beerpoll.rss

'"beerpoll" from "correlation_causation.Rmd" starting at line 1271

URN 52#1 26#0 drink
' Constitute the set of 52 beer drinkers, represented by 52 "1"s, and the
' set of 26 non-drinkers, represented by "2"s.
URN 57#1 21#0 sale
' The same set of individuals classified by whether they favor ("1") or
' don't favor ("0") the sale of beer.

' Note: F is now the vector {1 1 1 1 1 1 \... 0 0 0 0 0 \...} where 1 =
' people in favor, 0 = people opposed.
REPEAT 1000
    ' Repeat the experiment 1000 times.
    SHUFFLE drink drink$
    ' Shuffle the beer drinkers/non-drinker, call the shuffled set drink\*.

    ' Note: drink\$ is now a vector like {1 1 1 0 1 0 0 1 0 1 1 0 0 \...}
    ' where 1 = drinker, 0 = non-drinker.
END
SUBTRACT drink$ sale c
' Subtract the favor/don't favor set from the drink/don't drink set.
' Consistent responses are someone who drinks favoring the sale of beer (a
' "1" and a "1") or someone who doesn't drink opposing the sale of beer.
' When subtracted, consistent responses *(and only consistent responses)*
' produce a "0."
COUNT c =0 d
' Count the number of consistent responses (those equal to "0").
COUNT c <> 0 e
' Count the "inconsistent" responses (those not equal to "0").
SUBTRACT d e f
' Find the difference
SCORE f z
' Keep track of the results of each trial.

' End one trial, go back and repeat until all 1000 trials are complete.
HISTOGRAM z
' Produce a histogram of the trial result.

' Note: The file "beerpoll" on the Resampling Stats software disk contains
' this set of commands.