Set up
First, create an “InteRactModel” R6 object.
act <- interact(dictionary = "usfullsurveyor2015", equations = "us2010")
#> ✔ dictionary = list(dataset = "usfullsurveyor2015", group = "all")
#> ✔ equations = list(key = "us2010", group = "all")
act
#>
#> ── Interact Analysis ───────────────────────────────────────────────────────────
#> ℹ Dictionary: usfullsurveyor2015 (group: all)
#> ℹ Equations: us2010
#> impressionabo (group: all)Dictionary
act$dictionary
#> # A tibble: 2,403 × 5
#> term component ratings n sd
#> <chr> <chr> <list> <list> <list>
#> 1 abandon behavior <dbl [3]> <dbl [3]> <dbl [3]>
#> 2 abandoned modifier <dbl [3]> <dbl [3]> <dbl [3]>
#> 3 abduct behavior <dbl [3]> <dbl [3]> <dbl [3]>
#> 4 abet behavior <dbl [3]> <dbl [3]> <dbl [3]>
#> 5 abhor behavior <dbl [3]> <dbl [3]> <dbl [3]>
#> 6 able_bodied modifier <dbl [3]> <dbl [3]> <dbl [3]>
#> 7 abort behavior <dbl [3]> <dbl [3]> <dbl [3]>
#> 8 abortionist identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 9 absent_minded modifier <dbl [3]> <dbl [3]> <dbl [3]>
#> 10 abuse behavior <dbl [3]> <dbl [3]> <dbl [3]>
#> # ℹ 2,393 more rowsEquations
act$equations
#> Ae' Ap' Aa' Be' Bp' Ba' Oe' Op' Oa'
#> (Intercept) -0.16 -0.06 -0.05 -0.26 0.09 0.21 -0.18 0.00 -0.14
#> Ae 0.62 0.00 0.00 0.52 0.00 0.00 0.00 0.00 0.00
#> Ap 0.00 0.57 0.00 0.00 0.53 0.00 0.00 0.00 0.00
#> Aa 0.00 0.00 0.26 0.00 0.00 0.15 0.00 0.00 0.00
#> Be 0.46 -0.29 -0.18 0.50 -0.21 -0.12 0.26 0.51 0.33
#> Bp 0.00 0.62 0.12 0.00 0.64 0.00 0.00 -0.57 -0.41
#> Ba 0.00 0.00 0.86 0.00 0.00 0.91 0.00 0.00 0.31
#> Oe 0.00 0.00 0.00 0.00 0.00 0.00 0.91 0.00 0.00
#> Op 0.00 -0.20 0.00 0.00 0.00 0.00 0.00 0.47 0.00
#> Oa 0.00 0.00 0.00 0.00 0.00 0.09 0.00 0.00 0.54
#> Ae:Be 0.00 0.00 0.00 0.00 0.00 0.00 0.16 0.00 0.54
#> Be:Oe 0.29 0.00 0.00 0.32 0.00 0.00 0.11 0.00 0.00
#> Bp:Oe -0.22 0.00 0.00 -0.27 0.00 0.00 -0.15 0.00 0.00
#> Be:Oa -0.09 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
#> Aa:Ba:Oe 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.10 0.00interact() uses group = all by default for
both equations and dictionaries. If this option does not exist in the
actdata package, then you will see an error.
interact(dictionary = "indiana2003", equations = "nc1978")
#> ✔ dictionary = list(dataset = "indiana2003", group = "all")
#> ✔ equations = list(key = "nc1978", group = "all")
#> ! equations groups: male and female
#> Error:
#> ! `all` not found in `nc1978` equations in `actdata` packageYou can change the defaults by specifying a second element in either argument.
For example:
interact(dictionary = "indiana2003", equations = list("nc1978", "male"))
#> ✔ dictionary = list(dataset = "indiana2003", group = "all")
#>
#> ── Interact Analysis ───────────────────────────────────────────────────────────
#> ℹ Dictionary: indiana2003 (group: all)
#> ℹ Equations: nc1978
#> impressionabo (group: male)InteRactModel objects come with built in methods which
you can access via the $ operator.
act$fundamentals("deadbeat")
#> # Source: ()
#> # A data frame: 1 × 5
#> term component e p a
#> * <chr> <chr> <dbl> <dbl> <dbl>
#> 1 deadbeat identity -2.8 -2.57 -1.45Note. This is just a simple function that looks inside
act$dictionary.
