Data and Analysis Unit: liv001
Citation: Lawson et al. (2017)
Description
Free sorting experiments reported by Lawson et al. (2016). This DAU
contains raw freesort data, stimulus lists, and analysis scripts. The
following materials are covered by a Creative Commons Share-Alike 4.0
license.
Resources
- data.csv - Trial-level raw data, column
headings are as follows:
- Expt.PSYC. : Experiment code. Relative to Lawson et al. (2016),
exp.1 = 28, exp.2 = 9, exp.3 = 58, exp.4 = 25. The file also
contains experiment code 23. Experiment code 23 is an experiment not
reported in Lawson et al. (2016) due to space limitations.
- ExptGroup : Experimental condition; meaning is dependent on
experiment:
- PSYC28 : 1 = pictures, 2 = words
- PSYC9: 2 = tax primed, 3 = them primed, 4 = not primed
- PSYC58: 2 = them instructed, 3 = tax instructed
- PSYC25: 2 = them instructed, 3 = tax instructed
- SjNo : Participant number (unique within an experiment)
- CategoryNo : For each participant, CategoryNo ranges from 1 to the
number of groups that participant created. Within a given
participant, items with the same category number were put into the
same group. Where a classification decision was not recorded due to
human error, or where the participant placed the item into a 'junk'
category (an option only available in PSYC25), an 'NA' is recorded.
- ItemIndex : For each category for each participant, ItemIndex
ranges from 1 to the number of items in the category. Where a
classification decision was not recorded due to human error, an 'NA'
is recorded. Hence 'junk' decisions and missing data can be
distinguished because the former have NA for both CategoryNo and
ItemIndex, while the latter have NA only for CategoryNo.
- ItemNo : A number identifying the classified
stimulus. stim_names.csv maps these numbers on to object names.
- stim_names.csv - Mapping of ItemNo
in data.csv to the names of the objects presented. The mapping was
different for PSYC28 ('E1') than for the remaining studies PSYC9,
PSYC23, PSYC58, PSYC25 ('E234').
- analysis.R - R script for analyses
reported in Tables 1 and 2 of Lawson et al. (2016). In addition to
data.csv, it requires the following to run:
- Installation of R package freesortphi. Within R, type:
install.packages("freesortphi",
repos="http://R-Forge.R-project.org")
- mcout.rda - Archived results of Monte Carlo
simulations of the chance levels (see montecarlo.R)
- montecarlo.R - R script to produce
mcout.rda. Currently, it takes about 12 hours to run on an Intel i7
'8 core' (4 cores with hyperthreading) machine using package
freesortphi version 0.1.3. Requires R packages: 'freesortphi',
'snowfall', and 'rlecuyer'. The last two of these packages are
available from CRAN (i.e. are a standard install from within R). For
the first, see above. In the R script, set cpus appropriate to your
machine.
References
- Lawson, R., Chang, F. & Wills, A.J. (2017). Free classification of
large sets of everyday objects is more thematic than
taxonomic. Manuscript accepted by Acta Psychologica.