by Leah Liu
This led me to think—wouldn’t
controlling the Internet backfire for the governments? Since it is very
difficult to completely isolate a country’s network from the outside, at least some
people within the country will have access to denouncing opinions of their own
government. Blocking those voices shows the government’s lack of confidence in
itself as well as in its people. Wouldn’t people in return have less faith in
their government? Or do they still genuinely support their leaders because the
majority of the country are not exposed to unfavorable information of the
government? To find out the answer, I conducted the research and evaluated the
effect of Internet freedom on public opinion towards their governments by
comparing data of 42 countries in the world. The findings could give us better
insight of the rationality behind government censorship over the Internet and
possible policy implications.
To measure the public opinion of the
governments, I pulled the 2014 “Confidence in National Government” data from
the World Poll Survey of Gallup Analytics. The company did the annual survey in
over a hundred countries. For the scores of “Confidence in National
Government”, they measure the percentage of respondents with a positive answer to
the question “In this country, do you have confidence in the national
government?” and make a ranking of all surveyed countries. For the level of Internet
freedom, I used the Freedom On the Net scores from Freedom House. They rated 65
countries in 2014 through the examination of three categories: governmental
efforts to block certain technologies; governmental efforts to limit certain
contents; legal protection on Internet users’ rights. Higher score means
greater restriction over the Internet. Freedom House has so far the most
comprehensive data on Internet freedom, but the total number of the countries
they surveyed was not ideal for a quantitative research. Also,
cross-referencing to “Confidence in National Government” data eliminated some
choices in countries, thus the 42 observations in our study.

Besides the dependent and independent
variables, I also included three control variables: Internet accessibility,
economic development and regime type. These variables may also have an effect
on the popular support of governments, and by controlling them in the data
analysis, we can have more accurate result on the relations between Internet
freedom and popular support of governments. The accessibility of Internet is
measured by the number of Internet users per 100 people. Different from the
accessibility of Internet content, this serves as an indicator of government’s
performance on providing infrastructure. Together with economic development, as
measured in GDP per capita, Internet accessibility affects people’s
satisfaction with their governments. The third control variable is regime type
and the score of the variable ranges from +10 (full democracy) to -10 (full
autocracy). Countries with a democratic government generally enjoy greater
support and are less likely to impose restriction over Internet freedom.
Figure 1 shows the bivariate relationship
between Internet freedom and confidence in government. As indicated by the
graph, there is a positive correlation and the increase in Internet restriction
score is associated with the increase in government support.

Table 1 shows
us how much the independent variable and control variables determined people’s
confidence in their governments. Model 1 presents the unconditional effects,
and Model 3 presents conditional effects. Without taking into consideration of
other variables, every 1 point increase in Internet restriction is associated
with 0.005-unit increase in confidence in national governments. The result is
statistically significant (p<0.01) which means that there is only 1%
possibility that the observed relationship happened purely by chance. After
conditioning the effect of Internet accessibility, economic development and
regime type on popular support of the government, the effect of our independent
variable decreases, though not significantly, to 0.004. The result is still
statistically significant. It is also worth mentioning that an increase in
democracy score is associated with a slight decrease in government support,
though the result is not statistically significant.

The finding seems to support the rationality behind
controlling the Internet. What may be the possible explanation? First, the
distrust of government stems from all social aspects. Factors like uncertain
economic development, endangered national security and corrupted government can
all lead to decreasing confidence in the ruling party. These problems exist in
every country. As Internet growing to become the major source of information,
countries with little or no restriction over Internet have to deal with all the
reports on social problems, while countries with monopolies over information
simply eliminate any challenging or disapproving opinions. This of course does
not mean that countries like United States have more social problems than
countries like Russia, but people do tend to have greater support of their
governments when exposed to less unfavorable news. The theory may also be
supported by the negative relationship between democratic level and confidence
in government.
Does the finding mean that governments with a need to
increase people’s support should start monitoring the Internet? Maybe, if they
are desperate. But controlling information flow is never the solution to the
dissatisfaction and distrust of the government. Governments should accept the
fact that information is spreading faster than ever with the help of Internet.
Instead of trying to control it, they should have faster and more effective
response to possible accusation spreading online and also learn to use the Internet
to their own advantage in publicity.
The research question of this blog post is clear and concise. It is also quite an interesting premise. The author determines this is a “disappointing result” but it may not be. Just because the IV does not have an effect on the DV as expected does not mean nothing can be learned from this post. It can be argued that Internet freedom has no effect on Government confidence and thus can be an argument to dissuade governments from cracking down on internet access. If what they are doing has no effect, why do it? Plus, you could bring in the benefits of access to internet to back up this statement, for example, better innovation, increased economic activity (when you consider giant tech firms like Google having offices in a country or persons simply being able to connect with international buyers for their products). In terms of the World Poll Survey used to pull the confidence in government data, it should be clearer what year that is from. That way a reader can be sure that it is relevant to the internet freedom data of 2014. Also, the argument that Internet accessibility has an effect on government support is not altogether convincing. Perhaps there is a way to convert this to a dummy variable showing high access and low access to internet and then doing an interaction with the IV? To see if the level of accessibility may affect the results (Similarly to the problem set where we did high and low development using the median HDI). There is also no comment on reverse causality, this should be addressed.
ReplyDeleteBig Picture
ReplyDelete1. Clear research?
Yes, the intro and the tables explain very well what you are trying to study and why. The scatter plot doesn’t show on the blog. You also did a good job in explaining your sample and your variables. Great job in explaining that although your sample is small, the quality of it and of the variables is good and justifies the analysis.
IV: Internet Freedom
DV: Confidence in Government
2. Who cares? You did a good job in explaining the importance of the problem and why it is relevant in today’s world to study this problem. Maybe I would have included some data to situate us in the context.
3. Answered the question? Yes, you did answer the question. Internet freedom is not a determinant of government confidence.
Nuts and Bolts
4. Bivariate graph: The bivariate graph doesn’t show in the post
5. yes, both summary and regression table are in the post.
6. Both tables look good but the graph doesn’t show.
Modelling and inference
7. Evidence of transformation as needed? Yes, the author tried to look for linearity and found it as explained by the text.
8. Cofounds? The model control for cofounds really well. I would probably add education or political participation to the model.
9. Reverse causality? You controlled for it by adding well-chosen control variables.
10. Correct interpretation free of jargon? The interpretation is correct but I would have also said more about the other control variables that were not significant because they are findings as well. You use simple language so the general audience wouldn’t have had problems reading your post.
Overall comments:
The theme is really interesting and you did a great job in selecting your data base and your variables. I think you could further explain the results by incorporating the control variables although they were not significant. Your tables looked good and the only problem was that I couldn’t see the graph. Great job!
It's well organized with appropriate control variables. I;m thinking about the indicator for economic development, GDP per capita. I wonder if other indicators for development would have different effect.
ReplyDeleteThe study is relevant, and the explanations for why are good. I think it's valuable even though the results aren't what you predicted.Sample size and variable adjustments may yield something different, possibly. At least your analysis with these particular variables and sample meet the objective and answer the question.
The graph is hidden but your tables show the right information. This is one of those really interesting projects that can be done using so many independent and control variables, especially as internet usage stats change over time. Your explanation makes this very clear and easy to follow.