citysize: City-Size Paired-Comparison Task

Description

In a city-size paired-comparison task on each trial, participants judge which of two cities is more populous. After the paired comparisons, participants indicate for each city if they recognize its name. Hilbig, Erdfelder, and Pohl (2010) report a series of experiments to evaluate their model of recognition heuristic use at this task.

The WorldCities data are from a study designed to be similar to Hilbig et al.'s Experiment 6. The 17 cities were (in order of population; Wikipedia, 2016): Shanghai, Tianjin, Tokyo, Seoul, London, Bangkok, Chongqing, Wuhan, Santiago, Rangun, Ankara, Harbin, Kano, Busan, Durban, Ibadan, Montreal.

The ItalianCities data are from a study designed to be similar to Hilbig et al.'s Experiment 7. The 14 cities were: Milan, Naples, Turin, Palermo, Venice, Padua, Taranto, Prato, Reggio Emilia, Perugia, Cagliari, Foggia, Salerno, Ferrara.

Usage

data(citysize)

Arguments

Format

WorldCities A data frame containing 37 observations of six variables:

gender: factor. Participant gender.
age: participant age.
rt: median response time (in seconds) across paired comparisons.
group: factor. The control group (CG) received standard instructions, the experimental group (EG) was instructed to choose the city they recognized whenever possible.
country: number of cities whose country was correctly identified.
y: a matrix of aggregate response frequencies per participant. The column names indicate each of eight response categories: correct/false responses when both cities were recognized (KC, KF), when both were unrecognized (GC, GF), when only one was recognized and the recognized city was chosen (RC, RF), and when only one was recognized and the unrecognized city was chosen (UF, UC).

ItalianCities A data frame containing 64 observations of six variables:

gender, age, rt, y: see above.
group: factor. The control group (CG) received standard instructions, the experimental group (EG) was asked to compare the cities with respect to their elevation above sea level.
knowRH: factor. Does the participant have any knowledge about the recognition heuristic (RH)?

References

Hilbig, B.E., Erdfelder, E., & Pohl, R.F. (2010). One-reason decision-making unveiled: A measurement model of the recognition heuristic. Journal of Experimental Psychology: Learning, Memory, and Cognition, 36(1), 123--134. tools:::Rd_expr_doi("10.1037/a0017518")

Castela, M., Kellen, D., Erdfelder, E., & Hilbig, B.E. (2014). The impact of subjective recognition experiences on recognition heuristic use: A multinomial processing tree approach. Psychonomic Bulletin & Review, 21(5), 1131--1138. tools:::Rd_expr_doi("10.3758/s13423-014-0587-4")

Rettich, A. (2020). Application of the recognition heuristic: An experimental validation of the r-model. Bachelor thesis. University of Tuebingen, Germany. https://osf.io/mz47y/

Wikipedia. (2016). List of cities proper by population. Retrieved Jun 16 from https://en.wikipedia.org/wiki/List_of_cities_proper_by_population.

Examples

Run this code

data(citysize)

## Fit r-model separately for each instruction type
mpt(mptspec("rmodel"), unname(WorldCities[WorldCities$group == "CG",  "y"]))
mpt(mptspec("rmodel"), unname(WorldCities[WorldCities$group == "EG", "y"]))

## Test instruction effect on r parameter
city.agg <- aggregate(y ~ group, WorldCities, sum)
y <- as.vector(t(city.agg[, -1]))

m1 <- mpt(mptspec("rmodel", .replicates = 2), y)
m2 <- mpt(update(m1$spec, .restr = list(r2 = r1)), y)
anova(m2, m1)  # more use of RH with recognition instruction

## Fit r-model separately for each task type
mpt(mptspec("rmodel"),
    unname(ItalianCities[ItalianCities$group == "CG",  "y"]))
mpt(mptspec("rmodel"),
    unname(ItalianCities[ItalianCities$group == "EG", "y"]))

## Test task effect on r parameter
city.agg <- aggregate(y ~ group, ItalianCities, sum)
y <- as.vector(t(city.agg[, -1]))

m3 <- mpt(mptspec("rmodel", .replicates = 2), y)
m4 <- mpt(update(m1$spec, .restr = list(r2 = r1)), y)
anova(m4, m3)  # less use of RH with elevation task

## Plot parameter estimates
par(mfrow = 1:2)
dotchart(coef(m1)[c(4, 1:3)], xlim = 0:1, labels = c("a", "b", "g", "r"),
         xlab = "", main = "by instruction type")
points(coef(m1)[c(8, 5:7)], 1:4, pch = 16)
legend(0, 1, c("none", "recognition"), pch = c(1, 16),
       title = "Instruction", bty = "n")
dotchart(coef(m3)[c(4, 1:3)], xlim = 0:1, labels = c("a", "b", "g", "r"),
         xlab = "", main = "by task type")
points(coef(m3)[c(8, 5:7)], 1:4, pch = 16)
legend(0, 1, c("population", "elevation"), pch = c(1, 16),
       title = "Task", bty = "n")
title("Recognition heuristic use", outer = TRUE, line = -1)
mtext("Parameter estimate (r-model)", side = 1, outer = TRUE, line = -2)

## Compare with original results
Hilbig2010 <- rbind(
    WorldCities.CG = c(462, 204, 290, 272, 740, 205, 77, 62),
    WorldCities.EG = c(500, 307, 279, 264, 902, 235, 68, 29),
  ItalianCities.CG = c(232, 78, 135, 136, 465, 65, 56, 16),
  ItalianCities.EG = c(245, 176, 154, 150, 228, 160, 112, 140)
)
apply(Hilbig2010, 1, mpt, spec = mptspec("rmodel"))