plot.GeoLiftMarketSelection fails with "Two versions of estimated impacts do not correspond" for 2+ treated locations

[castroisabel]

Bug description

Calling plot() on a GeoLiftMarketSelection object fails with the internal error "Two versions of estimated impacts do not correspond. Serious error. Please contact package maintainers." The error originates in augsynth's get_placebo_gaps() during the permutation inference step (add_placebo_distribution).

This is fully reproducible using the package's own example dataset (GeoLift_PreTest), so it is not specific to user data. It occurs with multiple treated locations (the recommended markets here have 2–3 locations). The GeoLiftMarketSelection() call itself completes successfully and returns the recommendations table; only the subsequent plot() fails.

Session information

R version 4.6.0 (2026-04-24)
Platform: aarch64-apple-darwin23
Running under: macOS Tahoe 26.4.1

Matrix products: default
BLAS: /Library/Frameworks/R.framework/Versions/4.6/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.6/Resources/lib/libRlapack.dylib; LAPACK version 3.12.1

locale:
[1] C.UTF-8/UTF-8/C.UTF-8/C/C.UTF-8/C.UTF-8

time zone: America/Sao_Paulo
tzcode source: internal

attached base packages:
[1] stats graphics grDevices utils datasets methods base

other attached packages:
[1] dplyr_1.2.1 tidyr_1.3.2 augsynth_0.2.0 GeoLift_2.7.5

loaded via a namespace (and not attached):
[1] tidyselect_1.2.1 IRdisplay_1.1 farver_2.1.2
[4] S7_0.2.2 fastmap_1.2.0 GGally_2.4.0
[7] fixest_0.14.1 digest_0.6.39 lifecycle_1.0.5
[10] dreamerr_1.5.0 magrittr_2.0.5 compiler_4.6.0
[13] rlang_1.2.0 progress_1.2.3 rngtools_1.5.2
[16] tools_4.6.0 utf8_1.2.6 knitr_1.51
[19] labeling_0.4.3 prettyunits_1.2.0 doRNG_1.8.6.3
[22] plyr_1.8.9 repr_1.1.7 RColorBrewer_1.1-3
[25] abind_1.4-8 pbdZMQ_0.3-14 withr_3.0.2
[28] purrr_1.2.2 numDeriv_2016.8-1.1 grid_4.6.0
[31] xts_0.14.2 future_1.70.0 ggplot2_4.0.3
[34] globals_0.19.1 scales_1.4.0 iterators_1.0.14
[37] MASS_7.3-65 cli_3.6.6 mvtnorm_1.3-7
[40] crayon_1.5.3 generics_0.1.4 future.apply_1.20.2
[43] reshape2_1.4.5 directlabels_2026.4.23 proxy_0.4-29
[46] stringr_1.6.0 assertthat_0.2.1 parallel_4.6.0
[49] bsts_0.9.11 base64enc_0.1-6 stringmagic_1.2.0
[52] vctrs_0.7.3 Matrix_1.7-5 sandwich_3.1-1
[55] jsonlite_2.0.0 hms_1.1.4 Formula_1.2-5
[58] listenv_0.10.1 foreach_1.5.2 glue_1.8.1
[61] parallelly_1.47.0 dtw_1.23-2 ggstats_0.13.0
[64] codetools_0.2-20 fect_2.4.1 gsynth_1.4.0
[67] stringi_1.8.7 gtable_0.3.6 quadprog_1.5-8
[70] osqp_1.0.0 Boom_0.9.16 tibble_3.3.1
[73] pillar_1.11.1 htmltools_0.5.9 IRkernel_1.3.2
[76] R6_2.6.1 BoomSpikeSlab_1.2.7 panelView_1.3.1
[79] MarketMatching_1.2.1 doParallel_1.0.17 CausalImpact_1.4.1
[82] evaluate_1.0.5 lattice_0.22-9 Rcpp_1.1.1-1.1
[85] uuid_1.2-2 gridExtra_2.3 nlme_3.1-169
[88] xfun_0.57 zoo_1.8-15 pkgconfig_2.0.3

Reproduction steps

library(GeoLift)
data(GeoLift_PreTest)

GeoTestData_PreTest <- GeoDataRead(
data = GeoLift_PreTest,
date_id = "date",
location_id = "location",
Y_id = "Y",
X = c(),
format = "yyyy-mm-dd",
summary = TRUE
)

MarketSelections <- GeoLiftMarketSelection(
data = GeoTestData_PreTest,
treatment_periods = c(10),
N = c(2, 3),
Y_id = "Y",
location_id = "location",
time_id = "time",
effect_size = seq(0, 0.5, 0.05),
lookback_window = 1,
cpic = 1.0,
alpha = 0.1,
Correlations = TRUE,
fixed_effects = TRUE,
side_of_test = "two_sided"
)

this line throws the error:
plot(MarketSelections, market_ID = 1, print_summary = FALSE)

Expected behavior

plot(MarketSelections, market_ID = 1) should render the plot of observed vs. synthetic outcome for the selected market, as documented, instead of throwing an internal error from the inference step.

Error in get_placebo_gaps(augsynth, att = FALSE): Two versions of estimated impacts do not correspond.  Serious error.  Please contact package maintainers.