Each study draws n values and records the sample mean x̄ and standard deviation s. For normal data the normal-sample theorem promises three things: x̄ and s are independent (left: a shapeless cloud), (n − 1)s²/σ² follows the chi-squared distribution with n − 1 degrees of freedom (top right), and T = (x̄ − μ)/(s/√n) follows Student's t with n − 1 degrees of freedom (bottom right). These are what make the t interval and the variance interval exact. Switch to skewed or uniform data and watch each promise fail.
Pink bars are the studies whose 95% interval missed: the t interval when |T| is beyond the dashed lines, the variance interval when (n − 1)s²/σ² falls outside its dashed lines.