Findings
Initial Finding: Aggregated Terrorist Attacks & Advisory Levels suggest that generally countries with higher terrorist attacks have higher advisory levels.
Initial exploration of the aggregated number of terrorist attacks from 1999 to 2017 compared to the to the current U.S advisory levels per country shows an trend of higher terrorist attacks and higher advisory levels in Middle East & Africa region. In the map below, each country is color coded according to their advisory levels; the red countries are advised as level 4: "do not travel" whereas the blue countries are considered relatively safer than the rest. Overlaid on top is the aggregated total incidents from 1997 to 2017 per country, which are shown as the magenta circles, larger circles mean greater number of terrorist attacks in the country. Generally, countries with higher terrorist attacks have higher advisory levels (>= level 2). Countries in regions such as Middle East & Africa have higher rankings and total terrorist attacks recorded. From this generalization, let observe the hypothesis that terrorist attacks deter people from traveling, and thereby reducing the number of flights in for the respective regions.
However, the advisory levels are derived from others factors other than terrorism (see Definition). For example, Venezuela is a level 4 country but it has a relatively low number of terrorist attacks; this is because the country is political unstable due to its civil unrest and not influenced strongly by attacks organized by a none-state actor. On the other hand, for Somalia, Pakistan and Afghanistan have both political unrest and terrorist attacks corresponded here with higher advisory levels and bigger circles. Also, Columbia experienced the highest number terrorist attacks in South America, the U.K the highest in Western Europe, Russia the highest in Eastern Europe and Iraq highest in Middle East & Africa. It should also noted that this map has a limitation of time, its data on country advisory levels does not update in real time as the levels are updated frequently by the U.S. Depart of State. The levels here are as of April 2019 after the Easter Day Sri Lanka terrorist attack that changed its prior level 2: "exercise increased caution" to level 3: "reconsider travel".
Finding 2: An analysis of side-by-side total Terrorist Attacks & Flight Percentage Change Per Region show mixed results.
Middle East & Africa: The number of terrorist attacks steadily rose from 1999 and peaked at 2014 while the flight percentage change took 2 big dips in 2001 and 2006. The 2001 was the year of 9/11, the U.S government attributed the causes the Middle East, the start of the War on Terror and its associated national security policies maybe the most probable explanation to this drop in flights originated from the U.S. However, the data shows that 2006 drop is likely not connected to terrorism. The same is true for its drastic increases in 2004 and 2008.
Europe: The number of terrorist attacks steadily rose from 1999 and peaked at 2014-2015, a closer look at Europe subsets show that Eastern Europe's attacks make up the majority of attacks in both years (80% in 2014 and 67% in 2015). The fact the the flight data includes both Eastern and Western Europe as Europe could skew the outcome of this line chart. Flights could have decreased to Eastern Europe and increased in Western Europe for the two years, Western Europe flight increase could have a potential offset and overshadow Eastern Europe under one category Europe. On the other hand, one can see that there is a positive response from the increase in terrorist attacks since 2013, the percent change of flights continuous rose through out the years of heavy terrorist activities. However, this may be the increase of flights to Western European countries rather than Eastern European countries. There is no travel statistics for the these two subsets of Europe to verify this suspicion.
South America: The number of terrorist attacks flatten from 2004 to 2007 and started rising in 2008, this pattern is seen replicated in the flight percentage change that hovered between the range of positive 7% to positive 10%. The increase in number of terrorist attacks from 2008 corresponded with the decline in flight traffic. The 2000 dip in negative percentage change can be seen as a result of the 2001 peak in number of terrorist attacks. The 2010 dip on the line chart is explained under Other Consideration as having a separate and stronger causation. Given that South America relies heavily on tourism industry, the economic crises of 2008-2010 started in the U.S can contribute significantly to the decrease in flights to South America.
Asia: It is much trickier to analyze Asia's terrorist attacks and percentage change in flights because the while the U.S travel statistics dataset only gives flight data under Asia as one region, the terrorism dataset has categorized Asia into East, South, Southeast and Central Asia, these subcategories of Asia helped in identifying countries that have intensified terrorist attacks from countries with fewer attacks. Countries in South Asia usually make up the majority of the attacks, Southeast Asia's attacks have the second highest number of attacks. For 2014, similar to the increase of terrorist attacks in Middle East & Africa, South Asia (see Data and Methodology for list of countries) made up 80% of the terrorist attacks. The -11% change in 2003 could be attributed to the start of the War on Terror, where Pakistan is part of the South Asia and have known to be involved in housing terrorist groups. In addition to the global trend of negative flight decrease in 2010, there is another similar trend in peaked number of terrorist attacks in 2014 along with Eastern Europe and peaked positive flight increase in 2004 across Middle East & Africa, Europe and South America.
