Overview

This report examines the relationship between field goal attempts (GrAtt), game location (HomeIND, where 1 = Home and 0 = Away), and points scored (GrPoint) using the Hoops dataset. Models are evaluated using group equations, visual plots, and \(R^2\) values rather than p-values, t-tests, or confidence intervals.


Step 1: Single-Line Baseline Model (model1)

R Code

model1 <- lm(GrPoint ~ GrAtt, data = my_data)
coef(model1)

ggplot(my_data, aes(x = GrAtt, y = GrPoint)) +
  geom_point() +
  geom_abline(intercept = coef(model1)[1], slope = coef(model1)[2]) +
  labs(
    title = "Grinnell Points vs. Field Goal Attempts (Single Line Model)",
    x = "Field Goal Attempts (shots)",
    y = "Points Scored (points)"
  )

Questions & Answers

  1. Question: Which variable is the response variable and which is the explanatory variable for model1?
    • Answer: GrPoint (Points Scored) is the response variable, and GrAtt (Field Goal Attempts) is the explanatory variable.

Step 2: Indicator Variable / Parallel Lines Model (model2)

R Code

model2 <- lm(GrPoint ~ GrAtt + HomeIND, data = my_data)
coef(model2)

ggplot(my_data, aes(x = GrAtt, y = GrPoint, color = factor(Home), shape = factor(Home))) +
  geom_point() +
  geom_abline(intercept = 25.7126233, slope = 0.9052887, color = "#F8766D") +
  geom_abline(intercept = 35.2677201, slope = 0.9052887, color = "#00BFC4") +
  labs(
    title = "Grinnell Points vs. Field Goal Attempts (Parallel Lines Model)",
    x = "Field Goal Attempts (shots)",
    y = "Points Scored (points)",
    color = "Home (1 = Home, 0 = Away)",
    shape = "Home (1 = Home, 0 = Away)"
  )

Questions & Answers

  1. Question: What is the response variable and what are the two explanatory variables for model2?
    • Answer: GrPoint is the response variable, and GrAtt and HomeIND are the two explanatory variables.
  2. Question: What are all three coefficients ((Intercept), GrAtt, and HomeIND), and what are the intercept and slope for each group’s line?
    • Answer:
      • (Intercept): 25.7126233
      • GrAtt: 0.9052887
      • HomeIND: 9.5550968
      • Away games (HomeIND = 0): \(\hat{y} = 25.7126233 + 0.9052887(\text{GrAtt}) + 9.5550968(0)\) \(\rightarrow\) Intercept: 25.7126233, Slope: 0.9052887
      • Home games (HomeIND = 1): \(\hat{y} = (25.7126233 + 9.5550968) + 0.9052887(\text{GrAtt})\) \(\rightarrow\) Intercept: 35.2677201, Slope: 0.9052887
  3. Question: Interpret the explanatory variable’s coefficient (0.9052887) and the indicator coefficient (9.5550968) in context, using the phrase “for two games with the same…” in your explanations.
    • Answer (GrAtt slope): For two games with the same location (both home or both away), each additional field goal attempt is associated with an average increase of 0.91 points.
    • Answer (HomeIND indicator): For two games with the same number of field goal attempts, Grinnell is predicted to score 9.56 more points on average in a home game than in an away game.
  4. Question: What are the predicted points for an away game and for a home game when Grinnell takes 100 field goal attempts (GrAtt = 100), and how does the difference between those two predictions compare to the indicator coefficient?
    • Answer:
      • Away game (HomeIND = 0): \(\widehat{\text{Points}} = 25.7126233 + 0.9052887(100) + 9.5550968(0) = 116.24\) points.
      • Home game (HomeIND = 1): \(\widehat{\text{Points}} = 25.7126233 + 0.9052887(100) + 9.5550968(1) = 125.80\) points.
      • Comparison: The difference between the two predictions (\(125.80 - 116.24 = 9.56\) points) matches the indicator coefficient (9.5550968). Because model2 is a parallel lines model, the vertical distance between the two regression lines remains constant across all values of field goal attempts.

Step 3: Interaction Variable / Non-Parallel Lines Model (model3)

R Code

my_data <- mutate(my_data, HomeINT = HomeIND * GrAtt)

model3 <- lm(GrPoint ~ GrAtt + HomeIND + HomeINT, data = my_data)
coef(model3)

ggplot(my_data, aes(x = GrAtt, y = GrPoint, color = factor(Home), shape = factor(Home))) +
  geom_point() +
  geom_abline(intercept = 21.94086592, slope = 0.94729430, color = "#F8766D") +
  geom_abline(intercept = 38.69160661, slope = 0.86758618, color = "#00BFC4") +
  labs(
    title = "Grinnell Points vs. Field Goal Attempts (Interaction Model)",
    x = "Field Goal Attempts (shots)",
    y = "Points Scored (points)",
    color = "Home (1 = Home, 0 = Away)",
    shape = "Home (1 = Home, 0 = Away)"
  )

Questions & Answers

  1. Question: Which variable is the response variable, and what are all three explanatory variables that go into model3?
    • Answer: GrPoint (Points Scored) is the response variable, and GrAtt, HomeIND, and HomeINT are the three explanatory variables.
  2. Question: What are all four coefficients ((Intercept), GrAtt, HomeIND, and HomeINT), and what are the intercept and slope for each group’s line?
    • Answer:
      • (Intercept): 21.94086592
      • GrAtt: 0.94729430
      • HomeIND: 16.75074069
      • HomeINT: -0.07970812
      • Away games (HomeIND = 0): \(\hat{y} = 21.94086592 + 0.94729430(\text{GrAtt})\) \(\rightarrow\) Intercept: 21.94086592, Slope: 0.94729430
      • Home games (HomeIND = 1): \(\hat{y} = (21.94086592 + 16.75074069) + (0.94729430 - 0.07970812)(\text{GrAtt})\) \(\rightarrow\) Intercept: 38.69160661, Slope: 0.86758618
  3. Question: Interpret the interaction coefficient (-0.07970812) in context as the difference between the two groups’ slopes.
    • Answer: For every 1-unit increase in GrAtt, predicted points increase by 0.07970812 points less for home games (0.8676 points/attempt) than for away games (0.9473 points/attempt).
  4. Question: Looking at the two lines on your graph: do they look close to parallel, spread apart, or cross within the data?
    • Answer: The two lines look close to parallel across the data, slowly converging slightly from left to right.

Step 4: Model Comparison & Conclusion

R Code

summary(model1)$r.squared
summary(model2)$r.squared
summary(model3)$r.squared

Questions & Answers

  1. Question: What is the \(R^2\) value for each of the three models (model1, model2, and model3)?
    • Answer:
      • model1 \(R^2\): 0.3787588
      • model2 \(R^2\): 0.4383249
      • model3 \(R^2\): 0.4390333
  2. Question: Which model would you choose to best describe the relationship between field goal attempts, game location, and points scored—and why?
    • Answer: model2 best captures the relationship between field goal attempts, game location, and points scored. Adding game location to field goal attempts increases \(R^2\) from 0.3788 to 0.4383, reflecting the clear upward shift seen for home games on the graph. In contrast, moving to model3 by adding the interaction term raises \(R^2\) by a negligible 0.0007, confirming what the plots visually show: the slopes for home and away games are nearly parallel, so a shared slope with separate intercepts accurately captures the relationship.