| Type: | Package |
| Title: | Plot Full ROC Curves using Eyewitness Lineup Data |
| Version: | 0.1.0 |
| Description: | Enable researchers to adjust identification rates using the 1/(lineup size) method, generate the full receiver operating characteristic (ROC) curves, and statistically compare the area under the curves (AUC). References: Yueran Yang & Andrew Smith. (2020). "fullROC: An R package for generating and analyzing eyewitness-lineup ROC curves". <doi:10.13140/RG.2.2.20415.94885/1> , Andrew Smith, Yueran Yang, & Gary Wells. (2020). "Distinguishing between investigator discriminability and eyewitness discriminability: A method for creating full receiver operating characteristic curves of lineup identification performance". Perspectives on Psychological Science, 15(3), 589-607. <doi:10.1177/1745691620902426>. |
| BugReports: | https://github.com/yuerany/fullROC/issues |
| Language: | en-US |
| License: | GPL (≥ 3) |
| Encoding: | UTF-8 |
| LazyData: | true |
| RoxygenNote: | 7.1.1 |
| Imports: | stats, graphics |
| NeedsCompilation: | no |
| Packaged: | 2021-01-09 00:56:52 UTC; yueran |
| Author: | Yueran Yang [aut, cre] |
| Maintainer: | Yueran Yang <yuerany@unr.edu> |
| Repository: | CRAN |
| Date/Publication: | 2021-01-13 11:50:10 UTC |
Bootstrap AUCs
Description
A function to simulate bootstrap samples and calculate AUC.
Usage
auc_boot(
data,
group = NULL,
nboot = 1000,
byDR = FALSE,
ca_adj = FALSE,
lsize = 6,
csize = 3
)
Arguments
data |
A data frame or matrix saving both cp and ca frequencies. cp must precede ca. |
group |
A vector indicating group membership. Will calculate AUCs by group. |
nboot |
Number of bootstrap iterations for each group. Defaults to 1,000. |
byDR |
Whether to order ids by diagnosticity ratios. Defaults to FALSE. |
ca_adj |
Whether to adjust id rates for ca lineups after simulating a sample from the unadjusted rates. |
lsize |
Size of lineup (used to adjust id rates). Defaults to 6. |
csize |
Number of confidence levels (used to adjust id rates). Defaults to 3. |
Value
A list with simulated AUCs.
Examples
cpf <- c(100, 90, 80, 20, 10, 5)
caf <- c(6, 7, 15, 50, 75, 120)
auc_boot(cbind(cpf, caf), nboot = 100)
Bootstrap confidence intervals for AUC
Description
A function to simulate bootstrap samples and calculate CIs for AUC and differences.
Usage
auc_ci(cpf, caf, group = NULL, nboot = 1000, alpha = 0.05, ...)
Arguments
cpf |
A vector of cp frequencies. |
caf |
A vector of ca frequencies. |
group |
A vector indicating group membership. Will calculate AUC by group. |
nboot |
Number of bootstrap iterations. Defaults to 1,000. |
alpha |
Alpha level for the CIs. Defaults to 0.05. |
... |
Additional arguments in auc_boot. Will allow users to adjust ca id rates in each simulation. |
Value
A data frame of CIs for each group and group differences.
References
Yueran Yang & Andrew Smith. (2020). "fullROC: An R package for generating and analyzing eyewitness-lineup ROC curves" doi: 10.13140/RG.2.2.20415.94885/1
Andrew Smith, Yueran Yang, & Gary Wells. (2020). "Distinguishing between investigator discriminability and eyewitness discriminability: A method for creating full receiver operating characteristic curves of lineup identification performance". Perspectives on Psychological Science, 15(3), 589-607. doi: 10.1177/1745691620902426
Examples
cpf1 <- c(100, 90, 80, 20, 10, 5)
caf1 <- c(6, 7, 15, 50, 75, 120)
auc_ci(cpf1, caf1, nboot = 50)
cpf2 <- c(90, 40, 20)
caf2 <- c(10, 70, 80)
auc_ci(cpf2, caf2, nboot = 100)
## compare two groups
cpf <- c(cpf1, cpf2)
caf <- c(caf1, caf2)
group <- rep(letters[1:2], times = c(length(cpf1), length(cpf2) ) )
auc_ci(cpf, caf, group = group)
Simple adjustment
Description
A function to adjust the id rates for ca lineups using the 1/(lineup size) method; is applicable to ordered id rates with the same confidence levels for all responses.
Usage
id_adj(rate, lsize = 6, csize = 3)
Arguments
rate |
ID rate vector. |
lsize |
Lineup size. Defaults to 6. |
csize |
Number of confidence levels. Defaults to 3. |
Value
Adjusted ID vector.
Examples
ca_id <- c(rep(0,3), rep(c(0.1, 0.15, 0.25), 2))
id_adj(ca_id)
## change line size to 5
id_adj(ca_id, lsize = 5)
## For multiple groups
ca_id2 <- c(c(rep(0,3), rep(c(0.1, 0.15, 0.25), 2)),
c(rep(0,3), rep(c(0.1, 0.2, 0.3), 2)) )
group <- rep(c("a", "b"), each = 9)
## Adjust id rates by groups
by(ca_id2, group, id_adj)
Match by confidence levels
Description
A function to adjust the id rates for ca lineups using the 1/(lineup size) method; match and adjust id rates by names of confidence levels for both filler and suspect ids.