Central America & Caribbean: This region experienced a gentler decrease in flights over a range of 7 years (2003 - 2010). It is much harder to attribute the flight changes to terrorist attacks for this particular region because there are so few of them (ranging from 4 to 18 attacks per year). There is no attack in 2008 and 1 attack from 2010 to 2012 while the flights percent change rose steeply to 18% in 2012. The peak of terrorist attacks is in 2013 which corresponded almost directly to the drop in flights in the same year, followed by a decrease in attacks and increase in flights in 2014. However, this pattern does not seem to be entirely consistent over the observed period of time.
In Summary: All regions of the world experienced the decline in flight volume from the U.S in 2001 and 2010. This commonality could be explained by 2 major U.S events: the 9/11 terrorist attack and the economic recession. Therefore, 2001 decrease in flights confirms our hypothesis that fear in terrorist attack deter people from travel abroad. However, the the data show that after 2001 - 2002, variances in flights are not entirely associated to terrorist attacks per region.
Additionally, the 2010 decrease in flights tells us a story on world economy, which directly affected the tourism industry and number of people traveling abroad. When considering how terrorist attacks affected people traveling abroad from the U.S, 2010 is disregarded as a variable to the research question.
Finding 3: Regression shows an anomaly and insufficient proof for the association between flights and terrorist attacks per region.
For this linear regression:
X is the independent variable assigned to number of terrorist attacks
Y is the dependent or response variable assigned to number of flights
R is the correlation coefficient which tells us the degree of the relationship between flights and terrorist attacks
R Square is the coefficient of determination that measures of well the regression line represents the data.
b is the slope for the line of best fit
This linear regression model shows the line best fit for the two variable number of flights and terrorist attacks based on world regions: Middle East and Africa, Europe, South America, Asia, Central America and Caribbeans.
Middle East & Africa: There is a positive relationship between flights and terrorist attacks. The coefficient of determination, R Square, shows that 75% of total variance in flights can be explained by the terrorist attacks. Each unit increase in terrorist attacks is associated with an 244 increase in number of flights. This finding appears to be the anomaly, it's counter intuitive and rejects my hypothesis that people travel less to the region where there are more terrorist attacks.
Europe: There is a very small positive relationship between flights and terrorist attacks. Furthermore, the R Square shows that 0% of the total variance in flights can be explained by terrorist attacks. This indicates a bad fit, that there are factors other than terrorist attacks that might be more useful in understanding the flight data.
South America: There is a negative relationship between flights and terrorist attacks. However, the R Square shows that only 14% of total variance in flights can be explained by terrorist attacks and a unit increase in attack is associated with -1772 or 1772 decrease in flights. Overall, the finding appears to be insignificant in size.
Asia: There is a positive relationship between flights and terrorist attacks. Furthermore, the R Square shows that only 13% of total variance in flights can be explained by terrorist attacks and a unit increase in attack is associated with 121 increase in flights. Again, this tells us that there is little significance between terrorist attacks and oversea flights.
Central America & Caribbean: There is a negative relationship between flights and terrorist attacks. However, the R Square shows that only 9% of total variance in flights can be explained by terrorist attacks and each unit increase in terrorist attack is associated with 134100 decrease in flights. This is another insignificance confirmation of the relationship being studied.
Finding 4: Terrorist Attacks peaked globally in 2014 & A better view on regions affected each year.
The visualization below shows where the terrorist attacks took place from 1999 to 2917. Each region is color coded to its related regions and countries: (see Data and Methodology for more comprehensive list of countries)
Blue circles: Sub-Saharan Africa countries
Purple circles: Middle East & North Africa countries
Green circles: South Asia countries
Orange circles: Eastern Europe countries
Yellow circles in Europe: Western Europe countries
Yellow circles in Asia: Southeast countries
Pink circles in Asia: East Asia countries
Yellow circles in South America: South America countries
Brownish circles in North America: North America countries
The legend box to the right show the 'Progress' of the animation, the 'Duration' of the animation can be manipulated to a slower or faster rate through increasing or decreasing the number of seconds. The 'Current' shows the current year of the global distribution of terrorist attacks on the screen (please ignore the decimals after the year as part of the 'Progress' bar update). Viewer is encourage to click 'Pause' and 'Play' to explore where the attacks are located in a the interested year.
Pausing at 2014 on the map visualization of attacks and comparing the peak of attacks on the line chart are telling a story of maximum height of terrorist activities globally in 2014 from 1999 to 2017. More terrorist attacks are shown on Western Europe, and as far North as Sweden and Finland. Overlapping data points countries such as India, Afghanistan, Yemen, Pakistan, Somalia, Columbia and the Philippines may have some global connections to each other. Although our line charts show normal to increase flights in 2014 for all regions of the world, we do not know for sure if these attacks are not publicized on mass media, and did not affect people's decisions for going abroad to these regions or that because more people are abroad that they are targeted in terrorist attacks for any political goal or other reasons.
Finding 5: An Elevated level of warning from the U.S government.
