Model

Suppose that \(y_1, \ldots, y_n\) are observations on independent random variables \(Y_1, \ldots, Y_n\), each with probability density/mass function of the form \[ f_{Y_i}(y) = \exp\left\{\frac{y \theta_i - b(\theta_i) - c_1(y)}{\phi/m_i} - \frac{1}{2}a\left(-\frac{m_i}{\phi}\right) + c_2(y) \right\} \] for some sufficiently smooth functions \(b(.)\), \(c_1(.)\), \(a(.)\) and \(c_2(.)\), and fixed observation weights \(m_1, \ldots, m_n\). The expected value and the variance of \(Y_i\) are then \begin{align*} E(Y_i) & = \mu_i = b'(\theta_i) \\ Var(Y_i) & = \frac{\phi}{m_i}b''(\theta_i) = \frac{\phi}{m_i}V(\mu_i) \end{align*}

Hence, in this parameterization, \(\phi\) is a dispersion parameter.

A generalized linear model links the mean \(\mu_i\) to a linear predictor \(\eta_i\) as \[ g(\mu_i) = \eta_i = \sum_{t=1}^p \beta_t x_{it} \] where \(g(.)\) is a monotone, sufficiently smooth link function, taking values on \(\Re\), \(x_{it}\) is the \((i,t)th\) component of a model matrix \(X\), and \(\beta = (\beta_1, \ldots, \beta_p)^\top\).

Score functions and information matrix

The derivatives of the log-likelihood about \(\beta\) and \(\phi\) (score functions) are \begin{align*} s_\beta(\beta, \phi) & = \frac{1}{\phi}X^TWD^{-1}(y - \mu) \\ s_\phi(\beta, \phi) & = \frac{1}{2\phi^2}\sum_{i = 1}^n (q_i - \rho_i) \end{align*}

with \(y = (y_1, \ldots, y_n)^\top\), \(\mu = (\mu_1, \ldots, \mu_n)^\top\), \(W = {\rm diag}\left\{w_1, \ldots, w_n\right\}\) and \(D = {\rm diag}\left\{d_1, \ldots, d_n\right\}\), where \(w_i = m_i d_i^2/V(\mu_i)\) is the \(i\)th working weight, and \(d_i = d\mu_i/d\eta_i\). Furthermore, \(q_i = -2 m_i \{y_i\theta_i - b(\theta_i) - c_1(y_i)\}\) and \(\rho_i = m_i a'(-m_i/\phi)\) are the \(i\)th deviance residual (e.g. as implemented in the dev.resid component of a family object) and its expectation, respectively.

The expected information matrix about \(\beta\) and \(\phi\) is \[ i(\beta, \phi) = \left[ \begin{array}{cc} i_{\beta\beta}(\beta, \phi) & 0_p \\ 0_p^\top & i_{\phi\phi}(\beta, \phi) \end{array} \right] = \left[ \begin{array}{cc} \frac{1}{\phi} X^\top W X & 0_p \\ 0_p^\top & \frac{1}{2\phi^4}\sum_{i = 1}^n m_i^2 a''(-m_i/\phi) \end{array} \right]\,, \] where \(0_p\) is a \(p\)-vector of zeros.

Maximum likelihood estimation

The maximum likelihood estimator of \(\beta\) and \(\phi\) satisfies \(s_\beta(\hat\beta,\hat\phi) = 0_p\) and \(s_\phi(\hat\beta, \hat\phi) = 0\).

Bias-reducing adjusted score functions

Let \(A_\beta(\beta, \phi) = -i_\beta(\beta, \phi) b_\beta(\beta, \phi)\) and \(A_\phi(\beta, \phi) = -i_\phi(\beta, \phi) b_\phi(\beta, \phi)\), where \(b_\beta(\beta, \phi)\) and \(b_\phi(\beta, \phi)\) are the first terms in the expansion of the bias of the maximum likelihood estimator of the regression parameters \(\beta\) and dispersion \(\phi\), respectively. The results in Firth (1993) can be used to show that the solution of the adjusted score equations \begin{align*} s_\beta(\beta,\phi) + A_\beta(\beta, \phi) & = 0_p \\ s_\phi(\beta, \phi) + A_\phi(\beta, \phi) & = 0 \end{align*}

results in estimators \(\tilde\beta\) and \(\tilde\phi\) with bias of smaller asymptotic order than the maximum likelihood estimator.

