	Over the past couple months I have been working with professor Stromer-Galley and a team of 6 other individuals on identifying conspiracy theories in social media. Specifically, we focused on the 2016 Presidential election utilizing several Facebook posts made by both Hillary Clinton and Donald Trump. Before looking at the actual data our team created a working definition for conspiracy, in which we came up with “A conspiracy theory is a belief about an occurrence, or even that pertains to information not meant to be available to the public but that said individual intends to be shared or circulated.” After defining conspiracy, the team was able to move into the actual data set and do further analysis. 
	One area of this project that I chose to look into deeper was the average length of a conspiracy message compared to non-conspiracy. The expectations were that conspiracy would be longer due to the fact that a conspiracy needs some form of context, which turned out to be true. The process for calculating the length first involved downloading each of our 6 members completed 600 messages file from Google docs. Once downloaded, I averaged out the length of all messages, conspiracy and non-conspiracy. Next, I did an averageif formula that would average the length in all rows where conspiracy was marked ‘Yes’. I used the same formula to calculate the non-conspiracy but simply searched for ‘No’ rather than ‘Yes’. Lastly, once all of this data was calculated for each of our team members I averaged out the results to test against the initial hypothesis. The average length of a conspiracy message came out to be 478 characters while a non-conspiracy came out to only 129 characters. With conspiracy coming in at nearly 4 times as long as non-conspiracy based on our analysis it is safe to assume there is a strong correlation between the length of a message and whether it will contain conspiracy. 
	Another area that I further analyzed in this project was the occurrences of anti and pro data for each candidate. From what I remember from the 2016 election, at least my inner circle, it seemed to be strongly anti trump and pro Clinton. Naturally, this is what I thought would occur in the data set, but this was far from the case. Similarly to what I did for the length analysis, I started by downloading each member’s files that had completed the anti/pro analysis. I then did a countif statement for each of the columns where it was marked ‘Yes’. Once each file was calculated I averaged the results amongst the group. The actual results were quite surprising with anti-Clinton data occurring on average 151 times and anti-Trump only occurring 76 times. Pro-Clinton data, on the other hand, only occurred on average 36 times compared to 116 times for Pro-Trump. So not only is anti-Clinton data nearly twice as prevalent as anti-Trump but pro-Trump is 3 over times as prevalent as pro-Clinton. I think this data goes to show that people aren’t always aware of what is actually going on in a broad perspective when it comes to social media. People generally only see what is tailored to their social media agenda, hindering them from the general opinion and increasing confirmation bias. 
