Package {ramchoice}

ramchoice: Revealed Preference and Attention Analysis in Random Limited Attention Models

Description

Preferences and attention are important for understanding decision making, conducting welfare analysis, and providing robust policy recommendations. Decision makers may not pay full attention to all available alternatives, however, which can invalidate standard revealed preference analysis.

This package implements identification, estimation, inference, and specification procedures for the Random Attention Model of Cattaneo, Ma, Masatlioglu, and Suleymanov (2020; doi:10.1086/706861) and the Attention Overload Model of Cattaneo, Cheung, Ma, and Masatlioglu (2026).

The principal RAM and homogeneous-AOM interfaces are revealPref, ramTest, revealAtte, revealPrefModel, aomModel, aomTest, and aomIdentify. The heterogeneous list-based AOM interfaces are hlaoModel, hlaoTest, hlaoNoPITest, hlaoEvent, and hlaoRankings. Data preparation and simulation utilities include sumData, genMat, logitAtte, and logitSimu. The legacy rAtte interface and simulated ramdata dataset are retained for compatibility and illustration.

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

See Also

Useful links:

• https://github.com/mdcattaneo/ramchoice

• https://arxiv.org/abs/2110.10650

• Report bugs at https://github.com/mdcattaneo/ramchoice/issues

Population Identification for Homogeneous AOM

Description

'aomIdentify' implements the mixed-integer characterization of homogeneous Attention Overload. Binary variables encode pairwise comparisons, while totality, transitivity, and the observed 'succ'-Regularity inequalities characterize the sharp set of compatible strict preferences. The routine tests model feasibility and all pairwise revealed-preference conclusions without enumerating the factorial collection of rankings. A status of 2 is treated as a solver-certified infeasibility result; any other nonzero status is reported as a solver error rather than as model incompatibility.

Usage

aomIdentify(menu, prob, tolerance = sqrt(.Machine$double.eps), pairwise = TRUE)

Arguments

Value

An object of class 'ramchoiceAOMIdentification'. It contains model 'compatible', one feasible 'preference' when the model is nonempty, pairwise possibility and revelation results, solver diagnostics, and the mixed-integer system used in the calculation.

References

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

menu <- prob <- matrix(c(
  1, 1, 1,
  1, 1, 0,
  1, 0, 1,
  0, 1, 1
), ncol = 3, byrow = TRUE)
for (i in seq_len(nrow(prob))) {
  prob[i, menu[i, ] == 1] <- logitAtte(sum(menu[i, ]), 2)$choiceProb
}
aomIdentify(menu, prob)

Population Analysis for the Homogeneous Attention Overload Model

Description

'aomModel' evaluates the population choice-probability inequalities implied by a collection of candidate preference orderings under the homogeneous Attention Overload Model (AOM). It provides a model-specific interface to the AOM restrictions implemented by [revealPrefModel()].

Usage

aomModel(
  menu,
  prob,
  pref_list = NULL,
  tolerance = sqrt(.Machine$double.eps),
  attBinary = 1
)

Arguments

Value

An object of class 'ramchoiceAOMModel'. Its 'results' component has one row per candidate preference, including compatibility, inequality counts, and violation magnitudes. The object also contains 'preferences', candidate-specific 'inequalities', the classification 'tolerance', and the complete legacy [revealPrefModel()] result.

References

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

menu <- prob <- matrix(c(
  1, 1, 1,
  1, 1, 0,
  1, 0, 1,
  0, 1, 1
), ncol = 3, byrow = TRUE)
for (i in seq_len(nrow(prob))) {
  prob[i, menu[i, ] == 1] <- logitAtte(sum(menu[i, ]), 2)$choiceProb
}
aomModel(menu, prob, pref_list = rbind(1:3, 3:1))

Sample Inference for the Homogeneous Attention Overload Model

Description

'aomTest' tests candidate preference orderings under the homogeneous AOM and returns a tidy inference table. Row-i.i.d. calculations delegate to [revealPref()] with AOM restrictions only. When 'cluster' is supplied, the function uses cluster-level influence vectors and multiplier critical values while preserving the legacy result for backward-compatible auditing.

