Skip to main content

2018 Research

Brandon Grisham

The Reflector

Mississippi State University


What Reflector readers want from their paper


            There is nothing more important for a community than a newspaper. As an organization of writers, our purpose is to connect with others. We work alongside members of our society and keep everyone informed. Some might undervalue us or fail to pay attention, but those that read our content remain up to date on human affairs. That resource is needed at every scale, whether it be a small town or an entire nation. As the world moves closer to technology and away from print media, journalists must adapt to online mediums. Our prime imperative is to maintain the standards, integrity and values of news collection as we continue to venture into this digital age.

I work for the Reflector, Mississippi State’s college newspaper. It has served as a student-led news source since 1884. It also acts as the oldest running college newspaper in the SEC, leaving a ton of responsibility. The Reflector is available to the entire campus and the encompassing town of Starkville, Mississippi. Our staff is involved in every manner of affairs within our county, offering reports and analysis of local phenomenon. My job has served as a staff writer, but I was recently promoted to the Online Editor position. With this change came a new outlook on the status of journalism as a whole.

My position is specialized and I have enormous pride in the work that I do. I schedule every article to be posted on our social media pages at convenient times for our readers. The website allows me to archive copies of every digital issue produced by the paper since 2001. My edited videos are broadcasted to the entire state every week using the university’s television center. Needless to say, if anyone needs our news and cannot get it in print, they go through me. That is an enormous level of power that I try to wield correctly, ethically and responsibly. After holding the position for a few months, I can see the impact that I am making.

The task of maintaining the website, operating social media and editing videos requires knowledge that most writers do not possess. Phrases like ‘SEO keywords’, ‘data analytics’ or ‘RSS feeds’ tend to go over most people’s heads. Technical skill only comes from endless hours of practice, tinkering and problem solving with computers. These informative hours are great for technologically-oriented students, but not for writers that have deadlines. Explaining a content management system to our staff is the most that we can do. The alternative would be to place HTML, CSS and PHP coding requirements on all editorial positions.

My research began after automating my most tedious tasks. It all starts with how I began the position. My predecessor left abruptly, leaving me to fill the opening two weeks in advance. That also meant roughly a day of training before being thrown into the deep end of content production. My training included a synopsis of HTML coding, publishing online assets and exporting digital issues. It all wound up being a handful of tedious technical tasks. The expectation of doing this twice a week for our Tuesday and Friday issues was overwhelming. After two weeks of stress, I began to ease into the rhythm of production with organized tricks.

I found that if I upload the digital issue with the same date format, I only need to edit 4 characters of the HTML code. Scheduling tweets and Facebook posts require time, but building them in advance using a manager cuts down the hassle. My favorite automation was changing the online article organization, leaving the website looking dynamic and constantly updating. With these foundations set, I had ample time to conduct my research and find out which content generated the most traffic.

Using Google Analytics, I was able to parse the Reflector website data and sort everything into categories. These categories could be filtered down to base traffic, visitor time spent, spiked search results and overall content popularity. This data is crucial to adaption, as it serves as readership statistics. It functions in the same way as a reader poll, but with a hundred percent response rate. Everyone that visits the website pings our servers, which gives us details about their browsing habits on our website. With this data we can discover which types of stories generate the highest response, determine what stories have little interest and locate the areas that are not covered by the empty search results.

Data provides a ton of power, especially once the readership is in the thousands. If utilized properly, website data can act like the basis of a scientific study. I have collected the entirety of the traffic from 2018 and sorted it into month-long segments. These segments were then exported to an XML file and charted with Excel. Utilizing these tools, I know exactly which stories are the most and least popular. These Excel databases proved to be exceptional way to keep track of when stories were covered or if similar stories in the same vein would be worth the effort.

As students, we have limited time to cover material, so any way to reserve writers for more relevant content is an enormous benefit. An excellent example was when a student created a Kickstarter page for his own videogame. Once we saw the response in the traffic, we scheduled a follow up for the next semester when the game entered it’s beta development stage. This type of reaction would be hard to determine from a poll, but with viewer data we could find niche audiences for specific story categories. The basis for my research was to find what Reflector subscribers want to read, and I believe that I’ve found the answer.

For the school year, we generated 158,341 pageviews in the spring and fall semester. The summer months of May, June and July were not very helpful in terms of viewership. Due to the lack of fresh content and an empty campus, we only received 35,135 views. Our traffic mainly stemmed from old opinion pieces that remained relevant search topics for the Google search engine. The regularly updating homepage and section pages generated 21,370 total views, so our stories generated 172,106 views overall. That says a lot about how popular our content is when the student count of 22,000 is taken into consideration. If every visitor was a student, which is impractical, they still would have read over seven articles over the year.

A surprising factor in our viewership statistics is that only 20 articles received over a thousand views. Given that we average over 430 articles over two semesters, it really speaks to how most stories only captivate niche audiences. We run two papers a week and publish 24 full papers a semester. Some might consider these smaller articles to be frivolous, but the certainly add up to impressive web traffic. They allow for a large collection of topics to be covered, and our variety is the main draw for the bulk of our traffic.

The most popular section of the year was our news section, which had 61,740 views. This section is the only segment of the paper that receives two pages, so there are more stories overall. Our most viewed story was the result of our mascot statue being vandalized by unruly fans of a defeated football team. It generated 10,310 views, serving as the most seen article of the year. Other significantly popular articles include a new parking garage, students involved in a burglary and a local citizen becoming a suspect in a murder case.