Deflection
d <- act$deflection(data.frame(A = "god", B = "kill", O = "deadbeat"))
d
#> # Event deflection
#> # A data frame: 1 × 4
#> A B O deflection
#> * <chr> <chr> <chr> <dbl>
#> 1 god kill deadbeat 173.You can also extract useful metadata from these scores.
get_fundamentals(d)
#> # A tibble: 1 × 9
#> Ae Ap Aa Be Bp Ba Oe Op Oa
#> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 3.19 3.45 -0.46 -4.26 1.95 -0.11 -2.8 -2.57 -1.45
get_transients(d)
#> # A tibble: 1 × 9
#> Ae Ap Aa Be Bp Ba Oe Op Oa
#> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 3.96 4.86 0.737 4.56 4.06 0.422 -3.88 -4.48 -10.5
get_element_wise_deflection(d)
#> # A tibble: 1 × 9
#> Ae Ap Aa Be Bp Ba Oe Op Oa
#> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 0.597 2.00 1.43 77.8 4.46 0.283 1.16 3.64 81.9Behaviors and Reidentification
Optimal behaviors and/or reidentification uses the deflection scores
estimated by $deflection() and stored in d.
They require you specify the perspective of the one doing the behaving
or the reidentifying.
Behaviors
act$optimal_behavior(d, who = "actor")
#> # A tibble: 1 × 3
#> Be Bp Ba
#> <dbl> <dbl> <dbl>
#> 1 1.90 1.18 -0.637
act$optimal_behavior(d, who = "object")
#> # A tibble: 1 × 3
#> Be Bp Ba
#> <dbl> <dbl> <dbl>
#> 1 0.111 -0.525 0.233Reidentification
act$reidentify(d, who = "actor")
#> # A tibble: 1 × 3
#> Ae Ap Aa
#> <dbl> <dbl> <dbl>
#> 1 -0.999 2.35 0.848
act$reidentify(d, who = "object")
#> # A tibble: 1 × 3
#> Oe Op Oa
#> <dbl> <dbl> <dbl>
#> 1 -1.26 -4.85 -11.1Closest Terms
act$closest_terms(list(e = 1, p = 0, a = -1), component = "behavior", max_dist = 0.5)
#> obey bow_to nuzzle comply_with curtsey_to gaze_at
#> 0.0581 0.1419 0.2141 0.2822 0.2948 0.4301
#> stroke
#> 0.4806
deadbeat <- act$fundamentals("deadbeat")
deadbeat
#> # Source: ()
#> # A data frame: 1 × 5
#> term component e p a
#> * <chr> <chr> <dbl> <dbl> <dbl>
#> 1 deadbeat identity -2.8 -2.57 -1.45
act$closest_terms(deadbeat, component = "modifier", max_dist = 0.5)
#> uneducated incompetent helpless cowardly poor unemployed
#> 0.1581 0.2489 0.3126 0.3507 0.3901 0.4555Deference Score
You can also create the deference scores discussed by Freeland & Hoey (2018).
But this requires to modify the existing dictionary so we get the right combination of identities and behaviors.
This seems to be the easiest way to do this:
act <- interact()
#> ✔ dictionary = list(dataset = "usfullsurveyor2015", group = "all")
#> ✔ equations = list(key = "us2010", group = "all")
defer_to <- act$dictionary |> ## save for later use
dplyr::filter(term == "defer_to")
occ <- interact(dictionary = "occs2019")
#> ✔ equations = list(key = "us2010", group = "all")
#> ✔ dictionary = list(dataset = "occs2019", group = "all")
occ$dictionary
#> # A tibble: 650 × 5
#> term component ratings n sd
#> <chr> <chr> <list> <list> <list>
#> 1 911_dispatcher identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 2 accountant identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 3 actor identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 4 actress identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 5 actuary_for_an_insurance_company identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 6 acupuncturist identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 7 administrative_assistant identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 8 advertising_executive identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 9 advertising_salesman identity <dbl [3]> <dbl [3]> <dbl [3]>
#> 10 aerobics_instructor identity <dbl [3]> <dbl [3]> <dbl [3]>
#> # ℹ 640 more rows
occupations <- occ$dictionary$term ## save for later use
## replace original dictionary
occ$dictionary <- dplyr::bind_rows(defer_to, occ$dictionary)
#> ✔ added new dictionaryNote that a message appeared signaling that the replacement was successful.
occ
#>
#> ── Interact Analysis ───────────────────────────────────────────────────────────
#> ℹ Dictionary: External [!] (group: ?)
