Statistical and Analytical Terms

Standardized Beta (β)

A regression coefficient scaled so that all predictors are measured in standard deviation units. Allows comparison of relative predictor importance across factors measured on different scales.

Pearson Correlation (r)

A measure of linear association between two continuous variables, ranging from -1 to +1. This is used to identify factor overlap in the correlation matrix.

McLoone Index

Templeton et al. (2023) used this index to test whether below-median districts received equitable funding after HB 3. The index tracks how tightly the bottom half of districts cluster around the median funding level. A score of 1.0 means perfect clustering. The McLoone index fell from 0.876 to 0.858 after HB 3, meaning below-median districts lost ground.

Gini Coefficient

Borrowed from income inequality research, this index runs from 0 (every district gets the same funding) to 1 (one district gets everything). Templeton et al. (2023) applied it to per-pupil expenditure distributions in Texas and found the Gini rose from 0.124 to 0.146 under HB 3.

Concentration Index (CI)

Answers a specific question about whether funding flows toward higher-need or lower-need districts. A CI of 0 means funding tracks proportionally with need. The Texas value of 0.27 is positive, meaning funding concentrates among lower-need districts. Section 4E reports this finding.

Kakwani Index

Subtract the Gini coefficient of need (calculated from ARC scores) from the Concentration Index. A positive result means the funding distribution is more unequal than the need distribution. The Texas value of 0.09 confirms that funding skews toward districts with fewer at-risk students.

Spearman Rank Correlation (ρ)

Ranks districts by ARC score, then ranks them by per-pupil funding, and tests whether the two rankings align. The Texas result is ρ = −0.24 (p < 0.001, n = 1,201 districts). The negative sign means higher-need districts receive less funding per student.

Zero-Inflation

A data pattern where a variable contains an excess of zero values relative to what a standard distribution would predict. In the Texas dropout data, 780 of 2,087 secondary campuses (37.4%) report exactly 0.0% dropout. Among campuses with 200 or fewer students, 56.5% report zero. The pattern follows campus size, consistent with FERPA masking rather than genuine zero dropout.

FERPA Masking

Federal law (20 U.S.C. §1232g) requires TEA to suppress any student count below 10. A campus with 8 homeless students out of 50 total reports zero, losing its true 16% rate. A campus with 15 homeless students out of 500 keeps its 3% value. Pregnancy and parenting data is masked at 82.6% of campuses statewide. Foster care is masked at 40.9%. The ARC model treats all masked values as zero, producing lower-bound scores at small campuses.

Abbreviation Quick Reference