NAME: Drunkwalks.txt TYPE: Quasi-experiment SIZE: 151 Observations, 7 variables ARTICLE TITLE: The Not-so-Random Drunkard's Walk DESCRIPTIVE ABSTRACT: This dataset contains the results of a quasi-experiment testing Karl Pearson's "drunkard's walk" analogy for an abstract random walk. Inspired by the alternate hypothesis that drunkards stumble to their dominant hand's side, it includes observations of intoxicated test subjects walking a 10' line. Variables include: the direction to which subjects first stumbled, the side of the line on which they ended up, and a record of subjects' dominant hand. SOURCES: Data gathered by students in a class project. VARIABLE DESCRIPTIONS: CASE: each observation is numbered sequentially from 1-151. DRINKS: each subject self-reported how many alcoholic drinks they had consumed, as reported in this variable, with no distinction made for type of drink (beer, wine, etc) or time since consumption. Students were instructed, however, to only test subjects that were visibly drunk. The values range from 2-14. SEX: students noted the subject's sex. Males are coded as 1, females as 0. DOMINANT_SIDE: each subject self-reported their dominant side, measured as the hand with which they write. Right-handed observations were coded as 1, and left-handed as 0. FIRST_STUMBLE: students recorded the direction of each subjects' first stumble off the line, whether right (coded as a 1) or left (coded as 0). END_POSITION: students recorded each subjects' final position, whether right of the line (coded as a 1) or left of the line (coded as 0). In the original data, two observations have missing values for this variable: one accurately returned to the line, the other collapsed in mid-walk and was unable to finish, so these were removed during data cleaning. MATCH: The final variable measures whether the subjects stumbled to the same side as their dominant hand. A right-handed person would be coded 1 for stumbling right and 0 for stumbling left (and vice versa for left-handers) PEDAGOGICAL NOTES: Useful for teaching tests of sample proportions SUBMITTED BY: George Ehrhardt Appalachian State University Government and Justice Studies Department, Boone, North Carolina, 28608 ehrhardtgc@appstate.edu