| Booze + math = cool |
|
|
| Written by Nathan Wilke | |
|
Think you can guess how many beverages you need to buy for your house party this Saturday? Try predicting the drinking habits of an entire campus. Lori Voss, an undergraduate in mathematics and statistics, spent the summer working with a team of students from around the country on producing a mathematical model of student drinking trends. An accurate prediction method of what campus administrators could do to reduce the negative effects of student drinking could prove very useful for this country’s schools. In one year, almost 600,000 college students are injured in alcohol-related accidents. The amount of college students who binge drink (consuming more than five alcoholic drinks per sitting) is also alarming. Around 44% of college students claim to binge drinking at one point. At California State Polytechnic University in Pomona, Voss worked with a team of three other undergraduate students from universities around the country. The subject of their research was actually the subject that most students spend time on when their academic trials for the week are done and the weekend has arrived. o model other interest items in other organizations. This past Friday afternoon, Voss presented at a colloquium to the math department and other students, detailing the methods the team used to construct their models and the trends of the results they found from altering various environmental and input parameters. Sociological models have already been used to predict drinking habits and trends among student populations of campus. Until now, no mathematically powered model had been pioneered to quantitatively predict how many students would drink, as well as how often. The mathematical models work off of input factors such as individual tendencies, social influence norms common to a specific campus, and availability of alcohol. In her presentation, Voss outlined how the deterministic model was constructed. Students were categorized in groups similar to those used in student surveys seen on this campus in the past: from drinking zero or a few drinks a week to consuming five or more. Taking into account environmental factors, such as alcohol availability and social standards of drinking on a certain campus, the model goes through a set number of time step runs. In these steps, each student sample can move between categories based on their personal feelings and how they’re affected by the social aspects and other student samples around them. Some interesting results of the model are that it can be used to simulate wet or dry campuses. When reducing the value of the factor that represents how easy a student can get alcohol, the number of students in the higher-consumption categories (bingers and problem drinkers) decreases. Altering the social norms and influences to a increased drinking climate had a greater effect on how many students became binge or problem drinkers than increasing individual student risk factors. This means that students’ decisions are more likely to be influenced by their new social environment than the inherent traits they came to the campus with. A new type of model was created by the student team to better simulate trends, more accurately portraying how students actually interact on a campus. The deterministic assumed that everyone was known to and influenced by everyone else equally. So, the small world network model was created. Items representing the students reside within social ‘circles’ of influence, with random connections to individuals in other circles. This is the same way that, while people in circles certain dorms or departments will know fewer, random people from other groups on campus. In the social network model, students are also tagged with influence factors that represent how persuasive they could be in influencing others drinking habits, much in the way that campus and organization leaders are looked up to more by other students. The social network model still requires work to implement other factors, such as how easily students could get alcohol. However, it is a better representation of an actual campus environment. It offers the promise of ability to configure better for larger campus populations, students that might have fewer people in their close social circle, and more variation in the number of people any given student knows. No solid plans are in place to use this in real-world applications at this time, but there is certainly potential in the models for predicting risk propagation in campuses, and could even be used t |
| < Prev | Next > |
|---|