#> ℹ Equations: us2010
#> impressionabo (group: all)Now you just create another grid of events, calculate the deflection scores, and average over the As.
events <- tidyr::crossing(
A = occupations,
B = "defer_to",
O = occupations
)
output <- occ$deflection(events)
output |>
filter(A != O) |>
group_by(A) |>
summarize(avg = mean(deflection)) |>
arrange(desc(avg))
#> # A tibble: 650 × 2
#> A avg
#> <chr> <dbl>
#> 1 firefighter 16.6
#> 2 fireman 15.0
#> 3 paramedic 13.9
#> 4 professional_athlete 12.4
#> 5 fire_department_lieutenant 12.3
#> 6 ambulance_driver 11.4
#> 7 auctioneer 11.1
#> 8 dynamite_blaster 10.9
#> 9 911_dispatcher 10.9
#> 10 surgeon 10.6
#> # ℹ 640 more rowsExperimental
The things here are still experimental. The API may change a lot.
Modify Identity
You can create modified identities by combining them with modifier
terms in the ACT dictionaries. These use the traitid
equations in the actdata package.
The $modify_identity method does not work unless you
first specify the traitid equation.
act$modify_identity(list(M = "angry", I = "doctor"))
#> Error in `act$modify_identity()`:
#> ! must first set up `traitid` equation with the `$add_equation` methodIn order to do this you must use the $add_equation()
method.
act$add_equation(type = "traitid", group = "all")
#> ✔ traitid = list(key = "us2010", group = "all")
act$modify_identity(list(M = "angry", I = "doctor"))
#> # A tibble: 1 × 3
#> term component ratings
#> <chr> <chr> <list>
#> 1 angry__doctor identity <dbl [3]>
act$fundamentals("academic")
#> # Source: ()
#> # A data frame: 1 × 5
#> term component e p a
#> * <chr> <chr> <dbl> <dbl> <dbl>
#> 1 academic identity 2.34 2.26 -0.12
grid <- data.frame(
M = c("tired", "taciturn", "angry", "happy"),
I = c("academic", "academic", "academic", "academic")
)
out <- act$modify_identity(grid)
out |> tidyr::unnest_wider(ratings)
#> # A tibble: 4 × 5
#> term component e p a
#> <chr> <chr> <dbl> <dbl> <dbl>
#> 1 tired__academic identity -0.233 0.548 -1.53
#> 2 taciturn__academic identity 0.694 0.931 -0.834
#> 3 angry__academic identity -1.11 1.83 0.786
#> 4 happy__academic identity 3.10 2.42 0.276Characteristic Emotion
act$add_equation("emotionid", "male")
#> ✔ emotionid = list(key = "us2010", group = "male")
act$characteristic_emotion(list(I = "brute"))
#> # A tibble: 1 × 3
#> Me Mp Ma
#> <dbl> <dbl> <dbl>
#> 1 -1.40 3.05 1.46
grid <- act$dictionary |>
dplyr::filter(component == "identity") |>
dplyr::sample_n(size = 10) |>
dplyr::select(I = term)
glimpse(grid)
#> Rows: 10
#> Columns: 1
#> $ I <chr> "censor", "grandson", "infant", "australian", "job_seeker", "truck_d…
act$characteristic_emotion(grid)
#> # A tibble: 10 × 3
#> Me Mp Ma
#> <dbl> <dbl> <dbl>
#> 1 1.82 0.759 0.945
#> 2 -0.289 2.62 0.230
#> 3 1.81 2.11 0.377
#> 4 2.43 0.692 1.42
#> 5 -4.82 -0.686 0.599
#> 6 2.43 -2.71 1.26
#> 7 2.05 -0.0783 0.893
#> 8 2.49 3.90 0.527
#> 9 1.56 1.10 0.883
#> 10 1.66 -0.279 0.347