Skip to content

Commit

Permalink
Fix simdat_crossed for between-subject/between-item factors (#66)
Browse files Browse the repository at this point in the history 
* Adapted simdat_crossed for between-subject/between-item factors and added new testsets
  • Loading branch information
@jschepers
jschepers authored Feb 16, 2024
1 parent fe3b36a commit b8d6b75
Show file tree
Hide file tree
Showing 2 changed files with 173 additions and 6 deletions.
31 changes: 25 additions & 6 deletions src/simdat.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -72,19 +72,38 @@ function simdat_crossed(rng::AbstractRNG, subj_n=1, item_n=1;
item_names = vcat(["item"], ib_vars)
item = (; (Symbol(k) => v for (k,v) in zip(item_names, item_vals))...)

# set up within both table
if (isnothing(both_win))
# cross the subject and item tables
design = factorproduct(subj, item)
# Check whether there are experimental factors which are both between-subject and between-item
if isnothing(subj_btwn) || isnothing(item_btwn)
both_between = []
else
both_between = intersect(keys(subj_btwn), keys(item_btwn))
end

# Case where there are not factors that are both within subject and within item
if isnothing(both_win)
# and there are no factors that are both between subject and between item
if isempty(both_between)
# cross the subject and item tables
design = factorproduct(subj, item)
else
# make sure that each subject/item is only in one level of the between subject/between item factor
design = [merge(x, y) for x in rowtable(subj), y in rowtable(item) if all(x[var] == y[var] for var in both_between)]
end
else
# set up within both table
wc = values(both_win) |> collect
win_prod = Iterators.product(wc...)
win_vals = columntable(win_prod) |> collect
win_names = collect(keys(both_win))
win = (; (Symbol(k) => v for (k,v) in zip(win_names, win_vals))...)

# cross the subject and item tables with any within factors
design = factorproduct(subj, item, win)
if isempty(both_between)
# cross the subject and item tables with any within factors
design = factorproduct(subj, item, win)
else
# make sure that each subject/item is only in one level of the between subject/between item factor
design = [merge(x, y, z) for x in rowtable(subj), y in rowtable(item), z in rowtable(win) if all(x[var] == y[var] for var in both_between)]
end
end

dv = randn(rng, length(design))
Expand Down
148 changes: 148 additions & 0 deletions test/simdat.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -39,3 +39,151 @@ using Test
@test Tables.isrowtable(dat)
@test Set(Tables.columnnames(first(dat))) == Set([:subj, :item, :dv])
end

@testset "simdat_crossed between-subjects between-items - simple case" begin
# stimulate data for a case in which a factor is both between-subject and between-item
conditions = Dict(:cond => ["A", "B"])
cond_n = length(conditions[:cond])
subj_n = 2
item_n = 4

data = DataFrame(simdat_crossed(subj_n, item_n,
subj_btwn = conditions,
item_btwn = conditions))

@test nrow(data) == subj_n * (item_n/cond_n)

# Test whether each subject is only in one of the conditions
for s in nlevels(subj_n,"S")
@test length(unique(data[isequal.(data.subj, s), :cond])) == 1
end

# Test whether each item is only in one of the conditions
for i in nlevels(item_n,"I")
@test length(unique(data[isequal.(data.item, i), :cond])) == 1
end

end;

@testset "simdat_crossed between-subjects between-items - complex case" begin
# stimulate data for a case in which a factor is both between-subject and between-item
both_btwn = Dict(:cond1 => ["A", "B"],
:cond2 => ["C", "D", "E"])
subj_btwn = merge(Dict(:age => ["O", "Y"]), both_btwn)
item_btwn = merge(Dict(:pet => ["cat", "dog"]), both_btwn)
both_win = Dict(:time => ["morning", "evening"])

subj_n = 12
item_n = 12

data = DataFrame(simdat_crossed(subj_n, item_n,
subj_btwn = subj_btwn,
item_btwn = item_btwn,
both_win = both_win))

# Test whether each subject is only in one of the levels of the between-subject/between-items conditions
for s in nlevels(subj_n, "S"), cond in [:cond1, :cond2]
@test length(unique(data[isequal.(data.subj, s), cond])) == 1
end

# Test whether each item is only in one of the conditions
for i in nlevels(item_n, "I"), cond in [:cond1, :cond2]
@test length(unique(data[isequal.(data.item, i), cond])) == 1
end

end;

