Amidst the Pandemic, Citizens Create a Boost in Classifications
The following piece was written by OAS Communications Coordinator Ryan Bower for the Snapshot Wisconsin newsletter. To subscribe to the newsletter, visit this link.
We would like to thank everyone who has helped classify Snapshot photos during the pandemic. We are happy to see so many of our volunteers connecting with nature while at home.
We want to share how the pandemic has affected the Snapshot Wisconsin team and offer a look at the surge in classifications. Jennifer Stenglein, Research Scientist at the Wisconsin DNR, shares some encouraging findings about the rise in daily classifications and Zooniverse classifiers during the pandemic.
The Snapshot Team During the Pandemic
“I think Snapshot has been a success story,” said Stenglein. “The effects of COVID-19 haven’t been very detrimental to the team. Change happened, and we adjusted. Some of our workload shifted, but we continued to be Snapshot – just in a slightly different way.”
The Snapshot team is very collaborative. Each week, team members must heavily coordinate with each other to keep photos moving to the next steps and interacting with a volunteer base of over 2,000 people. However, since Wisconsin’s Safer at Home order went into effect in late March, the Snapshot team has been teleworking from home. Stenglein said, “To me, that’s been the biggest change: not having face-time with the team. But the team has transitioned really well to having online meetings.”
There is now a bigger focus on getting photos to Zooniverse for our volunteers to classify. With more people at home, Snapshot photos are being classified at a faster rate than before. This increase in daily classifications is what first alerted the team to the boost.
Investigating the Surge
Stenglein has been a member of the Snapshot team since its beginning in 2013. She leads the scientific program at Snapshot and plays a vital role in turning quantitative data into usable metrics (for example, calculating county fawn-to-doe ratios). Stenglein was the first to offer to investigate the boost in detail, as she was excited to learn more about our volunteers and their behavior.
The first step to investigating this boost was to determine what time frames should be compared. Two time frames were needed: One from before the pandemic and one during the pandemic. “Our Zooniverse volunteers come from all over the world, and the pandemic has affected different places differently. But a lot of our Zooniverse traffic comes from Wisconsin, so we thought selecting a timeframe that was relevant to Wisconsin was a good approach.” Additionally, because Snapshot Wisconsin is a Wisconsin DNR program, a Wisconsin-centric time frame seemed natural.
To narrow down the time frames even more, the team needed to choose a specific date as the delineator between the two time frames. “March 15th correlates to about the time when the Department of Health Services started collecting their stats. It was less than two weeks later that the Safer at Home order went into effect. We felt like this date was a good mix of a time frame important to Wisconsin and a line in the sand. We needed that [break] point between pre-pandemic and mid-pandemic.” At least two months of data was needed to ensure reliability, so the two months prior (pre-pandemic: Jan. 15–March 14, 2020) and the two months after (mid-pandemic: March 15-May 15, 2020) were chosen.
With the time frames set, Stenglein began compiling the data for the analysis. However, before we get to Stenglein’s analysis, it helps to understand the difference between a photo, a trigger and a classification. Those who have classified Snapshot photos before will recall that you were given three photos at a time to classify as a set. Volunteers are asked to classify the set as one whole, meaning they would tag the set as having a red fox if a red fox was present in at least one of the three photos. Each set of photos is called a trigger and gets classified by multiple volunteers before it is retired on Zooniverse to ensure accuracy. Each time a trigger is classified by a volunteer, that is called a classification. Simply said, three photos make a trigger, and a classification is each time a volunteer looks at that trigger before the trigger retires.
The Boost: Analyzed
In total, our volunteers made 460,604 classifications during the pandemic – a huge increase from the 255,208 classifications prior to the pandemic. While each classification doesn’t perfectly translate to the number of triggers retired, this boost is a huge increase in Snapshot’s turnover rate for triggers. Additionally, a total of 1,168 unique classifiers logged in during the pandemic, almost double the number from pre-pandemic.
