rollmatch {rollmatch}

Rolling entry matching

Description

rollmatch is the last of 3 main functions in the rollmatch package <rollmatch> implements a comparison group selection methodology for interventions with rolling participant entry over time. A difficulty in evaluating rolling entry interventions is that a suitable "entry" date is not observed for non-participants. This method, called rolling entry matching, assigns potential comparison non-participants multiple counterfactual entry periods which allows for matching of participant and non-participants based on data immediately preceding each participant's specific entry period, rather than using data from a fixed pre-intervention period.

Usage

rollmatch(
  scored_data,
  data,
  treat,
  tm,
  entry,
  id,
  vars,
  lookback,
  alpha = 0,
  standard_deviation = "average",
  num_matches = 3,
  replacement = TRUE
)

Arguments

scored_data

Output from scored_data() or the output from reduce_data() with propensity scores labeled "score".

data

Original dataset before reduce_data() was ran.

treat

String for name of treatment variable in data.

tm

String for time period indicator variable name in data.

entry

String for name of time period in which the participant enrolled in the intervention (in the same units as the tm variable).

id

String for individual id variable name in data.

vars

Vector of column names used in the propensity score algorithm. This is used when creating the balance table.

lookback

The number of time periods to look back before the time period of enrollment (1-...).

alpha

Part of the pre-specified distance within which to allow matching. The caliper width is calculated as the alpha multiplied by the pooled standard deviation of the propensity scores or the logit of the propensity scores - depending on the value of match_on.

standard_deviation

String. 'average' for average pooled standard deviation, 'weighted' for weighted pooled standard deviation, and 'None' to not use a standard deviation multiplication. Default is "average".

num_matches

Number of comparison beneficiary matches to attempt to assign to each treatment beneficiary. Default is 3.

replacement

Assign comparison beneficiaries with replacement (TRUE) or without replacement (FALSE). If replacement is TRUE, then comparison beneficiaries will be allowed to be used with replacement within a single quarter, but will not be allowed to match to different treatment beneficiaries across multiple quarters. Default is TRUE.

Details

Rolling entry matching requires preliminary steps. This package will assist the user in steps 2 and 3. First, a quasi-panel dataset is constructed containing multiple observations of non-participants (one for each entry period). Participants enter the data once in the baseline period immediately preceding their unique entry into the intervention. Time-varying covariates (e.g., health conditions, spending, utilization) are dynamic for each entry period's non-participant observations. The user of rollmatch is expected to have already created this quasi-panel dataset. Second, the pool of potential comparisons for each participant is restricted to those that have the same "entry period" into the intervention (see function "reduce_data"). Finally, a predicted probability of treatment is obtained for participants and non-participants (e.g. through propensity score matching). The user can use function "score_data" to complete this step, or create use their own propensity score calculation.

The final step consists of the matching algorithm. The algorithm selects the best matched comparison(s) for each participant from the pool of non-participants with the same entry period. This is completed via the function "rollmatch".

Value

rollmatch returns an object of class "rollmatch".

An object of class "rollmatch" is a list containing the following components:

model

The output of the model used to estimate the distance measure.

scores

The propensity scores used in the matching algorithm.

data

The original dataset with all matches added.

summary

A basic summary table with counts of matched and unmatched data.

ids_not_matched

A vector of the treatment IDs that were not matched.

total_not_matched

The number of treatment IDs not matched.

matched_data

R data.frame of matches with scores, matching information, and the weights of the individuals

balance

table showing the full treatment, full control, matched treatment, and matched comparison group means and standard deviations for the variables used in the model.

Examples

data(package="rollmatch", "rem_synthdata_small")
reduced_data <- reduce_data(data = rem_synthdata_small, treat = "treat",
                            tm = "quarter", entry = "entry_q",
                            id = "indiv_id", lookback = 1)
fm <- as.formula(treat ~ qtr_pmt + yr_pmt + age)
vars <- all.vars(fm)
scored_data <- score_data(reduced_data = reduced_data,
                          model_type = "logistic", match_on = "logit",
                          fm = fm, treat = "treat",
                          tm = "quarter", entry = "entry_q", id = "indiv_id")
output <- rollmatch(scored_data, data=rem_synthdata_small, treat = "treat",
                    tm = "quarter", entry = "entry_q", id = "indiv_id",
                    vars = vars, lookback = 1, alpha = .2,
                    standard_deviation = "average", num_matches = 3,
                    replacement = TRUE)
output