The U.S Department of State have released the Travel Warning which includes destination descriptions and safety information on 211 countries in the world. Looking at the keywords, displayed as bigger words in the wordcloud, shows the type of message that can potentially instill fear and anxiety in others. The frequent use of words such as "border", "crime", "demonstration", "terrorism", "violent", "threat", etc. not only expresses the U.S's proactive warning of the safety and stability for each country but also shapes the perceptions of those countries as well as discourages travel for most people, especially those who have never been outside of the U.S.
The U.S creates a free Smart Traveler Enrollment Program (STEP) that encourages its citizens to enroll before going abroad. This program directly communicates and advises U.S citizens about the safety conditions in their upcoming destination. The goal of this elevated level of warning is unknown. It could be seen as added precautionary practice in the interest of Americans well-being. The outcome of this elevated warning can also be a heightened sense of fear from traveling to a particularly country, region, and in effect limits the mobility abroad.
Finding 6: A sample of twitter posts shows a awareness and sentiment toward travel and terrorism in economically advanced countries & countries effected by terrorism.
This analysis is done from a sample of 136 tweets from June 2009 to April 2019. Due to limited project time, this project does not use the Standard Twitter API because it only yields 7 most recent days of tweets, which cannot contribute to this project research question about terrorism and travel from 1999 to 2017.
By searching for tweets with hashtags #travel and #terrorism, the result page shows 136 posts dated to the earliest post in June 2009.
Out of 136 tweets found with both hashtags #travel and #terrorism, about 40% are from travel agencies and marketing, 20% from the news and research organizations such as World Economic Forum, RAND Corporation, National Geographic, The Economist and other news agencies from countries such as India and Israel. The rest of the posts are made by citizens from the countries.
Not all users have their locations listed on their profiles and locations that are duplicated are not counted but weighted for node size. A bigger node means that more users who posted are from that specific location. Users from U.S and U.K made up the majority of the users. Other users from other countries are represented by small blue nodes because they only post 1 or 2 posts.
On the first node graph, it appears that more awareness of terrorist attacks is seen in developed countries and countries that are more involved in terrorist activities such as Israel and Jerusalem.
On the second pie chart, the positive, negative and neutral post are analyzed using Vader sentiment analysis's compound score.
According to Vader developers, the compound score is computed by summing the valence scores of each word in the lexicon and normalized to be between -1 and 1 being most negative and most positive. For sorting the 136 posts to either positive, negative or neutral: I determine that a post is neutral if the compound score is 0.0, positive if compound score > 0.0 and negative otherwise.
It appears that more than half of the total post are negative posts about travel and terrorism, about 40 posts are neutral and 27 posts are positive. Text analysis show that posts range from warning about a location of recent terrorist attacks to advertising for technologies and services for weary travelers to encourage travel to previously known as 'dangerous' places such as Pakistan. It is uncertain whether this small sample size really represents the global awareness or perception about terrorism and travel. Certainly, the concerns come mostly from citizens of economically advanced countries that have more capability to 'travel' overseas for vacation, work or other reasons that citizens of developing countries do not have.
Finding 7: Americans travel to Mexico and Canada more than Europe and Asia starting 2011.
While looks like U.S travel to regions such as Middle East and Africa increased by almost by 50% in 2008, Middle East and Africa flights make up only 3% of all U.S flights overseas in 2008. This figure hovers from 0% to 5% max through 1999 -2018. By looking at the total flights abroad and converted them to percentages per region for a total of 100%, Americans have been traveling to these top 4 destinations: Asia, Canada, Europe and Mexico. The sum of those destinations' percentages make up 60% of all travel in 1999 and 73% of all travel in 2018. This finding shows that Americans have not been traveling to destinations outside of this list of regions, which makes it harder to be confident about answering the research question using U.S travel statistics.
There is a drop in total percentage of flights to Europe and Asia starting in 2011 and a rise for flights to Asia and Canada. There are multiple factors that may have caused this noticeable drop, it could be attributed to the 2008-2010 recession. The common feature that both Mexico and Canada share is the their relational distances to the U.S as both are more accessible being closer to U.S than all other countries. There is a slight connection with the concept of 'staycation' for British citizens who express that "increasing numbers of us taking our holidays at home in Britain of flying to far-flung hot destinations."
A 2017 Gallup poll survey shows that about 46% of Americans are less willing to travel overseas due to terrorism concerns, the number went down from 48% in 2001. Unfortunately, this may show that more Americans are accustomed to the terrorist activities that rose steadily post 9/11.
If more people are accustomed to the terrorist attacks, then this would answer why there is no statistically negative correlation between post 9/11 flights and post 9/11 terrorist attacks.
It appears that we can only be more certain about the fact that the U.S flights data recorded the most observable decrease in flights for all regions of the world in 2002 and 2010, both are attributable to post 9/11 and the 2008 economic recession.