The results in either Kosmidis and Firth (2009) or Cordeiro and McCullagh (1991) can then be used to re-express the adjustments in forms that are convenient for implementation. In particular, and after some algebra the bias-reducing adjustments for generalized linear models are \begin{align*} A_\beta(\beta, \phi) & = X^\top W \xi \,, \\ A_\phi(\beta, \phi) & = \frac{(p - 2)\phi\sum_{i = 1}^n m_i^2 a''(-m_i/\phi) + \sum_{i = 1}^n m_i^3 a'''(-m_i/\phi))}{2\phi^2\sum_{i = 1}^n m_i^2 a''(-m_i/\phi)} \end{align*}

where \(\xi = (\xi_1, \ldots, \xi_n)^T\) with \(\xi_i = h_id_i'/(2d_iw_i)\), \(d_i' = d^2\mu_i/d\eta_i^2\) and \(h_i\) is the “hat” value for the \(i\)th observation (see, e.g. ?hatvalues).

Input

• \(s_\beta(\beta, \phi)\), \(i_{\beta\beta}(\beta, \phi)\), \(A_\beta(\beta, \phi)\)
• \(s_\phi(\beta, \phi)\), \(i_{\phi\phi}(\beta, \phi)\), \(A_\phi(\beta, \phi)\)
• Starting values \(\beta^{(0)}\) and \(\phi^{(0)}\)
• \(\epsilon > 0\): tolerance for the \(L1\) norm of the direction before reporting convergence
• \(M\): maximum number of halving steps that can be taken

Output

• \(\tilde\beta\), \(\tilde\phi\)

Iteration

Initialize outer loop

1. \(k \leftarrow 0\)

2. \(\upsilon_\beta^{(0)} \leftarrow \left\{i_{\beta\beta}\left(\beta^{(0)}, \phi^{(0)}\right)\right\}^{-1} \left\{s_\beta\left(\beta^{(0)}, \phi^{(0)}\right) + A_\beta\left(\beta^{(0)}, \phi^{(0)}\right)\right\}\)

3. \(\upsilon_\phi^{(0)} \leftarrow \left\{i_{\phi\phi}\left(\beta^{(0)}, \phi^{(0)}\right)\right\}^{-1} \left\{s_\phi\left(\beta^{(0)}, \phi^{(0)}\right) + A_\phi\left(\beta^{(0)}, \phi^{(0)}\right)\right\}\)

Initialize inner loop

4. \(m \leftarrow 0\)

5. \(b^{(m)} \leftarrow \beta^{(k)}\)

6. \(f^{(m)} \leftarrow \phi^{(k)}\)

7. \(v_\beta^{(m)} \leftarrow \upsilon_\beta^{(k)}\)

8. \(v_\phi^{(m)} \leftarrow \upsilon_\phi^{(k)}\)

9. \(d \leftarrow \left|v_\beta^{(m)}\right|_1 + \left|v_\phi^{(m)}\right|\)

Update parameters

10. \(b^{(m + 1)} \leftarrow b^{(m)} + 2^{-m} v_\beta^{(m)}\)

11. \(f^{(m + 1)} \leftarrow f^{(m)} + 2^{-m} v_\phi^{(m)}\)

Update direction

12. \(v_\beta^{(m + 1)} \leftarrow \left\{i_{\beta\beta}\left(b^{(m + 1)}, f^{(m + 1)}\right)\right\}^{-1} \left\{s_\beta\left(b^{(m + 1)}, f^{(m + 1)}\right) + A_\beta\left(b^{(m + 1)}, f^{(m + 1)}\right)\right\}\)

13. \(v_\phi^{(m + 1)} \leftarrow \left\{i_{\phi\phi}\left(b^{(m + 1)}, f^{(m + 1)}\right)\right\}^{-1} \left\{s_\phi\left(b^{(m + 1)}, f^{(m + 1)}\right) + A_\phi\left(b^{(m + 1)}, f^{(m + 1)}\right)\right\}\)

Continue or break halving within inner loop

14. if \(m + 1 < M\) and \(\left|v_\beta^{(m + 1)}\right|_1 + \left|v_\phi^{(m + 1)}\right| > d\)

    14.1. \(m \leftarrow m + 1\)

    14.2. GO TO 10

15. else

    15.1. \(\beta^{(k + 1)} \leftarrow b^{(m + 1)}\)

    15.2. \(\phi^{(k + 1)} \leftarrow f^{(m + 1)}\)

    15.3. \(\upsilon_\beta^{(k + 1)} \leftarrow v_b^{(m + 1)}\)

    15.4. \(\upsilon_\phi^{(k + 1)} \leftarrow v_f^{(m + 1)}\)

Continue or break outer loop

16. if \(k + 1 < K\) and \(\left|\upsilon_\beta^{(k + 1)}\right|_1 + \left|\upsilon_\phi^{(k + 1)}\right| > \epsilon\)

    16.1 \(k \leftarrow k + 1\)

    16.2. GO TO 4

17. else

    17.1. \(\tilde\beta \leftarrow \beta^{(k + 1)}\)

    17.2. \(\tilde\phi \leftarrow \phi^{(k + 1)}\)

    17.3. STOP