Correlogram: Dietary Diversity, Iron Intake & Excess Food × NFHS-4 & NFHS-5 Health Indicators

Methodology Note

Each cell displays the pairwise correlation between two variables computed across n = 35 states/UTs of India. Two correlation methods are available via toggle:

Pearson r measures the linear association between two variables. For variables X and Y with means x̄ and ȳ:   r = Σ(X_i−x̄)(Y_i−ȳ) / [Σ(X_i−x̄)² · Σ(Y_i−ȳ)²]^½. Sensitive to outliers and assumes a linear relationship.

Spearman ρ replaces each value with its rank (1 = smallest) and computes Pearson r on the ranks. Equivalently, ρ = 1 − 6Σd_i² / n(n²−1), where d_i is the difference in ranks for state i. Captures monotonic (not necessarily linear) relationships and is robust to outliers.

95% Confidence Intervals are computed via Fisher z-transformation: z = arctanh(r),  SE = 1/√(n−3),  CI = tanh(z ± 1.96·SE). A correlation is marked significant (★ yellow border) when the 95% CI excludes zero.

Two datasets: NFHS-5 correlates HCES dietary variables with 2019-21 health outcomes; NFHS-4 uses 2015-16 outcomes.

Weighting: Unweighted treats all states equally (standard Pearson/Spearman). Pop. Weighted weights each state by its population, so larger states contribute more to the correlation. Confidence intervals under weighting use the Kish effective sample size.

Income Deciles: The HCES dietary variables (Shannon indices, excess food, iron intake, real income) are computed separately for each national income decile within each state. Selecting D1 (bottom 10%) through D10 (top 10%) shows how the dietary patterns of a specific income group across states correlate with state-level health outcomes. "All" uses the overall state average across all income groups. NFHS health outcomes are state-level aggregates and do not vary by decile.

Variable Descriptions:

HCES Dietary Variables: Real Income (MPCE) = real monthly per-capita consumption expenditure; Shannon (Gram) = Shannon diversity index computed on gram-weights of food groups; Shannon (Cereal Adj) = Shannon index adjusted for cereal dominance; Cereal Share = share of cereals in total food quantity; Shannon (Ratio Adj) = Shannon index based on ratio-adjusted food quantities; MAR = Mean Adequacy Ratio, measuring overall micronutrient adequacy of the diet.

Excess Food: Excess Cereal (g/day) = grams per day of cereal consumption above recommended levels; Excess Cereal (% pop) = proportion of the population with excess cereal consumption; Excess Fats (g/day) = grams per day of fats/oils consumption above recommended levels; Excess Fats (% pop) = proportion of the population with excess fats consumption.

Iron Intake: Shannon (Iron) = Shannon diversity index of iron sources across food groups; Iron Intake Adj (mg) = energy-adjusted total iron intake in milligrams; Iron Intake Unadj (mg) = unadjusted (raw) total iron intake in milligrams; Iron Non-Cereal Adj (mg) = energy-adjusted iron intake from non-cereal sources; Iron Non-Cereal Unadj (mg) = unadjusted iron intake from non-cereal sources.

NFHS Health Outcomes: Stunting, Wasting, Underweight = prevalence among children under 5; Anemia = prevalence among children (6–59 months), women (15–49), and men (15–49); Women/Men Thin = BMI below 18.5; Overweight = BMI ≥ 25; Hypertension = elevated blood pressure (Ch. 12, NFHS-5); High Glucose = elevated blood glucose (Ch. 12, NFHS-5).