Usage

aomTest(
  menu,
  choice,
  pref_list = NULL,
  method = "GMS",
  alpha = 0.05,
  nCritSimu = 2000,
  BARatio2MS = 0.1,
  BARatio2UB = 0.1,
  MNRatioGMS = NULL,
  attBinary = 1,
  cluster = NULL
)

Arguments

Value

An object of class 'ramchoiceAOMTest'. Its 'results' component has one row per preference, method, and nominal level. The object also contains 'preferences', candidate-specific 'inequalities', menu-level 'summary' estimates, 'constraints', inference 'options', elapsed computation time, and the complete legacy [revealPref()] result.

References

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

set.seed(42)
simulated <- lapply(4:2, function(size) {
  logitSimu(n = 10, uSize = 4, mSize = size, a = 2)
})
menu <- do.call(rbind, lapply(simulated, `[[`, "menu"))
choice <- do.call(rbind, lapply(simulated, `[[`, "choice"))
aomTest(
  menu,
  choice,
  pref_list = rbind(1:4, 4:1),
  nCritSimu = 100
)

Generate Constraint Matrices

Description

genMat generates constraint matrices for a range of preference orderings according to (i) the monotonic attention assumption proposed by Cattaneo, Ma, Masatlioglu, and Suleymanov (2020), (ii) the attention overload assumption proposed by Cattaneo, Cheung, Ma, and Masatlioglu (2021), and (iii) the attentive-at-binaries restriction.

This function is embedded in revealPref.

Usage

genMat(
  sumMenu,
  sumMsize,
  pref_list = NULL,
  RAM = TRUE,
  AOM = TRUE,
  limDataCorr = TRUE,
  attBinary = 1
)

Arguments

Value

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

# Load data
data(ramdata)

# Generate summary statistics
summaryStats <- sumData(ramdata$menu, ramdata$choice)

# Generate constraint matrices
constraints <- genMat(summaryStats$sumMenu, summaryStats$sumMsize)
constraints$ConstN
constraints$R[1:10, 1:10]

Structured H-LAO Preference Event

Description

'hlaoEvent' describes the event that one alternative is strictly preferred to every alternative in a supplied comparison set. Unlike an indicator over enumerated rankings, this representation can be priced directly by the H-LAO column-generation algorithm.

Usage

hlaoEvent(alternative, preferred_to, name = NULL)

Arguments

Value

An object of class 'ramchoiceHLAOEvent'.

References

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

hlaoEvent(4, 1, name = "4 above 1")
hlaoEvent(1, c(2, 3, 4), name = "1 top ranked")

Population Analysis for Heterogeneous List-Based Attention Overload

Description

'hlaoModel' recovers list-based reach probabilities and prefix masses on a suffix-closed menu domain. It evaluates recovered-attention restrictions, constructs sharp independent, dependence-robust, or path-independence-robust preference polytopes, and computes sharp bounds for supplied preference events. The first two modes require a suffix-closed domain because they use Sequential Path Independence to recover attention. The '"noPI"' mode treats prefix masses as latent and is available on any observed-menu domain. With full menu data and positive terminal reach, the SPI modes also recover the full-attention choice rule and report Block–Marschak diagnostics. Optional agreement targets measure whether observed and full-attention choices agree. Under benchmark independence, structured events support status-checked column generation with mixed-integer pricing over linear orders. Returned diagnostics report solver statuses, tolerance, reduced costs, primal and dual residuals, an optimality-gap bound, and whether the numerical certificate checks succeeded.

Usage

hlaoModel(
  menu,
  prob,
  outside_prob = NULL,
  list_order = NULL,
  events = NULL,
  dependence = c("independent", "robust", "noPI", "both", "all"),
  tolerance = sqrt(.Machine$double.eps),
  agreement = FALSE,
  algorithm = c("auto", "enumerate", "column_generation"),
  max_rankings = 5000L,
  max_iterations = 1000L
)

Arguments

Value

An object of class 'ramchoiceHLAOModel' containing recovered 'attention', attention 'diagnostics', population 'pairwise' shares, compatibility by dependence mode, event 'bounds', full-attention 'agreement', ranking columns used by the selected algorithm, computation diagnostics, and, when available, 'full_attention' and 'block_marschak' results.