Classifiers were also, on average, classifying 11.6% more triggers each time they logged on. Once we saw that people were classifying for longer, we wondered what other behavioral shifts have occurred since the pandemic started. One big shift is seen in our established classifiers, or volunteers who were already active on Zooniverse before the pandemic. Stenglein saw that established classifiers were logging in and classifying on significantly more days than before, and they were classifying more triggers each time. However, Stenglein noticed that there were fewer sessions per day. A session is similar to a sitting in that one volunteer could classify triggers over multiple sittings in a single day. Altogether, our established classifiers were classifying more triggers, on more days, but fewer times per day than pre-pandemic.
The most classifications in a single day occured on March 29, less than a week after the Safer at Home order went into effect in Wisconsin. A total of 17,155 triggers were classified on that single day.
On intuition, Stenglein had an idea to check if there was a difference among the days of the week. Do volunteers classify differently on the weekends than on weekdays? “The weekend/weekday idea was something that I thought of because it was coming up on the weekend. I was thinking, ‘Man, these days all feel the same.’ I got intrigued by this question and wanted to know,” said Stenglein. “But I didn’t even know what the baseline was.”
Stenglein continued, “I was very interested to find out that [normally, pre-pandemic] weekends have much fewer classifications compared to weekdays. I could have hypothesized the opposite and convinced myself.” In fact, the total number of triggers classified rose for both weekends and weekdays, although unequally. We saw an increase of 34% during weekdays (1,507 more triggers classified each weekday) and 86.7% during weekends (2,452 more triggers classified each day of the weekend).
“I was really fascinated that both weekdays and weekend classifications went up [during the pandemic], but when you compare them now, there no longer is a distinction between them. I think that was my favorite finding,” said Stenglein. “With fewer options of things to do on the weekend, perhaps [volunteers] are willing to sit down and spend more time on Zooniverse.”
Stenglein also compared the mid-pandemic period (March 15–May 15, 2020) to a similar time frame from the year prior (March 15–May 15, 2019). Stenglein said, “It helped me to really believe in this boost idea when I looked at the same time period from the previous year and found almost identical results. There is really something unique about this [mid-pandemic] time period.”
Stenglein said, “There is value in communicating this data back to volunteers who are working extra hard right now by classifying photos. You are increasing our capacity to turn over photos and helping us get through our backlog.” In fact, classifiers helped process over 10% of Snapshot’s backlog since the pandemic started.
The Snapshot staff were excited to see this rise in classifications because a quicker turnaround time for photos means more up-to-date data for wildlife management decisions. Continuing to reduce the backlog of photos is an important way that the public (and Snapshot volunteers specifically) can contribute to the project and wildlife monitoring.
“Back in January, we had a meeting about how to deal with unclassified photos. We came up with all these great new ideas, but a pandemic was not among them. The pandemic is never good news, but it provided an opportunity to get through so many unclassified photos in a way that wouldn’t have been possible,” said Stenglein.
But in order to provide key measures to decision makers about animal distribution and abundance, Snapshot needs continued support from the public to keep classifying photos and monitoring wildlife in Wisconsin. Stenglein said, “Snapshot is a volunteer program. There is no Snapshot without its volunteers, and they are what makes Snapshot successful. Citizen science [projects] are a wonderful thing to participate in, and it’s a way to work with others towards research, science and wildlife decision support. People care about helping monitor wildlife and helping the DNR understand the wildlife in their own backyards.”
Stenglein and the Snapshot team want to thank the thousands of volunteers who have helped classify photos during the pandemic. “We are super grateful for the increased traffic and time that people have been putting into Zooniverse.”
Stenglein added, “On a personal note, I’ve been classifying photos with my son, which has been really fun. I’m glad that Snapshot provides an opportunity to see wildlife [while we are stuck indoors].” The Snapshot team encourages classifying photos together with loved ones. Check out the next article in this newsletter to see more ways to stay connected with nature!
Let’s discover our wildlife together!
Snapshot Wisconsin is a partnership to monitor wildlife year-round, using a statewide network of trail cameras. The project provides data needed for wildlife management decision support at the Wisconsin Department of Natural Resources. It also provides a unique opportunity for individuals, families, and students to get involved in monitoring the state’s valuable natural resources!
Learn more about the Snapshot Wisconsin project.