In JOUR 309, students were tasked with compiling, sorting, filtering and analyzing data for a data story. For this project, I used data from the Douglas County Sheriff’s Office Twitter account. I compiled the data and engagement rate to find helpful insights for the company to further their twitter goals. I also created to infographics in Adobe Illustrator to illustrate the data.
View the presentation here.
Douglas County Sheriff’s Office Tweet Analytics
Below is an analysis of the project:
For the final project in Journalism 309, I worked with a dataset from George Diepenbrock, public information officer at the Douglas County Sheriff’s Office. Diepenbrock sent me a list of 155 tweets spanning from July-October 2021. My task was to take this dataset and find useful insights for the Sheriff’s Office to use in the future.
Origin
- This data comes from George Diepenbrock, the public information officer at the Douglas County Sheriff’s Office. Diepenbrock created the spreadsheet using the Douglas County Sheriff’s Office twitter account. Diepenbrock pulled impressions, engagements, engagement rate, retweets, replies, likes, user profile clicks, url clicks, hashtag clicks, detail expansions, follows, media views and media engagements.
- This dataset is credible because I too could easily look up this data on my own– it’s all on Twitter. Really, the only thing Diepenbrock did was compile the data into a spreadsheet. It’s also for his organization– why would he lie about data he also needs?
Spreadsheet
- (Here is a link to my dataset. I couldn’t get it to link as an excel sheet, so I had to convert it to a google sheet.) The dataset originally had more columns than listed currently. However, I deleted any columns that only had the value of “0” or matching values. The current columns are: tweet permalink, tweet text, time, impressions, engagements, engagement rate, retweets, replies, likes, user profile clicks, url clicks, hashtag clicks, detail expansions, follows, media views, media engagements and category.
- The dataset has, in total, 155 rows (excluding the label/title row). Each row contains a different tweet and its identifying data.
Transform
- I transformed this data by deleting any columns that had the variable “0” or deleting any columns that only had matching variables. For example, if a dataset only had 0’s and 1’s. These columns also tended to be the least important factors while sorting and filtering, so I deleted them. I also added a column categorizing the tweets, so that I could create separate pivot tables for them.
Analyze
- The main analysis I found was that tweets about the Sheriff’s Office had the highest engagement rate. These tweets had the highest engagement rate, likes, replies, engagements and impressions, all by a fairly large margin. I found that the section with the lowest engagement rate was animals: tweets about lost pets, loose animals, etc.
- A surprising analysis that sprung from this data was that despite dominating every other category, Sheriff’s Office tweets did not have the most retweets: #DriveSafe did. #DriveSafe encompasses tweets that have the hashtag “drive safe” or any other tweets about driving/vehicles.
Insights
- The biggest two insights were that tweets about the Sheriff’s Office have the highest engagement rate and that #DriveSafe had the greatest number of retweets. The final important insight is that tweets about animals do particularly poorly with the DGSO’s audience.
- These insights are important because they show us what is doing well, what could be doing better/expelled from the tweet deck and they show us a category that has the potential to be better.
Caroline G Zimmerman 