Mapping the Hometowns of University of Nebraska-Lincoln Football Recruits, 2000-2018

Abstract

The University of Nebraska switched athletic conferences in 2011 from the Big XII Conference to the Big Ten Conference. The geography of athletic conferences has impact on what schools’ recruits chose, which is important given the financial implications of college athletics. The purpose of this study is to ascertain whether the geography of Nebraska’s football recruits changed with the conference switch. Using the r-package Rvest, hometowns of Nebraska’s recruits were extracted from the University of Nebraska’s football website from the years 2000-2018, then geocoded. The geocode and subsequent calculations of mean centers revealed a possible correlation between the conference switch and the hometowns of Nebraska recruits, however the results are inconclusive.

Background

The purpose of this study is to understand geographic variability in the University of Nebraska’s football recruiting. The University of Nebraska switched athletic conferences in 2011, from the Big Twelve Conference to the Big Ten Conference. The main difference between the two conferences is the geography. Schools in the Big XII Conference include states with high prevalence of high school football culture, like Texas, while schools in the Big Ten Conference include states where high school football may not be as culturally significant (i.e. the midwest). With Nebraska’s switch into the Big Ten, I hypothesize the mean center of recruits has shifted north. Dumand, Lynch, and Platania (2008) state the number one factor in a high school recruits’ choice in college is the geographic distance between hometown and college. Although Nebraska and Texas aren’t that close geographically, Nebraska did travel down to Texas at least once a year for an away game, which Texas recruits can invite their family and friends to those games. However, with Nebraska’s switch into the Big Ten Conference, they stopped traveling to Texas on an annual basis, which may have affected their ability to recruit in Texas.

This study is important because of the enormous financial impact college athletics have on universities. College athletics provide direct financial gain for universities, in that universities directly profit off their football programs (given they are successful) (Goff 2000). Collegiate athletics also can increase exposure for universities, which in turn increases financial contributions from people and increase enrollment numbers (Goff 2000). And the more successful the program, the more financially successful the university is.

Methods and Data

19 years of recruits’ hometowns were compared between 2000 and 2018. Those years were selected because they represent Nebraska in the Big XII conference, their transition into the Big Ten Conference, and Nebraska rooted in the Big Ten Conference. The University of Nebraska’s football website was used, in which they have an extensive database on every player and their hometown throughout their football history.

Once the years were determined, the data was extracted using R, specifically the package “Rvest.” The package Rvest is used to scrape data from webpages and using the code below I extracted the hometowns of football players.

 web.page <- read_html(page)
  #web.page <- read_html("http://www.huskers.com/SportSelect.dbml?DB_OEM_ID=100&SPID=22&SPSID=4&KEY=&Q_SEASON=2018") # sample for single year

  # used selector gadget to get what i wanted
  home.town <- web.page %>%
    html_nodes(".hometown .data")

  # further process data with rvest
  home.town.text <- html_text(home.town)

  # remove new line and tab characters
  home.town.text <- gsub("[\n\t]", "", home.town.text)

  # remove high school name (in parentheses)
  home.town.text <- gsub(" \\(.*","", home.town.text)

  # get the city on its own
  city <- sub('\\s*,.*','', home.town.text)

# get csv of old/new state names
states <- data.frame(read.csv("Q:/StudentCoursework/Haffnerm/GEOG.435.001.2191/MOENAD/Projects/Project_Part_3/State_Abbr_Changes.csv"))

# get the bad state abbreviation on its own
  state.old <- sub('.*, ', '', home.town.text)

  # get new state name based on our manually looked up changes
  state.new <- states$good[match(state.old, states$bad)]

  # create new column of city, state
  city.state <- paste0(city, ", ", state.new)
  
  # use osm geocoder to extract the data
  home.town.loc <- osm_search(query = city.state, key = key)

  # put results in a dataframe
  df <- data.frame("hometown" = home.town.text[1:nrow(home.town.loc)],
                   "lon" = home.town.loc$lon,
                   "lat" = home.town.loc$lat)

  # save the .csv to a file
  write.csv(df, paste("nb-recruits-", year, ".csv", sep = ""), row.names = FALSE)
}

Results

The data produced relatively accurate geocoded results, as evident by the map below. The map below is an example for the year 2000. There were a couple of inaccurate geocodes, which points in Czech Republic and Namibia, however most points geocoded correctly.

Discussion

Nebraska switching conferences seems to exhibit some impact on the where their recruits come from. The general trend of mean centers seems to be moving eastward away from Texas, which might mean Nebraska is recruiting players from Big Ten Conference states. However, the mean center trend isn’t moving eastward on a consistent year-to-year basis. The mean center for 2014 is the farthest east, however the trend moves west in 2015 and 2016. The mean centers move east again for the years 2017 and 2018, but there doesn’t seem to be a definitive correlation between mean center and year other than it generally moved east.

Standard deviations also seem to illustrate a general trend after the conference switch. The standard deviations seem to shrink after the conference switch, especially in the years 2016-2018 when Nebraska firmly rooted itself in the Big Ten Conference. The standard deviations in 2016-2018 fall between 11-13 longitude and 4-5 latitude, while in the years preceding they fall generally fall between 14-21 longitude and 5-7 latitude.

The data not geocoding correctly is a limitation in my study. The 2000 Geocoded Result Map showed some points in Czech Republic and Namibia, while upon further examination, no recruits came from those countries. The inaccurate geocodes might have thrown off the mean center, however only 2-3 points geocoded inaccurately per year. Another limitation of my study is Nebraska potentially losing recruits from Texas may not be directly correlated with the conference switch. Nebraska has had four head coaches since 2004, and coaches might have relationships with a certain geographical area that doesn’t account for a conference switch (i.e. the University of Wisconsin has a recruiting relationship with New Jersey running backs). Also, mean centers are not perfect representations of averages because of longitude disparity. Longitudes are not the same distances between every line on Earth, so while mean centers are relatively accurate, they are not perfect.

Conclusion

Nebraska might be losing recruits from Texas because of the conference switch, however the analysis is inconclusive. The general trend of the mean centers seems to move eastward, possibly meaning Nebraska is recruiting from Big Ten Conference states. Nebraska switching conferences has implications on where they obtain their recruits from, which might determine how well they do if my hypothesis. Texas is a high school recruiting hotspot; in 2018, they produced the second most recruits the ESPN Top 300 Recruits. Recruits can change the culture of a football program to a winning culture. However, if Nebraska misses out on Texas recruits because of their conference switch, they might never land a big-name recruit from Texas to change their program, which has direct financial implications (Goff 2000). For future studies, I would like to analyze the recruiting characteristics of each coach in Nebraska’s history, and where each coach recruited from.

References

Dumond, J. Michael, Allen K. Lynch, and Jennifer Platania. 2008. “An Economic Model of the College Football Recruiting Process.” Journal of Sports Economics 9, no. 1 (February): 67-87.

Goff, B. 2000. “Effects of University Athletics on the University: A Review and Extension of Empirical Assessment.” Journal of Sports Management 14, no. 2 (April): 85-104.