet (newspapers, social media, etc.). And as we’ve seen since the 2016 election, the internet also plays a substantial role in domestic politics, whether it be through neutral journalism, activism, whistle-blowing, or misinformation. It is both good and bad, and as Stanford professor Evgeny Morozov writes, the internet’s impact “will depend on individual conditions” such as a given country’s political atmosphere (peaceful, volatile, conservative, liberal, etc.), ease of access to the internet, and how united a population is either in favor or against their current government (“The Internet, Politics and the Politics of Internet Debate”). So, what /is/ the relationship between internet use and domestic politics?
Based on my personal and academic experience, I hypothesize that while internet usage definitely does affect people’s confidence in their government, I do not think that it automatically hinders government effectiveness. If a certain country’s subjects have extremely little confidence in its government, the government would have no legitimacy (and would therefore be ineffective); a government’s power largely comes from its subjects’ willingness to accept it. However, there are instances where a country’s population has little confidence in its government with no effect on its efficacy. But, again, for this project I will be focusing on how internet usage impacts confidence in a given regime.
Now to transcribe my research question: Does internet usage (social media, news, etc.) have an impact on people’s confidence in their government (‘confidence’)? If so, is said government more or less effective in its duties?
First, lets look at some summary statistics.
world_filter <- world %>%
select(country,regime_type3,confidence,effectiveness,unnetuse)
summary(world_filter) country regime_type3 confidence effectiveness
Afghanistan: 1 Dictatorship :61 Min. : 6.495 Min. : 7.801
Albania : 1 Parliamentary democ:41 1st Qu.:38.889 1st Qu.: 28.132
Algeria : 1 Presidential democ :31 Median :49.508 Median : 39.007
Angola : 1 NA's :34 Mean :48.900 Mean : 46.036
Argentina : 1 3rd Qu.:59.523 3rd Qu.: 62.884
Armenia : 1 Max. :99.862 Max. :100.000
(Other) :161 NA's :99 NA's :14
unnetuse
Min. : 0.00
1st Qu.: 5.25
Median :18.65
Mean :26.78
3rd Qu.:42.35
Max. :87.70
NA's :3 regime_type3 – There are 167 entries in the filtered down dataset I am using. Dividing that, 61 countries are classified as dictatorships, 41 as a parliamentary democracy, 34 N/A responses/unclassified regimes, and 31 presidential democracies. Not having doen any statistical analysis yet I can’t speak for the values’ mathematical significance, but politically, it is important to note that 61/167 countries in the data set are classified as dictatorships.
confidence – Confidence scores in this data set are scaled from 1-100. The minimum confidence score is 6.495 and the maximum, obviously is 100.000. The mean value, 48.508, is below 50.000 which indicates that there are slightly more scores under than 50.000 than over. The median is 49.508 and the 1st and 3rd quantiles are 38.889 and 59.523 respectively. However, these values are not so low that there’s an implication of a widespread (and severe) sense that peoples’ governments are incompetent.
effectiveness – Similar to ‘confidence’, ‘effectiveness’ is scaled out of 100. Unlike ‘confidence’, however, this variable presents a numerical representation of how effective regimes are at governance (actual governance, welfare, infrastructure/services). The summary values for this variable are, overall, lower than their counterparts under ‘confidence’. The minimum value is 7.801, and the mean is 46.036. The median, 1st, and 3rd quantiles are 39.007, 28.132, and 62.884 respectively. While currently untested, these figures indicate that whether or not people have confidence in their government, said government is not necessarily effective to the same degree.
unnetuse – Unlike the previous 2 variables, this one is scaled as a percentage of each country’s total population, specifically the number of internet users per 100 people. This variable is more practically tangible; the values are calculated as a percentage, but since it’s also out of 100 people, it’s easy to conceptualize the 1st quantile value for instance: 5.25, which would be about 5 people per 100. On that note, the maximum value, 87.70, is indicative of a high percentage of internet users in a given state.
I will start with some preliminary graphs/visualizations.
ggplot(data=world_filter, aes(x=unnetuse,y=confidence, color=regime_type3)) + geom_point() Warning: Removed 100 rows containing missing values (geom_point).