@testset "simdat_crossed test all combinations" begin
both_btwn = Dict(:cond1 => ["A", "B"],
:cond2 => ["C", "D", "E"])
subj_btwn = merge(Dict(:age => ["O", "Y"]), both_btwn)
item_btwn = merge(Dict(:pet => ["cat", "dog"]), both_btwn)
both_win = Dict(:time => ["morning", "evening"])


subj_n = 12
item_n = 12
data = DataFrame(simdat_crossed(subj_n, item_n,
subj_btwn=subj_btwn,
item_btwn=item_btwn,
both_win=both_win))
s2 = subset(data, :subj => ByRow(==("S02")))
@test all(==("A"), s2.cond1)
@test all(==("Y"), s2.age)
@test length(unique(s2.pet)) == 2 # from item between

#-----
# no subject between
data = DataFrame(simdat_crossed(subj_n, item_n,
item_btwn=item_btwn,
both_win=both_win))

@test nrow(data) == 288 # many more rows because many more effects are within-subject
s2 = subset(data, :subj => ByRow(==("S02")))
@test length(unique(string.(s2.cond1) .* string.(s2.cond2))) == 6 # test all 6 combinations are there
@test length(unique(s2.cond1)) == 2 # now we have both conditions within subject

i2 = subset(data, :item => ByRow(==("I02")))
@test all(==("A"), i2.cond1) # but only one because cond1 is between-items here


#-----
# no item between
data = DataFrame(simdat_crossed(subj_n, item_n,
subj_btwn=subj_btwn,
both_win=both_win))
@test nrow(data) == 288 # many more rows because many more effects are within-subject
i2 = subset(data, :item => ByRow(==("I02")))
@test length(unique(string.(i2.cond1) .* string.(i2.cond2))) == 6 # test all 6 combinations are there
@test length(unique(i2.cond1)) == 2 # now we have both conditions within item

s2 = subset(data, :subj => ByRow(==("S02")))
@test all(s2.cond1 .== "A") # but only one because cond1 is between-subject here

#---------
data = DataFrame(simdat_crossed(subj_n, item_n; both_win))
@test nrow(data) == subj_n * item_n * 2
i2 = subset(data, :item => ByRow(==("I02")))
s2 = subset(data, :subj => ByRow(==("S02")))
@test length(unique(i2.subj)) == 12
@test length(unique(s2.item)) == 12


#------

data = DataFrame(simdat_crossed(subj_n, item_n))
@test nrow(data) == subj_n * item_n


#-----
data = DataFrame(simdat_crossed(subj_n, item_n; subj_btwn, item_btwn))
@test nrow(data) == 24
i2 = subset(data, :item => ByRow(==("I02")))
@test nrow(i2) == 2
@test length(unique(i2.subj)) == 2 # because age is within item, but between subjects, we have two subjects here
@test length(unique(i2.age)) == 2 # because age is within item

#----
data = DataFrame(simdat_crossed(subj_n, item_n,
subj_btwn=subj_btwn))
@test nrow(data) == 12*12
i2 = subset(data, :item => ByRow(==("I02")))
@test length(unique(i2.subj)) == 12
@test length(unique(i2.cond1 .* i2.cond2 .* i2.age)) == 12 # everything is within item
s2 = subset(data, :subj => ByRow(==("S02")))
@test length(unique(i2.subj)) == 12
@test length(unique(s2.cond1 .* s2.cond2 .* s2.age)) == 1 # everything is within item

#--- same as previous but for item
data = DataFrame(simdat_crossed(subj_n, item_n,
item_btwn=item_btwn))
@test nrow(data) == 12 * 12
i2 = subset(data, :item => ByRow(==("I02")))
@test length(unique(i2.subj)) == 12
@test length(unique(i2.cond1 .* i2.cond2 .* i2.pet)) == 1 # everything is within item
s2 = subset(data, :subj => ByRow(==("S02")))
@test length(unique(i2.subj)) == 12
@test length(unique(s2.cond1 .* s2.cond2 .* s2.pet)) == 12 # everything is within item


end

2 comments on commit b8d6b75

@palday
Copy link
Member

@palday palday commented on b8d6b75 Feb 16, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@JuliaRegistrator register

@JuliaRegistrator
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Registration pull request created: JuliaRegistries/General/101007

Tip: Release Notes

Did you know you can add release notes too? Just add markdown formatted text underneath the comment after the text
"Release notes:" and it will be added to the registry PR, and if TagBot is installed it will also be added to the
release that TagBot creates. i.e.

@JuliaRegistrator register

Release notes:

## Breaking changes

- blah

To add them here just re-invoke and the PR will be updated.

Tagging

After the above pull request is merged, it is recommended that a tag is created on this repository for the registered package version.

This will be done automatically if the Julia TagBot GitHub Action is installed, or can be done manually through the github interface, or via:

git tag -a v0.2.8 -m "<description of version>" b8d6b75cdbc0711359b96e309a957331ffa1024f
git push origin v0.2.8

Please sign in to comment.