F_ADF = (s - σ(γ)’)’W⁻¹(s − σ(γ))

Where:

- s = the strung-out vector of nonredundant elements of S (observed matrix)

- σ (γ) = the similar vector of their counterparts in the reproduced matrix Σ(γ),

- γ = the vector of all model parameters (that is variances, and covariances between independent variables, regression coefficients and factor loadings)

- W = a weight matrix representing a consistent estimate of the large sample covariance matrix of the elements of S (regarded as random variables).