Our second most popular section was Opinion, which held 59,324 total views. Given that this section does not need to rely on local events, there is an innate creativity that comes with the topics of these articles. While they can be topical, most of the articles featured here are of incredible variety. Serving as evergreen stories, they draw in considerable search traffic year-round. The most popular article from 2018 was “Professional Gaming Should Be Considered a Sport”, which received over 7,627 views. The irony from that story is that less than two months later Mississippi State went to the first annual Esports Egg Bowl.

Following that was our Sports section with 21,268 views. The drop off between Sports and the other two sections was surprising, but there are solid reasons for the diminished viewership. As an SEC school, most students trust professional networks and search engines for faster coverage. Our Sports section covers columns and game recaps, which ultimately results in a niche audience. As the internet increased the speed and depth of sports coverage, articles decreased in value. We still support our Sports section, as it provides readers with important information regarding the category. We plan to restructure our sports coverage utilizing these digital viewership statistics to be more personal and cover a wider variety of sports. 

Our Life section was the least popular section with 14,591 views. This section has the issue of competing with three other local newspapers that focus on covering events and entertainment. It can be hard to avoid re-treading the same topics, so the lower viewership makes sense. Only a single story broke over 1,000 views and it was based on the headliner for an annual concert. Every other story hovered within a few hundred views, focusing on the niche audiences for local events and performances. Through 2019, we have utilized this data to improve our viewership and cover diverse events.

After breaking down the section data, various trends started to emerge. These trends were able to be charted through Excel and sorted into distinct categories. The primary category that proved to be useful was sorting by week. These weeks displayed on three-month graphs which weeks held the most responsive content. Through this analysis, we were able to pinpoint various spikes in popularity and cross-reference the summary data to the viewership data. Our most popular weeks were in September, which held an average of 9,119 views. The second most popular weeks were in November, holding a weekly average of 7,672 readers.

Following the data analytics that I was able to pull from the website, my research moved toward other online outlets. It primarily focused on response data from social media platforms. The Reflector utilizes Facebook, Instagram and Twitter to help with content pushing. These sources wind up generating over 30 percent of our overall traffic, so the data provided is rather influential. Instead of viewership from the click-through rate, I paid closer attention to comments, likes, and shares.

Facebook is the highest pull towards our website outside of search engines. By analyzing the data from the entire year of 2018, we were able to determine various specifics about our readers. Two-thirds of our 2,985 Facebook followers are in their late 20’s. This gives us an expectation for the content that they would respond heavily toward. The majority of the Facebook followers are female, yet the 42 percent of male users tend to be more responsive. Comment threads on this platform tend to have the most controversy, leading to skewed data, so I focused on comment data from a different source.

Instagram is our newest social media platform. Considering the posts to this platform are image-related, we upload preview photos for future articles. We have operated the account for three years and have finally passed the milestone of 1,000 followers in November of 2018. With a post count of 46, we received 2,956 total hearts and gained over 156 followers through the year. The demographics with this platform are eerily similar to Facebook, as 65 percent of the followers are female. They also fall within the age range of 18 to 35, supporting the hypothesis that the bulk of our followers are alumnus or Starkville residents.

Twitter is by far the most active and popular of social accounts. We ended 2018 with a follower count of 9,596. Being the substantially more popular medium also came with a few benefits when organizing data. All of our stories are scheduled to run corresponding tweets at specific time slots of each day. By doing this, we could cross-reference the website traffic to the engagement from Twitter. This utilization allowed for better popularity metrics and clearly displayed that posts with corresponding images held a better click-through rate.

By merging the data collected from social media with our own website statistics, I was able to run analysis tools using Excel. ANOVA testing, regression, and chi-square tests proved to be extraordinarily valuable for determining improvement. The alternative hypothesis that we have gained more readers was correct. A few questions turned out to be incorrect, as our null hypotheses proved to remain as the status quo. These questions were based on social media response or if we were significantly increasing web-traffic. These data-analysis tools were crucial to determining if these corelating trends were actually valid at a statistical level.   

After starting this project, I was not sure how to proceed with all of my data. It took over a month of trial and error to compile everything together into a cohesive format. Then it took weeks of testing to determine how significant the results were. My research solved several of my questions about section popularity and readership demographics. It also disproved my expectations about social media interactivity and our improvement in total web-traffic. While there were several frustrating aspects, I’m happy to have run this analysis. It has only proved to be useful to all future production cycles and I guarantee that these statistics will help us grow in the future.

This project gave me insight into how data analysis could help newspapers. It also showed me how crucial website traffic statistics are for understanding readership. Without this data, all of our information would be inference-based. That is a horrible way to determine what subscribers want to read, as it breaks the main rule of statistical analysis. Correlation does not equal causation, proving that just because it looks one way does not mean that it is accurate. Moving forward, I can help set the standard for truly understanding our readers compared to the age-old method of guessing.

There are dozens of potential future applications with this information. As the Online Editor, I plan on running this analysis every semester to gauge our progress as a paper. It will serve as more secure foundation for determining story assignments. That will only benefit our college campus, local community and the information our readers have access to. We collect news but we need to keep adapting toward digital information. This research has proved itself to be useful and it takes very little time to adjust after setting up the template. With these methods, I am certain that the Reflector will grow into a stronger paper. So far, the results are promising.