Chi_Square Test Using NCES Public School Data (2022–23)

[module2]

code

```Detected columns: locale_code: ULOCALE locale_text: None level: GSLO charter-like columns: ['CHARTER_TEXT'] virtual-like columns: ['VIRTUAL']

Sample values in ULOCALE: ULOCALE 21-Suburb: Large 26614 11-City: Large 15228 41-Rural: Fringe 11979 42-Rural: Distant 10103 13-City: Small 6538 43-Rural: Remote 6332 32-Town: Distant 6199 12-City: Mid-size 6053 33-Town: Remote 4104 22-Suburb: Mid-size 3305 31-Town: Fringe 2940 23-Suburb: Small 1995 Name: count, dtype: int64

Sample values in GSLO: GSLO PK 32857 KG 22804 09 16905 06 13579 07 5060 05 2558 03 1681 M 1227 04 1145 01 958 08 706 02 569 Name: count, dtype: int64

Sample values in CHARTER_TEXT: CHARTER_TEXT No 89527 Yes 8155 N 3708 Name: count, dtype: int64

Sample values in VIRTUAL: VIRTUAL Not Virtual 67717 Supplemental Virtual 17192 Missing 13886 Full Virtual 1323 Not Reported 819 Virtual with face to face options 453 Name: count, dtype: int64

Selected charter column: CHARTER_TEXT CHARTER_TEXT 0 93235 1 8155 Name: count, dtype: int64

Fallback virtual column: VIRTUAL VIRTUAL 1.0 86685 NaN 14705 Name: count, dtype: int64

Mapped place_type counts: place_type NaN 101390 Name: count, dtype: int64

Mapped urban_rural counts: urban_rural NaN 101390 Name: count, dtype: int64

Mapped level_std counts: level_std Elementary 62572 Middle 19345 High 17802 Other 1671 Name: count, dtype: int64

[SKIP] Charter (Yes/No) ~ Place Type (City/Suburb/Town/Rural): not enough data (n=0, response levels=0, exp levels=0).

[SKIP] Virtual (Yes/No) ~ Place Type (City/Suburb/Town/Rural): not enough data (n=0, response levels=0, exp levels=0).

[SKIP] Charter (Yes/No) ~ Urban vs Rural: not enough data (n=0, response levels=0, exp levels=0).

[SKIP] Virtual (Yes/No) ~ Urban vs Rural: not enough data (n=0, response levels=0, exp levels=0).

[SKIP] Virtual (Yes/No) ~ School Level: not enough data (n=86685, response levels=1, exp levels=4).

=== Charter (Yes/No) ~ School Level === charter_bin 0 1 level_std Elementary 57270 5302 High 16385 1417 Middle 18049 1296 Other 1531 140

Chi2=63.31, df=3, p=1.1522e-13, Cramér’s V=0.025

Post‑hoc pairwise (Bonferroni-adjusted): Pair Test p_raw p_Bonf Chi2 YesRate_first(%) YesRate_second(%) Counts Elementary vs Middle Chi2 2.589352e-15 1.553611e-14 62.556303 8.5 6.7 [[57270, 5302], [18049, 1296]] High vs Middle Chi2 3.416050e-06 2.049630e-05 21.567424 8.0 6.7 [[16385, 1417], [18049, 1296]] Middle vs Other Chi2 1.049669e-02 6.298011e-02 6.548598 6.7 8.4 [[18049, 1296], [1531, 140]] Elementary vs High Chi2 3.005128e-02 1.803077e-01 4.706357 8.5 8.0 [[57270, 5302], [16385, 1417]] High vs Other Chi2 5.783371e-01 1.000000e+00 0.308931 8.0 8.4 [[16385, 1417], [1531, 140]] Elementary vs Other Chi2 9.255453e-01 1.000000e+00 0.008733 8.5 8.4 [[57270, 5302], [1531, 140]]```

[module2]

[module2]

[module2]

What the code did?

Loaded the NCES dataset for public schools (2022–23) and applied NCES data notes (numeric special codes like −1/−2/−9 treated as missing, text values left intact). Detected the right columns (e.g., ULOCALE for place type and any CHARTER_* fields for charter status), then standardized values into clear categories (City/Suburb/Town/Rural; Yes/No for charter). Built a contingency table (rows = place types; columns = Not Charter / Charter) and ran a Chi_Square Test of Independence to see if charter status and place type are associated. Calculated effect size (Cramér’s V, 0–1) to summarize how strong the association is. If there were more than two place types and the overall test was significant, it ran post‑hoc pairwise tests (with Bonferroni correction) to show which specific pairs of place types differ.

[module2]

Variables

Variables (what we analyzed) We examined whether being a charter school (Yes/No) is distributed differently across place types—City, Suburb, Town, Rural—using NCES’s urban‑centric locale codes (City = 11–13, Suburb = 21–23, Town = 31–33, Rural = 41–43). In plain words: Do charter schools show up more often in some kinds of communities than others? If the 4‑group test couldn’t run (e.g., due to missing values), the code automatically simplified the place types into Urban vs Rural and tried that instead. Sources: NCES dataset description; NCES locale definitions. Public School Characteristics 2022–23 (NCES EDGE) • NCES Locale Classifications (codes 11–43)

[module2]

Summary

This project examines whether charter school status (Yes/No) varies across different community types using the NCES Public School Characteristics 2022–23 dataset. Schools were classified into City, Suburb, Town, and Rural using NCES urban‑centric locale codes, which describe the population density and geographic context of school locations. A Chi‑Square Test of Independence was used to determine whether charter schools are evenly distributed across these place types or if some community types have higher or lower charter prevalence. When the overall test showed an association, we conducted post‑hoc pairwise comparisons (with Bonferroni correction) to identify which specific place types differ. Graphs visualizing percentage charter by place type and charter vs non‑charter counts help illustrate how the distribution changes across community categories. Overall, the project provides statistical evidence about whether place matters in the presence of charter schools in the U.S. public education landscape. Sources: NCES Public School Characteristics 2022–23 dataset; NCES Locale Classifications.