Denigration of academic experts and expertise, amid a resurgence of political populism, poses a challenge to the legitimacy of academic research. Addressing this challenge requires us to continually demonstrate the importance of basing policy interventions on reliable evidence, rather than unevidenced assertions that gain traction through communication echo chambers. However, unconscious confirmation biases in collection and analysis of evidence can impair the reliability of our research insights. A key source of such confirmation biases are unchallenged ideologies and other taken-for-granted assumptions underlying any research (sub)field. This essay argues that informal and formal peer review processes at many stages of research need to highlight and challenge both conscious selectivity bias and unconscious confirmation bias. However, they are unlikely to do so where researchers only take on board feedback from peers in the same (sub)field who share ideological commitments and taken-for-granted assumptions. In such circumstances, self-referential peer review echo chambers can develop that entrench rather than challenge weaknesses in a research (sub)field. This can be a major risk to the effectiveness and reputation of any academic research (sub)field; a risk we need to confront.