Joins a TGES per-pupil spend metric to an outcome percentile and labels each district's spend-vs-outcome position. This is the comparative fiscal analyst's headline product, and the literal answer to the MarGrady brief's open "cost-effectiveness" question: outcomes rose, but at what cost, and was that efficient relative to peers?
Spend is ranked within the chosen peer group via
tges_percentile_rank(). The outcome percentile is supplied by the
caller (e.g. from grate_percentile_rank() or
parcc_percentile_rank()), so the outcome metric and its peer definition
stay the caller's choice. The efficiency residual is the vertical distance
from a per-year regression of outcome percentile on spend percentile: positive
means more outcome than the spending would predict.
Usage
tges_efficiency(
spend_df,
outcome_df,
spend_col = "Per Pupil costs",
outcome_percentile_col,
spend_peer = c("tges_group", "dfg", "county", "statewide"),
year_col = "end_year"
)Arguments
- spend_df
A tidied TGES table (one report year, ideally filtered to a single
calc_type) carryingdistrict_idandspend_col.- outcome_df
A district-level frame with
district_id, the year column, andoutcome_percentile_col(a 0-100 percentile).- spend_col
Character. Spend column in
spend_df. Default"Per Pupil costs".- outcome_percentile_col
Character. Percentile column in
outcome_df.- spend_peer
Character. Peer group for ranking spend. See
tges_percentile_rank(). Default"tges_group".- year_col
Character. Year column, present in both frames. Default
"end_year".
Value
A tibble: entity columns, end_year, the spend value,
spend_percentile, outcome_percentile, efficiency_residual,
and quadrant.
Examples
if (FALSE) { # \dontrun{
library(dplyr)
# Graduation outcomes ranked within DFG A, joined to per-pupil spend
grate <- fetch_grad_rate(2023, methodology = "4 year") %>%
add_dfg() %>%
filter(dfg == "A", is_district, subgroup == "total") %>%
grate_percentile_rank(peer_type = "dfg")
spend <- fetch_tges(2024)$CSG1 %>%
filter(calc_type == "Actuals", end_year == 2023)
tges_efficiency(
spend, grate,
outcome_percentile_col = "grad_rate_percentile",
spend_peer = "dfg"
) %>%
filter(district_id == "3570")
# The "watch" quadrant: high spend, low outcome
eff <- tges_efficiency(spend, grate,
outcome_percentile_col = "grad_rate_percentile")
eff %>% filter(grepl("watch", quadrant))
} # }