guilty_bot: guilty_bot

In lorenc5/Rmturkcheck: Tools for Wave 1, Wave 2 Panel Surveys

Description

Detects guilty bots in Mturk data based on Qualtrics lat/long coordinates

Usage

  guilty_bot(lat, long, dat, more_than=10)

Arguments

lat	Column name (character) of survey latitude coordinate, typically generated via Qualtrics
long	Column name (character) of survey longitude coordinate, typically generated via Qualtrics
dat	data.frame() object, housing survey responses, probably downloaded from Qualtrics
more_than	more_than = numeric number, default = 10

Value

Returns dataframe() object of all rows fitting "more_than" criterion. Users should then examine lat/long overlaps. Turk IDs, IP addresses, and demographics may all be different, but 15 responses from the exact same lat/long coordinates seems unlikely. Users can then examine these cases for botness with various validity checks.

Author(s)

Loren Collingwood <[e-mail]>

Examples

######################
# guilty_bot Example #
######################

n <- 100
lat <- c( rep("42.999967", n), rnorm(n*4, mean = 44))
long <- c( rep("-80.444825", n), rnorm(n*4, mean = -83))
toy1 <- rnorm(n*5, 20)
toy2 <- rnorm(n*5, 30)
toy3 <- rnorm(n*5, 10)
toy_turk_id <- paste("faux_id:", round(runif(n*5, 0, 2),6), sep="")
df <- data.frame(lat, long, toy1, toy2, toy3, toy_turk_id, stringsAsFactors = F)

# Prints out problem coordinates and returns original dataset but only bots
no_soup_for_you <- guilty_bot(lat = "lat", long = "long", dat = df)

# Keep only non-bot data
keep <- !df$toy_turk_id %in% no_soup_for_you$toy_turk_id
df_bot_free <- df[keep,]; uniqueN(df_bot_free$toy_turk_id)
head(df_bot_free)

lorenc5/Rmturkcheck documentation built on June 5, 2019, 10:59 a.m.