References

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

menu <- rbind(c(1, 0), c(0, 1), c(1, 1))
prob <- rbind(c(.8, 0), c(0, .75), c(.56, .24))
rankings <- hlaoRankings(1:2)
event <- rankings[, 1] == 2
hlaoModel(menu, prob, events = list(`2 above 1` = event))
hlaoModel(
  menu, prob,
  events = hlaoEvent(2, 1, name = "2 above 1"),
  agreement = TRUE
)

Path-Independence-Robust Inference for H-LAO

Description

'hlaoNoPITest' projects a simultaneous confidence region for primitive menu-choice probabilities through the sharp H-LAO model that retains prefix consideration, attention overload, and a stable marginal preference distribution but does not impose Sequential Path Independence. Prefix masses and menu-specific preference–stopping couplings are latent variables. Every reported event endpoint is obtained by linear programming, and the observed menu domain need not be suffix closed.

Usage

hlaoNoPITest(
  menu,
  choice,
  outside = NULL,
  list_order = NULL,
  events = NULL,
  alpha = 0.05,
  band_method = c("hoeffding", "gaussian"),
  n_band_draws = 2000L,
  boundary_count = 5L,
  tolerance = sqrt(.Machine$double.eps),
  max_rankings = 5000L,
  cluster = NULL
)

Arguments

Value

An object of class 'ramchoiceHLAONoPITest' containing event 'intervals', simultaneous probability 'bands', the LP 'projection', options, and elapsed time.

References

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

menu <- rbind(
  matrix(rep(c(1, 0), 20), ncol = 2, byrow = TRUE),
  matrix(rep(c(1, 1), 20), ncol = 2, byrow = TRUE)
)
choice <- matrix(0, nrow = nrow(menu), ncol = 2)
choice[1:15, 1] <- 1
choice[21:30, 1] <- 1
choice[31:36, 2] <- 1
rankings <- hlaoRankings(1:2)
hlaoNoPITest(
  menu, choice,
  events = list(`2 above 1` = rankings[, 1] == 2)
)

Enumerate Strict Preference Rankings

Description

'hlaoRankings' returns all strict rankings of a supplied alternative set in the deterministic ordering used by the H-LAO linear-programming routines.

Usage

hlaoRankings(alternatives)

Arguments

Value

A matrix with one strict ranking per row, from best to worst.

References

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

hlaoRankings(1:3)

Inference for Heterogeneous List-Based Attention Overload

Description

'hlaoTest' forms simultaneous bands for all observed menu–outcome probabilities. The default Hoeffding method is finite-sample valid. The correlated-Gaussian method uses the estimated block-multinomial covariance and retains exact binomial bands for sparse or degenerate cells. When 'cluster' is supplied, the Gaussian component instead uses cluster-level influence vectors and multiplier draws, and the fallback is a cluster-Hoeffding band. When both types of cells are present, each component receives half of the common error budget. The function inverts the undivided binary-menu moments to obtain simultaneous pairwise preference-share intervals that remain valid at zero reach. It also reports a Bonferroni-calibrated studentized inversion of the same moments. The studentized set is obtained by exact quadratic inversion and may therefore contain more than one component; exactly degenerate moments return '[0,1]'. For supplied general preference events, the function also computes the dependence-robust outer projection intervals described in the Supplemental Appendix.

Usage

hlaoTest(
  menu,
  choice,
  outside = NULL,
  list_order = NULL,
  events = NULL,
  alpha = 0.05,
  band_method = c("hoeffding", "gaussian"),
  diagnostic_method = c("outer", "delta"),
  n_band_draws = 2000L,
  boundary_count = 5L,
  tolerance = sqrt(.Machine$double.eps),
  max_rankings = 5000L,
  cluster = NULL
)

Arguments

Value

An object of class 'ramchoiceHLAOTest' containing aggregated choice 'summary', plug-in 'attention' and 'full_attention' estimates, simultaneous probability 'bands', weak-reach 'pairwise' intervals, studentized 'pairwise_studentized' sets and their 'pairwise_studentized_components', optional general-event 'event_intervals', simultaneous specification diagnostics, projection dimensions, options, and elapsed time.