This scatterplot, which measures the relationship between internet usage and popular confidence in government, has no obvious relationship or correlation. The points are grouped by color based on regime type, but there is no notable increase or decrease in the points’ location as a variable increases or decreases. The vast majority of values are located in the middle of the graph.
ggplot(data=world_filter, aes(x=confidence,y=effectiveness, color=regime_type3)) + geom_point() Warning: Removed 99 rows containing missing values (geom_point).
While there still isn’t a traditional positive or negative relationship between ‘confidence’ and ‘effectiveness’, it is noteworthy that the majority of values are in the center of the graph. They have a wide range, but they are mostly within 25 and 75 on the x-axis, which does make sense giving that the 1st and 3rd quantiles are included in the initial summary statistics.
ggplot(world_filter, aes(x=unnetuse,y=confidence,color=regime_type3)) + geom_point() + stat_smooth(method="lm",col="blue")`geom_smooth()` using formula 'y ~ x'Warning: Removed 100 rows containing non-finite values (stat_smooth).Warning: Removed 100 rows containing missing values (geom_point).
This visualization is a recreation of the first scatterplot, but with a multiple regression line. The line, at first glance, is straight, but upon closer inspection it ever so slightly slopes down toward the right. This indicates a slightly negative relationship between internet usage (x) and confidence in government institutions (y).
ggplot(world_filter, aes(x=unnetuse,y=effectiveness,color=regime_type3)) + geom_point() + stat_smooth(method="lm",col="blue")`geom_smooth()` using formula 'y ~ x'Warning: Removed 15 rows containing non-finite values (stat_smooth).Warning: Removed 15 rows containing missing values (geom_point).
The above scatterplot, however, depicts a very clear positive relationship between variables ‘unnetuse’ (x) and ‘effectiveness’ (y). This is indicative of a strong relationship between internet use and government effectiveness. There is what appears to be a heavier concentration of points below 50 on the x-axis, but the general trend is pretty clear that government effectiveness increases as nationals’ internet use does. Although ‘unnetuse’ comes from a 2008 survey, I would argue that this trend still holds up; especially due to COVID-19, the main way that politicians engage with their constituents is on the internet (social media). There is not really a culture of “visit your politician and talk to them” culture anymore, the new norm is for sucht things to be done over the phone, over email, or in some form, over the internet.
My next step is to go into modeling. I’m going to start with a linear regression model. ‘confidence’ is the dependent variable, and ‘unnetuse’, ‘effectiveness’, and ‘regime_type3’ are all independent.
confidence_lm <- lm(confidence~unnetuse+effectiveness+regime_type3, data=world_filter)
confidence_lm
Call:
lm(formula = confidence ~ unnetuse + effectiveness + regime_type3,
data = world_filter)
Coefficients:
(Intercept) unnetuse
43.0725 -0.5805
effectiveness regime_type3Parliamentary democ
0.7574 -15.4561
regime_type3Presidential democ
-19.0101 And here is a visualization of the above values.
plot(fitted(confidence_lm))
abline(h=50, lty=2)
summary(confidence_lm)
Call:
lm(formula = confidence ~ unnetuse + effectiveness + regime_type3,
data = world_filter)
Residuals:
Min 1Q Median 3Q Max
-49.097 -6.527 0.075 7.993 40.399
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 43.0725 5.6175 7.668 4.82e-10 ***
unnetuse -0.5805 0.1561 -3.719 0.000498 ***
effectiveness 0.7574 0.1700 4.455 4.61e-05 ***
regime_type3Parliamentary democ -15.4561 6.3577 -2.431 0.018608 *
regime_type3Presidential democ -19.0101 6.1730 -3.080 0.003336 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 15.37 on 51 degrees of freedom
(111 observations deleted due to missingness)
Multiple R-squared: 0.3376, Adjusted R-squared: 0.2856
F-statistic: 6.497 on 4 and 51 DF, p-value: 0.0002642I’m honestly not entirely sure what these values mean, but I input what I had into the function and I got results.
AIC(confidence_lm)[1] 471.6936BIC(confidence_lm)[1] 483.8457