Direct answer
Is the chi square distribution symmetric? No. A chi-square random variable takes only nonnegative values, so its distribution is right-skewed and cannot be symmetric.
Why a chi-square distribution cannot be symmetric
A distribution is symmetric about a center \(c\) when values equally far from \(c\) have the same probability density: for all \(a\), the density at \(c+a\) matches the density at \(c-a\).
A chi-square random variable \(X \sim \chi^2_\nu\) has support \([0,\infty)\). That means \(P(X<0)=0\), and its density is zero for negative values.
If symmetry about some \(c\) were possible, then whenever \(c+a\) is in the support, \(c-a\) would also need to be in the support. Choose any \(a>c\). Then \(c-a<0\), which lies outside \([0,\infty)\), so the density at \(c-a\) is \(0\) while the density at \(c+a\) is positive for many \(a\). This contradiction shows that a chi-square distribution is not symmetric for any degrees of freedom \(\nu\).
How the shape changes with degrees of freedom
A chi-square distribution with \(\nu\) degrees of freedom can be defined as a sum of squared standard normals:
\[ X=\sum_{i=1}^{\nu} Z_i^2,\quad \text{where each } Z_i \sim N(0,1)\text{ and the }Z_i\text{ are independent.} \]
For small \(\nu\), most probability mass sits near \(0\) with a long right tail (strong right skew). As \(\nu\) increases, the distribution becomes more mound-shaped and less skewed, but it still remains on \([0,\infty)\), so it is not truly symmetric.
| Quantity | Expression for \(X \sim \chi^2_\nu\) | Interpretation for shape |
|---|---|---|
| Mean | \(\mathbb{E}(X)=\nu\) | Center increases linearly with \(\nu\) |
| Variance | \(\mathrm{Var}(X)=2\nu\) | Spread increases, but relative spread decreases as \(\nu\) grows |
| Mode (for \(\nu \ge 2\)) | \(\nu-2\) | Peak shifts right as \(\nu\) increases |
| Skewness | \(\sqrt{\dfrac{8}{\nu}}\) | Skewness decreases toward \(0\) as \(\nu\) increases |
Numerical intuition from skewness
The skewness formula \(\sqrt{8/\nu}\) quantifies how quickly the chi-square distribution becomes less skewed:
| \(\nu\) | Skewness \(\sqrt{8/\nu}\) | Qualitative shape |
|---|---|---|
| 1 | \(\sqrt{8}\approx 2.828\) | Very right-skewed |
| 4 | \(\sqrt{2}\approx 1.414\) | Strong right skew |
| 10 | \(\sqrt{0.8}\approx 0.894\) | Moderate right skew |
| 30 | \(\sqrt{8/30}\approx 0.516\) | Mild skew, near bell-shaped |
Visualization: chi-square curves for different degrees of freedom
Normal approximation (why it can look “almost symmetric” for large \(\nu\))
Since \(X=\sum_{i=1}^{\nu} Z_i^2\), consider the centered sum \[ X-\nu=\sum_{i=1}^{\nu}(Z_i^2-1). \]
Each term has mean \(0\) and variance \(2\). For large \(\nu\), the Central Limit Theorem gives the approximation \[ \frac{X-\nu}{\sqrt{2\nu}} \approx N(0,1), \] so \(X\) is approximately normal with mean \(\nu\) and variance \(2\nu\). This explains why large-\(\nu\) chi-square curves can appear nearly symmetric, even though exact symmetry is not possible.
Common implication in inference
Treating a chi-square distribution as symmetric is a frequent mistake. In chi-square goodness-of-fit and independence tests, p-values commonly come from the right tail because large chi-square values indicate greater disagreement between observed and expected counts.