References

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

menu <- rbind(
  matrix(rep(c(1, 0), 10), ncol = 2, byrow = TRUE),
  matrix(rep(c(0, 1), 10), ncol = 2, byrow = TRUE),
  matrix(rep(c(1, 1), 10), ncol = 2, byrow = TRUE)
)
choice <- matrix(0, nrow = nrow(menu), ncol = 2)
choice[1:8, 1] <- 1
choice[11:17, 2] <- 1
choice[21:26, 1] <- 1
choice[27:28, 2] <- 1
hlaoTest(menu, choice)

Compute Choice Probabilities and Attention Frequencies for the Logit Attention Rule

Description

logitAtte computes choice probabilities and attention frequencies for the logit attention rule considered by Brady and Rehbeck (2016). To be specific, for a choice problem S and its subset T, the attention that T attracts is assumed to be proportional to its size: |T|^a, where a is a parameter that one can specify. It will be assumed that the first alternative is the most preferred, and that the last alternative is the least preferred.

This function is useful for replicating the simulation results in Cattaneo, Ma, Masatlioglu, and Suleymanov (2020; doi:10.1086/706861) and Cattaneo, Cheung, Ma, and Masatlioglu (2026).

Usage

logitAtte(mSize = NULL, a = NULL)

Arguments

Value

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

R. L. Brady and J. Rehbeck (2016). Menu-Dependent Stochastic Feasibility. Econometrica 84(3): 1203-1223. doi:10.3982/ECTA12694

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

logitAtte(mSize = 5, a = 2)

Choice Data Simulation Following the Logit Attention Rule

Description

logitSimu simulates choice data according to the logit attention rule considered by Brady and Rehbeck (2016). To be specific, for a choice problem S and its subset T, the attention that T attracts is assumed to be proportional to its size: |T|^a, where a is a parameter that one can specify. It will be assumed that the first alternative is the most preferred, and that the last alternative is the least preferred.

This function is useful for replicating the simulation results in Cattaneo, Ma, Masatlioglu, and Suleymanov (2020; doi:10.1086/706861) and Cattaneo, Cheung, Ma, and Masatlioglu (2026).

Usage

logitSimu(n, uSize, mSize, a)

Arguments

Value

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

R. L. Brady and J. Rehbeck (2016). Menu-Dependent Stochastic Feasibility. Econometrica 84(3): 1203-1223. doi:10.3982/ECTA12694

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

set.seed(42)
logitSimu(n = 5, uSize = 6, mSize = 5, a = 2)

Internal function.

Description

Internal function.

Usage

## S3 method for class 'ramchoiceRevealAtte'
print(x, ...)

Arguments

Internal function.

Description

Internal function.

Usage

## S3 method for class 'ramchoiceRevealPref'
print(x, ...)

Arguments

Internal function.

Description

Internal function.

Usage

## S3 method for class 'ramchoiceRevealPrefModel'
print(x, ...)

Arguments

Revealed Preference Analysis in Random Limited Attention Models

Description

This has been replaced by revealPref.

Usage

rAtte(
  menu,
  choice,
  pref_list = NULL,
  method = "GMS",
  nCritSimu = 2000,
  BARatio2MS = 0.1,
  BARatio2UB = 0.1,
  MNRatioGMS = NULL,
  RAM = TRUE,
  AOM = TRUE,
  limDataCorr = TRUE,
  attBinary = 1
)

Arguments

Value

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Sample Inference for the Random Attention Model

Description

'ramTest' provides tidy candidate-ranking inference for the Random Attention Model of Cattaneo, Ma, Masatlioglu, and Suleymanov (2020). Row-i.i.d. calculations use [revealPref()]. When 'cluster' is supplied, the function retains the same RAM inequalities but estimates their joint covariance from cluster influence vectors and uses multiplier critical values.

Usage

ramTest(
  menu,
  choice,
  pref_list = NULL,
  method = "GMS",
  alpha = 0.05,
  nCritSimu = 2000,
  BARatio2MS = 0.1,
  BARatio2UB = 0.1,
  MNRatioGMS = NULL,
  attBinary = 1,
  limDataCorr = TRUE,
  cluster = NULL
)

Arguments

Value

An object of class 'ramchoiceRAMTest' with the same tidy components as [aomTest()] and a complete legacy [revealPref()] result.

