What does Twitter have to say about Trump’s Executive Order on Immigration?
In just a few weeks, President Trump has changed the tone of the national conversation. His executive orders have sparked debate on both sides of the aisle. The most discussed of these is his 90 day ban on immigration (and travel) from seven Muslim majority countries. Critics have accused him of discriminating against an entire religion while supporters have pointed out that his ban is temporary and that such a step is needed for national security.
Both sides have tried to be as loud as possible. Social media sites have been flooded with messages of support and calls to protest. The most important site for this discussion appears to be Twitter, where calls have been made to delete apps, donate money and pick a side.
For this article, we used Twitter hashtags to examine the support (or the lack of it) for the ban. We manually identified four hashtags, two for and two against. The two hashtags used to support the ban were ‘#faketears’ and ‘#boycottstarbucks’, while, the two used to oppose the ban were ‘#deleteuber’ and ‘#resist’. As the search was restricted to a certain time period, it is unlikely that the hashtags were being used for any other purpose during those days. The recent mentions of these tags were collected through Twitter’s API. Then, we used Google Maps to identify the US state where the tweets (or re-tweets) came from. Further, we compared this data to the election results to identify the possibility of a shift in support, that is, states that supported Trump but may not support the recent executive order.
In the interest of clarity, we would like to acknowledge that this paper is limited by both its low sample size (approximately 3000 Tweets) and by the preponderance of young users on Twitter. As a result, the output of this paper is not intended to be a definitive account of support (or the lack of it) for the ban. Instead, it is supposed to be a look at patterns of activity on Twitter and use them to better understand the atmosphere around the executive order. Any attempt to use this data to make grand claims would be unfounded. However, it should be noted that despite the low sample size, the distribution of the tweets aligns quite closely with the voting patterns observed in the 2016 election.
MAPPING THE BAN
The first thing to note is the surprising (or perhaps not so surprising) similarity of the two maps, namely, the map of election results and the map of support for the ban. For the second figure, we added the two hashtags on either side and compared them in each state. If the raw number of hashtags in favour of the ban were greater than those opposed to the ban, then we coloured the state red. If however, the hashtags opposed to the ban were greater, the state was coloured blue. In both cases, only the winning percentage is displayed on the map. Yellow states are those where the numbers were even.
States that were close in the election, like Michigan, which voted 47.6%–47.3% in favour of Trump, are close in our map. Michigan leans in favour of the ban by less than 1%. Similarly, Pennsylvania, which voted 48.8%-47.6% in favour of Trump, is opposed to the ban by just 3%.
In terms of larger states (by population), like Texas, which went 52%-43% in favour of Trump, our map shows similar numbers, with support for the ban at nearly 57%, while the rest of the tweets were opposed. Other large states also show a resemblance. California, which went 62%-33% in favour of Clinton shows great opposition to the ban with nearly 62% of the tweets being opposed to it.
With regard to smaller states like Hawaii and Alaska, the low number of tweets makes it difficult to be certain of a trend. Thus, we don’t analyse these in depth.
For most people, this may be simple enough — states that voted Trump seem to be mostly in favour of the ban while states that voted against him are opposed to it. However, for others, this may not have been expected. The results show that not only did people vote for Trump, but, they also voted for his policies (well, at least this policy).
Of course, our data is not an exact match. States like South Carolina, which voted heavily in favour of Trump, have a greater number of tweets opposed to the ban. Similarly, Connecticut, which Clinton won comfortably, is red on our map. There are several possible reasons for this –
- There was extraordinary activity on social media, which led to this discrepancy. After all, a sample does not represent the views of everyone and support (or the lack of it) may exist on the ground but, has not been captured by the tweets that we have collected.
- Between the elections and now, that is over the course of the last two months, people in these states have changed their mind.
- Voting takes a lot of ‘effort’. On the other hand, tweeting is comparatively easier. Also, in order to vote, you need to meet certain criteria. We cannot comment on the ‘eligibility criteria’ of Twitter, but, clearly, this may leads to differences between the two.
- These states may have voted for someone, but, when it comes to the ban, they have different views. This will be further analysed below.
WHERE DO ALL THE REFUGEES GO?
A number of people have talked about the ban as a way to keep refugees out of the country. Due to this, we decided to see if support for the ban was higher is areas where there were a large number of refugees. However, this comparison has mixed results. There are some states that follow this trend like North Dakota, Arizona, Texas and Georgia, among others, however, states like Washington States, Vermont and Maine, among others, accept a high number of refugees and yet, show opposition to the ban.
The same applies to states that have a low number of refugees. In those states, there may or may not be support for the ban. Thus, the ban may represent something more than just the people’s views on refugees.
SOME STATES REFUSE TO CONFORM
While it is generally true that the voting patterns of the 2016 presidential election are similar to the support/opposition to the ban, some states are acting all on their own. For the figures below, we have taken the tweets opposing the ban as a percentage of the total tweets in that state and compared that to Clinton’s vote count in the presidential election.
Iowa is one of the most puzzling states as Clinton lost there by a substantial margin and yet, the opposition to the ban seems to be extremely high. There have been protests in the capital and some politicians have expressed their anger toward the ban.
Wisconsin is another similar case, although Clinton’s margin of defeat was smaller. In these cases, local factors may be the reason for the difference, thus showing that these voters may prefer Trump, but, don’t like this particular policy.
On the other side, there are states that Trump lost that show high levels of support for the ban. New Jersey is next to New York, which shows high levels of opposition and has had demonstrations at several airports. Further, Jersey voted for Clinton. However, it shows a great deal of support for the ban. We could find no particular reason for this support based on what we read.
Similarly, Nevada shows the same high levels of support. Based on some articles, there was talk of opposition to the ban in Las Vegas. However, these tweets may be coming from other parts of the state. Since we do not have data on regions within the state, we can’t say whether Las Vegas is different from the rest of the state.
NOW WHAT?
This data captures an instant reaction to the executive order and shows that in many parts of the country, there may be high levels of support for it. Further, this support tends to align with those states that previously voted for Trump. In the 8 states where Trump won more than 60% of the vote, 7 of them show Twitter support for the ban. Similarly, of the 5 states that Clinton won more than 56% of the vote, 4 of them oppose the ban. Thus, the data shows a split along party lines.
Using Twitter to map out support for such action is quite unusual. There are bound to be problems with the data due to the way that the medium is used. This can be seen in the case of states like Iowa and Nevada that show large differences between their voting patterns and their tweets regarding the ban. Nonetheless, this could be due to the other factors mentioned previously. And, it would be unlikely for every state that voted for Trump to show support for the ban. After all, people vote on the basis of a number of different factors.
There has been a shift in the way that social media is perceived. It is now the playground of political parties, companies, public figures and, of course, ordinary civilians. Given its power over our democracy, the analysis of trends on these new portals is key to understanding the world that we live in.
