To obtain a critical value function for the likelihood ratio statistic, the conditional technique is used. When identification is high, it performs similarly to the unconditional likelihood ratio test and appears to outperform the test suggested by Anderson and Rubin (1949) and a specific score test.
The likelihood-ratio test in statistics compares the quality of fit of two competing statistical models based on the ratio of their likelihoods, especially one obtained by maximizing over the whole parameter space and another obtained after imposing some restriction.