When in doubt about your assumption that the population variances are equal when comparing two population means, the best strategy is to proceed with the unequal variances test?