Understanding R's shuffle() Function

Group	Outcome	Score_shuf

Initial b₁ (group difference) = —

Shuffle 1

Model

Shuffle 2

Dist of b₁

Graphing Shuffles

df$Score_shuf <- shuffle(df$Score)
gf_point(Score_shuf ~ Condition, data=df)

A Two-Group Model: Y_i = b₀ + b₁X_i + e_i

▶ Graph the data

gf_point(Score ~ Condition, data=df)

▶ Fit the model

Y_i = b₀ + b₁X_i + e_i

lm(outcome ~ group, data=df)

▶ Generate b₁

b1(outcome ~ group, data=df)

Simulating a World Where β₁ is 0

▶ Graph shuffles

df$Score_shuf <- shuffle(df$Score)
gf_point(Score_shuf ~ Condition, data=df)

▶ Model shuffles

df$Score_shuf <- shuffle(df$Score)
b1(Score_shuf~Condition, data=df)

A Sampling Distribution of b₁ Assuming β₁=0

df$Score_shuf <- shuffle(df$Score)
b1(Score_shuf~Condition, data=df)

Sampling Distribution of b₁

0 shuffles

Original difference Shuffles Most recent

Mean of Sampling Distribution