Usage
id_adj_name(rate, conf = NULL, fid, sid, lsize = 6)
Arguments
rate |
ID rate vector. |
conf |
Confidence levels for the id rate vector. Default to be NULL. |
fid |
Mapping confidence levels from filler id. |
sid |
To-be-matched confidence levels for suspect id. Must have equal length as fid. |
lsize |
Lineup size. Defaults to 6. |
Value
Adjusted ID vector.
Examples
ca_id <- c(rep(0,3), rep(c(0.1, 0.15, 0.25), 2))
names(ca_id) <- paste0(rep(c("IDS", "IDF", "REJ"), each = 3), c("high", "medium", "low"))
fid_conf <- paste0("IDF", c("high", "medium", "low"))
sid_conf <- paste0("IDS", c("high", "medium", "low"))
id_adj_name(ca_id, fid = fid_conf, sid = sid_conf)
Match by position
Description
A function to adjust the id rates for ca lineups using the 1/(lineup size) method; match and adjust id rates by positions of filler and suspect ids.
Usage
id_adj_pos(rate, fid, sid, lsize = 6)
Arguments
rate |
ID rate vector. |
fid |
Mapping positions from filler id. |
sid |
To-be-matched positions for suspect id. Must have equal length as fid. |
lsize |
Lineup size. Defaults to 6. |
Value
Adjusted ID vector.
Examples
ca_id <- c(rep(0,3), rep(c(0.1, 0.15, 0.25), 2))
id_adj_pos(ca_id, fid = 4:6, sid = 1:3, lsize = 5)
A function to calculate responses from simulated memory distribution
Description
A function to calculate responses from simulated memory distribution
Usage
response_calculate(memory, criterion, id_criterion = NULL, suspect = TRUE)
Arguments
memory |
A simulated data matrix with suspect (column 1) and filler memory. |
criterion |
A vector of judgment criterion. Must have odd number of elements if id_criterion is not specified. |
id_criterion |
A number to define the criterion for id or rejection. Use the middle element of criterion if not specified. |
suspect |
Whether there is a designated suspect. Defaults to TRUE. |
Value
A data matrix of id responses (IDS, IDF, or REJ) and confidence levels.
Simulate witness responses
Description
A function to simulate both CP and CA responses
Usage
response_simu(
guilt_diff,
inno_diff = 0,
n_sim = 1000,
size = 6,
inno_suspect = FALSE,
criterion,
id_criterion = NULL
)
Arguments
guilt_diff |
Mean difference between guilty suspect and filler distributions. |
inno_diff |
Mean difference between innocent suspect and filler distributions. Defaults to 0. |
n_sim |
Number of simulations per condition. Defaults to 1,000. |
size |
Number of lineup members. Defaults to 6. |
inno_suspect |
Whether there is a designated innocent suspect. Defaults to FALSE. |
criterion |
A vector of response criteria. Must have odd number of elements if id_criterion is not specified. |
id_criterion |
A number to define the criterion for id or rejection. Use the middle element of criterion if not specified. |
Value
A data frame including both CP and CA ID responses and confidence levels.
Examples
## Set up response criteria
rc1 <- seq(-1, 3, length.out = 5)
## no designated innocent suspect
response_simu(guilt_diff = 2, criterion = rc1)
## with a designated innocent suspect
response_simu(guilt_diff = 2, inno_diff = 0.2, inno_suspect = TRUE, criterion = rc1)
## define a criterion for id/rejection instead of using the middle criterion
response_simu(guilt_diff = 2, criterion = 0:3, id_criterion = 1)
A function to calculate AUC using non-cumulative response rates.
Description
A function to calculate AUC using non-cumulative response rates.
Usage
roc_auc(cpr, car, group = NULL, byDR = FALSE)
Arguments
cpr |
A vector of cp id rates. |
car |
A vector of ca id rates. |
group |
A vector indicating group membership. If specified, will calculate AUC by group. |
byDR |
Whether to order ids by diagnosticity ratios. Defaults to FALSE. |
Value
Area under the curve.
A function to plot ROC curves.
Description
A function to plot ROC curves.
Usage
roc_plot(cp, ca, group = NULL, byDR = FALSE, grayscale = FALSE, ...)
Arguments
cp |
A vector of cp id rates or frequencies. |
ca |
A vector of ca id rates or frequencies. |
group |
Grouping variable to indicate group membership. Will create an ROC curve and calculate AUC for each group. |
byDR |
Whether to order ids by diagnosticity ratios. Defaults to FALSE. |
grayscale |
Whether to produce the plot in grayscale. Defaults to FALSE. |
... |
Additional plotting parameters.
For example, users can change x-axis and y-axis labels using |
Value
Plot ROC curves and calculate AUCs as side effects.
References
Yueran Yang & Andrew Smith. (2020). "fullROC: An R package for generating and analyzing eyewitness-lineup ROC curves" doi: 10.13140/RG.2.2.20415.94885/1
Andrew Smith, Yueran Yang, & Gary Wells. (2020). "Distinguishing between investigator discriminability and eyewitness discriminability: A method for creating full receiver operating characteristic curves of lineup identification performance". Perspectives on Psychological Science, 15(3), 589-607. doi: 10.1177/1745691620902426
Examples
cpf1 <- c(100, 90, 80, 20, 10, 5)
caf1 <- c(6, 7, 15, 50, 75, 120)
roc_plot(cpf1, caf1)
cpf2 <- c(90, 40, 20)
caf2 <- c(10, 70, 80)
roc_plot(cpf2, caf2)
## plot two ROC curves
cpf <- c(cpf1, cpf2)
caf <- c(caf1, caf2)
group <- rep(letters[1:2], times = c(length(cpf1), length(cpf2) ) )
roc_plot(cpf, caf, group = group)