"nearest" error

[Huang-Yilin]

Dear author,
I am writing to report a potential issue I encountered while using the MatchIt package (version 4.6.0) with the method = "nearest" option. Below are the details:
Description of the Issue:
When using the nearest method with ratio > 2, I noticed that each treated case is only matched to a single control instead of the expected ratio number of controls. And the last case will match all rest unmatched controls,like 1:2000. This occurs even when there are enough controls to meet the specified ratio.
R version: 4.2.2
MatchIt version: 4.6.0
Operating system: macOS（15.1.1）
I have included a minimal reproducible example below for your reference:

set.seed(123)

扩展到上千的病例和对照数据

n_cases <- 1000 # 病例数量
n_controls <- 5000 # 对照数量

生成模拟数据

test_data <- data.frame(
case_control = c(rep(1, n_cases), rep(0, n_controls)), # 病例与对照分组
age = c(rnorm(n_cases, 60, 5), rnorm(n_controls, 60, 5)), # 年龄分布
sex = c(
sample(0:1, n_cases, replace = TRUE), # 病例性别分布
sample(0:1, n_controls, replace = TRUE) # 对照性别分布
),
ID = 1:(n_cases + n_controls) # 唯一标识符
)

使用 MatchIt 进行匹配

library(MatchIt)

m.out <- matchit(case_control ~ age + sex,
data = test_data,
method = "nearest", # 使用最近邻匹配
ratio = 10, # 每个病例匹配5个对照
distance = "mahalanobis", # 使用 Mahalanobis 距离
caliper = c(age = 0.32), # 对年龄限制在 ±1 岁内
exact = "sex") # 确保性别完全一致匹配

matched_data <- match.data(m.out)

Could you please confirm if this is an intended behavior or a potential bug? If it is the latter, I would be happy to provide additional details or assist in testing any fixes.
I attempted to revert to an earlier version of MatchIt to resolve this problem, but unfortunately, the error persists.

This issue is impacting my current research, and as I am on a tight schedule to finalize an article for publication, I would greatly appreciate any guidance or a quick resolution.

Thank you for your time and for all your efforts in maintaining this fantastic tool for the community.

[ngreifer]

Thanks for reporting. This has been fixed in #209, and version 4.7.0 is now on CRAN.