References

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

ramdata: Simulated Choice Data

Description

The file contains a standard choice data of 9,000 observations. There are five alternatives in the grand set.

See revealPref for revealed preference analysis, and revealAtte for revealed attention. sumData is a low-level function that computes summary statistics, and genMat generates constraint matrices subject to given preferences.

Format

menu

Numeric matrix of 0s and 1s, choice problems (1 indicates an alternative in the choice problem and 0 otherwise).

choice

Numeric matrix of 0s and 1s, choices (1 indicates an alternative being chosen).

Revealed Attention Analysis in Random Limited Attention Models

Description

Given a random sample of choice problems and choices, revealAtte returns the upper and lower bounds on the attention frequency following the construction of Cattaneo, Cheung, Ma, and Masatlioglu (2026).

sumData is a low-level function that generates summary statistics. For revealed preference analysis, see revealPref.

Usage

revealAtte(
  menu,
  choice,
  alternative = NULL,
  S = NULL,
  lower = TRUE,
  upper = TRUE,
  pref = NULL,
  nCritSimu = 2000,
  level = 0.95
)

Arguments

Value

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

# Load data
data(ramdata)

# Set seed, to replicate simulated critical values
set.seed(42)

# preference
pref <- matrix(c(1, 2, 3, 4, 5), ncol=5, byrow=TRUE)
# list of choice problems
S <- matrix(c(1, 1, 0, 0, 0,
              1, 1, 1, 0, 0,
              1, 1, 1, 0, 1,
              1, 1, 1, 1, 1), ncol=5, byrow=TRUE)
result <- revealAtte(menu = ramdata$menu, choice = ramdata$choice,
  alternative = c(1,2), S = S,
  lower = TRUE, upper = TRUE,
  pref = pref)
summary(result)

Revealed Preference Analysis in Random Limited Attention Models

Description

Given a random sample of choice problems and choices, revealPref returns test statistics, critical values and p-values against a collection of preferences. Five methods for choosing critical values are available: (i) GMS: generalized moment selection (plug-in (estimated) moment conditions with shrinkage); (ii) PI: critical values based on plug-in estimated moment conditions (this is not uniformly valid); (iii) LF: critical values based on the least favorable model (plug-in 0 for the moment conditions); (iv) 2MS: two-step moment selection; and (v) 2UB: refined moment selection (plug-in upper bound of moment inequalities).

sumData is a low-level function that generates summary statistics, and genMat can be used to construct the constraint matrices. The simulated dataset ramdata is also provided for illustration. For revealed attention analysis, see revealAtte.

Usage

revealPref(
  menu,
  choice,
  pref_list = NULL,
  method = "GMS",
  nCritSimu = 2000,
  BARatio2MS = 0.1,
  BARatio2UB = 0.1,
  MNRatioGMS = NULL,
  RAM = TRUE,
  AOM = TRUE,
  limDataCorr = TRUE,
  attBinary = 1
)

Arguments

Value

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

# Load data
data(ramdata)

# Set seed, to replicate simulated critical values
set.seed(42)

# list of preferences
pref_list <- matrix(c(1, 2, 3, 4, 5,
                      2, 1, 3, 4, 5,
                      2, 3, 4, 5, 1,
                      5, 4, 3, 2, 1), ncol=5, byrow=TRUE)

# revealed preference using only RAM restrictions
result1 <- revealPref(menu = ramdata$menu, choice = ramdata$choice, method = "GMS",
  pref_list = pref_list, RAM = TRUE, AOM = FALSE)
summary(result1)

# revealed preference using only AOM restrictions
result2 <- revealPref(menu = ramdata$menu, choice = ramdata$choice, method = "GMS",
  pref_list = pref_list, RAM = FALSE, AOM = TRUE)
summary(result2)

# revealed preference using both RAM and AOM restrictions
result3 <- revealPref(menu = ramdata$menu, choice = ramdata$choice, method = "GMS",
  pref_list = pref_list, RAM = TRUE, AOM = TRUE)
summary(result3)

