Hiring interviews are critical because they allow companies to increase the diversity of their workforce. However, managers may not be aware of subtle forms of bias that can shape interviews or how to structure the interview process to level the playing field for all candidates.

Once managers are more aware of what biases to look for, they can improve their interview process. This will not only help interrupt bias, but it will make managers better at selecting the best candidates.

Here are the five most common patterns of bias and how they play out in interviews:

1. Tightrope bias. The most common pattern emerges because, while white men need to be authoritative and ambitious to succeed, members of other groups often need to find ways of displaying authority and ambition that strike white men as “appropriate.” Women are expected to be modest and self-effacing, complying with prescriptive gender stereotypes that have increased in strength in recent decades. For members of a group seen as lower in status, notably people of color and women, being deferential may be the price of being seen as reasonable — and that women and people of color are more likely than white men to be expected to be worker bees, not leaders brimming with ambition.

2. Prove-it-again bias. The second pattern of bias that often plays out in interviews is called prove-it-again bias, which reflects that some groups must prove themselves more than others. White men are often judged on potential while other groups must have demonstrated competence already, which, of course, creates an invisible escalator for white men. A related prove-it-again pattern is in-group favoritism: like likes like. This means that, if most of the hiring is being done by white men, they typically will give the benefit of the doubt to other white men, but not to other groups.

3. Maternal wall bias. This pattern describes the negative competence and commitment assumptions often made about mothers, especially if they have a “gap in their resume.”

4. Cultural stereotypes. This pattern concerns racial stereotypes. One example is the assumption that Asian Americans will be well suited to technical roles, but not to leadership. Members of the Latinx community may be seen as “too emotional” regarding behavior that, in a white man, would be seen as admirable passion for the business. African Americans may be seen as “intimidating” or angry when they behave assertively rather than deferentially.

5. Tug-of-war bias. The final pattern is when bias against a group fuels conflict within the group. Research shows that women and people of color who advocate for members of their own group may well be seen as playing favorites. The same does not apply to white men (favoritism threat). Women and people of color also may hold other members of their own group to higher standards, because “that’s what it takes to succeed here” or “I want to make sure they don’t make me look bad” (comparison threat).

In addition to being aware of these patterns, managers can level the playing field by making interviews more structured. To interrupt tightrope bias, they can keep track of who is faulted for personality issues and look for demographic patterns. If they find that only women or people of color are faulted, then they should ask themselves whether a broader range of behavior is accepted from white men than from other groups.

The most effective way to interrupt prove-it-again bias is structure. Before the interview, managers should write down the most important things they are looking for and create a form in which they evaluate each competency, along with pieces of evidence they used to reach their rating for that competency. This, of course, means that they will need an interview protocol that asks the questions needed to elicit that evidence, asking each candidate the same questions (ideally in the same order, the research shows). The form should have enough evidence that someone else could pick it up and understand the ratings or rankings of each candidate.

The maternal wall and the tug-of-war patterns are easy to interrupt. Candidates should be allowed to explain gaps in their resume. This is more effective than pretending to ignore them. Moreover, gaps related to child rearing should not be counted against the candidate. It’s also important to reiterate that it’s illegal to discriminate against a candidate due to pregnancy or for having children.

For the tug-of-war, it should be enough to describe the favoritism and comparison threats. Doing so will help ensure that majority groups do not put pressure on minority groups that lead to those dynamics.

In addition, managers should provide tips and questions to all candidates that prepare them for the interview. Experimental evidence suggests that this is effective in leveling the playing field.