All free tools
Free · no signup

Statistical symbols and abbreviations — a lookup table

What every statistical letter and Greek symbol stands for, the sample-vs-population pairs that cause the most confusion, and how APA style wants each one typeset.

Let the AI do this — sign up freeUpload your data, get the write-up back.
Sample statistics vs. population parameters

The core convention: Latin letters describe your SAMPLE; Greek letters describe the POPULATION they estimate.

Sample (Latin)Population (Greek)Quantity
M or x̄μ (mu)Mean
s or SDσ (sigma)Standard deviation
σ²Variance
rρ (rho)Correlation
p̂ or pπ (pi)Proportion
bβ (beta)Regression coefficient
n (subgroup) / N (total)NSample / population size
Test statistics and inference
SymbolNameMeaning / where you meet it
pp-valueProbability of a result at least this extreme if H0 is true. Italic, no leading zero: p = .03.
α (alpha)Significance levelThe false-positive rate you accept in advance, usually .05. (Same letter, different meaning: Cronbach’s α.)
β (beta)Type II error rateProbability of missing a true effect; power = 1 − β. (Also: standardized regression coefficient.)
tStudent’s tt-tests and regression coefficients; reported with df: t(24) = 2.31.
FF statisticANOVA and overall regression tests; two dfs: F(2, 57) = 4.40.
χ²Chi-squareCategorical association and goodness of fit: χ²(1, N = 120) = 5.24.
zStandard normal deviateLarge-sample and proportion tests; also standardized scores.
UMann–Whitney UNonparametric two-group comparison.
WWilcoxon / Shapiro–WilkSigned-rank statistic; also the normality test statistic.
dfDegrees of freedomIndependent information behind an estimate; determines the reference distribution.
CIConfidence intervalRange of parameter values compatible with the data: 95% CI [0.21, 1.03].
SE / SEMStandard error (of the mean)SD of a statistic’s sampling distribution: SE = SD/√n.
H0 / H1 (Ha)Null / alternative hypothesisThe "no effect" default and its rival.
λ (lambda)Wilks’ lambdaMANOVA test statistic (also: Poisson rate parameter).
ε (epsilon)Sphericity correction / errorGreenhouse–Geisser ε in repeated measures; residual error term in models.
Effect sizes and reliability
SymbolNameMeaning
dCohen’s dStandardized mean difference (benchmarks .2/.5/.8).
gHedges’ gSmall-sample-corrected d.
rPearson correlationLinear association, −1 to +1; r² = shared variance.
ρ / rsSpearman correlationRank-based (monotonic) association.
Coefficient of determinationVariance explained by a regression model; adjusted R² penalizes predictors.
η² / η²p(Partial) eta squaredANOVA variance explained; the partial version is not comparable across designs.
ω²Omega squaredLess-biased alternative to η².
fCohen’s fANOVA effect size used by power software (G*Power).
VCramér’s VEffect size for contingency tables, 0–1.
φ (phi)Phi coefficientCorrelation for 2 × 2 tables.
OR / RROdds ratio / risk ratioBinary-outcome effect sizes; OR > RR when the outcome is common.
κ (kappa)Cohen’s kappaChance-corrected inter-rater agreement.
αCronbach’s alphaInternal consistency of a scale (≥ .70 conventional floor).
ICCIntraclass correlationReliability of continuous ratings; also clustering in multilevel models.
Δ (delta)Change / differenceΔR² = increment in variance explained between nested models.

Typesetting rules (APA 7)

Italicize Latin-letter statistical symbols: N, n, M, SD, df, t, F, p, r, d, R². Do NOT italicize Greek letters (α, β, χ², η²) or ordinary abbreviations (CI, ANOVA, HR). Subscripts that are labels stay roman: in "M_control", the M is italic and "control" is not.

Numbers: statistics that cannot exceed 1 (p, r, η², α as reliability) drop the leading zero (p = .04, r = .31); those that can exceed 1 keep it (d = 0.50, t = 2.31). Report exact p-values to two or three decimals, with p < .001 as the floor.

Spacing: put spaces around = and < (t(24) = 2.31, p < .001), report df in parentheses immediately after the statistic letter, and give χ² its sample size: χ²(1, N = 120) = 5.24.

Frequently asked questions

What is the difference between M and μ (and s and σ)?
M (or x̄) is the mean of YOUR sample — a computed number. μ is the mean of the population your sample came from — an unknown you are estimating. The same logic pairs s with σ and r with ρ. Mixing them up in a methods section signals conceptual confusion, which is why reviewers pounce on it.
Why does α mean two different things?
Historical accident. In hypothesis testing α is the significance level (Type I error rate). In psychometrics α is Cronbach’s coefficient of internal consistency. Context disambiguates, but write "Cronbach’s α" on first use in a reliability context to be safe.
How do I type these symbols?
Word/Google Docs: Insert → Special characters, or shortcuts like Alt+0956 for μ on Windows. LaTeX: \mu, \sigma, \alpha, \chi^2, \eta^2_p. Unicode also works directly in most editors: μ σ α β χ² η² κ Δ — this page’s symbols can be copied straight from the tables.

Related free tools

Have a question the table can't answer?

The AI inside Kahubi works with your actual data and library — it runs analyses, checks assumptions, and writes the results up in your style.

Kahubi AI
Hi! Ask me anything you’d normally google mid-study — study design, statistics, literature, writing.

Free accounts include AI chat, literature search, and every tool on this site.

Stop copying numbers between tools

Inside Kahubi, the AI agent runs this analysis directly on your uploaded dataset — then writes the results section in your own writing style, with the statistics reported correctly.