<p>This paper extends the statistical inference for the alpha power inverse Weibull distribution within the framework of Type-I censored step-stress partially accelerated life tests. Both the model parameters and the acceleration factor are estimated within the maximum likelihood paradigm. Confidence intervals are constructed via asymptotic approximation and two bootstrap methods, namely percentile bootstrap and bootstrap-t. The simulation results demonstrate that the estimators are consistent. Furthermore, percentile bootstrap confidence intervals achieve coverage probabilities nearest the nominal level, outperforming those derived from asymptotic theory. The methodology is validated on two real datasets, where the proposed model demonstrates superior fit compared to standard lifetime distributions.</p>