# revealed preference employing additional restrictions for binary choice problems
result4 <- revealPref(menu = ramdata$menu, choice = ramdata$choice, method = "GMS",
  pref_list = pref_list, RAM = TRUE, AOM = TRUE, attBinary = 2/3)
summary(result4)

Model Falsification with Random Limited Attention

Description

Given a collection of choice problems and corresponding choice probabilities, revealPrefModel determines if they are compatible with the Random Attention Model (RAM) of Cattaneo, Ma, Masatlioglu, and Suleymanov (2020; doi:10.1086/706861) and/or the Attention Overload Model (AOM) of Cattaneo, Cheung, Ma, and Masatlioglu (2026).

See revealPref for revealed preference analysis with empirical choice data.

Usage

revealPrefModel(
  menu,
  prob,
  pref_list = NULL,
  RAM = TRUE,
  AOM = TRUE,
  limDataCorr = TRUE,
  attBinary = 1
)

Arguments

Value

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

# Logit attention with parameter 2
# True preference: 1 2 3 4 5 6
menu <- prob <- matrix(c(1, 1, 1, 1, 1, 1,
                         0, 1, 1, 1, 1, 1,
                         1, 0, 1, 1, 1, 1,
                         1, 1, 0, 1, 1, 1,
                         1, 1, 1, 0, 1, 1,
                         1, 1, 1, 1, 0, 1,
                         1, 1, 1, 1, 1, 0), ncol=6, byrow=TRUE)
for (i in 1:nrow(prob)) prob[i, menu[i, ]==1] <- logitAtte(sum(menu[i, ]), 2)$choiceProb

# List of preferences to be tested
pref_list <- matrix(c(1, 2, 3, 4, 5, 6,
                      2, 3, 4, 5, 6, 1), ncol=6, byrow=TRUE)
# RAM only
result1 <- revealPrefModel(menu = menu, prob = prob, pref_list = pref_list, RAM = TRUE, AOM = FALSE)
summary(result1)

# AOM only
result2 <- revealPrefModel(menu = menu, prob = prob, pref_list = pref_list, RAM = FALSE, AOM = TRUE)
summary(result2)

# Both RAM and AOM
result3 <- revealPrefModel(menu = menu, prob = prob, pref_list = pref_list, RAM = TRUE, AOM = TRUE)
summary(result3)

Generate Summary Statistics

Description

sumData generates summary statistics. Given a collection of choice problems and corresponding choices, sumData calculates the number of occurrences of each choice problem, as well as the empirical choice probabilities.

This function is embedded in revealPref.

Usage

sumData(menu, choice)

Arguments

Value

Author(s)

Matias D. Cattaneo (maintainer), Princeton University. matias.d.cattaneo@gmail.com.

Paul Cheung, University of Maryland. hycheung@umd.edu

Xinwei Ma, University of California San Diego. x1ma@ucsd.edu

Yusufcan Masatlioglu, University of Maryland. yusufcan@umd.edu

Elchin Suleymanov, Purdue University. esuleyma@purdue.edu

References

M. D. Cattaneo, X. Ma, Y. Masatlioglu, and E. Suleymanov (2020). A Random Attention Model. Journal of Political Economy 128(7): 2796–2836. doi:10.1086/706861

M. D. Cattaneo, P. H. Y. Cheung, X. Ma, and Y. Masatlioglu (2026). Attention Overload. Working paper.

Examples

# Load data
data(ramdata)

# Generate summary statistics
summaryStats <- sumData(ramdata$menu, ramdata$choice)
nrow(summaryStats$sumMenu)
min(summaryStats$sumN)

summaryStats$sumMenu[1, ]
summaryStats$sumProb[1, ]
summaryStats$sumN[1]

Internal function.

Description

Internal function.

Usage

## S3 method for class 'ramchoiceRevealAtte'
summary(object, ...)

Arguments

Internal function.

Description

Internal function.

Usage

## S3 method for class 'ramchoiceRevealPref'
summary(object, ...)

Arguments

Internal function.

Description

Internal function.

Usage

## S3 method for class 'ramchoiceRevealPrefModel'
summary(object, ...)

Arguments