Non-stereotypic association and hiring decisions (2005) - Kawakami, Dovido & Van Kamp - Article

In the last few decennia there has been an increase in the representation of women in the workplace, but disparities in the status of women relative to men persist. Many scholars think that stereotypes have to do with this type of discrimination. Stereotypes are characteristics associated with a particular social group and discrimination is the biased treatment of others based on their group membership. Some scientists have shown that men may be chosen more often than women for supervisory roles in employment contexts, because power and leadership are qualities associated with men and gentility and submissiveness are associated with women. So one way of decreasing sex discrimination may be to reduce stereotypes.

Two basic processes are involved in stereotypes. Stereotype activation involves the increased cognitive accessibility of characteristics associated with a particular group and stereotype application represents the use of stereotypes when responding to a group member. One goal of this study is to examine sex discrimination in an experimental setting and to explore a strategy of reducing the activation of stereotypes. In the past stereotype activation has been described as a bad habit. Recently, extensive training in negating category-trait associations has been used to reduce automatic stereotype activation. Some scientists found that practice is essential to unlearning the stereotype associations. The present study wants to examine whether the training strategy that can reduce stereotype activation would also ameliorate the preference of men over women for managerial positions. Training seems to be a good strategy in reducing stereotype activation, but it’s not clear what the impact of training would be on behavioural responses to group members (discrimination). The training may be effective in reducing the activation of stereotypes, but can people control their evaluations? One research shows that reducing automatic activations may not necessarily reduce a more controlled form of bias. Implicit measures of stereotyping are related to spontaneous responses and are not correlated with deliberative responses because these responses may be different. This means that it’s important to examine the impact of strategies related to the reduction of biases on explicit and implicit processes separately.

Research on mental correction has shown that when people are motivated and able, they may modify their assessments in a direction opposite to the perceived bias. If the participant thinks that the training will bias his/her decision in favour of female job candidates, he/she may try to minimize this influence and be less favourable towards a female than towards a male candidate. So if people correct for the anticipated impact of the training, they may exhibit a level of sex discrimination which is similar to participants in the no-training condition. But, in order for correction to occur, people must first recognize that they have been influenced and then have the mental resources to control their response. If they are not aware of an unwanted influence and/or if they’re not able to adjust their responses because they lack the cognitive capacity, they will not modify their response. The writers of the current article examined whether correction processes could moderate the effects of training on subsequent hiring decision. They questioned participants in a pilot study about their awareness of the potential impact of training and their anticipated reactions. The participants indicated that they were aware that the training could systematically influence their responses and that they believed that the experimenter expected the training to reduce sex biases. According to correction theory, it is therefore possible that participants would modify their responses to adjust for that influence under controlled processing conditions. The writers therefore decided that in the ‘real’ study, the possible influence of correction processes related to non-stereotypic association training would be examined in two ways: disassociating the two tasks and adding a cognitive load manipulation. The writers expected that participants would correct for the effect of the perceived training when they received training and then directly performed the candidate selection task. They also expected correction processes to be reduced when the training was disassociated from the job application task by including a filler task. They thought that if the relationship between the non-stereotypic association training and the candidate selection task was less obvious, participants would be less likely to correct for the training. The writers thought that without the correction processes the non-stereotypic association training would be effective in reducing sex discrimination against women in the candidate selection task.

Mental resources are also needed for correction to occur. In previous studies, researchers showed that when not under cognitive load, participants corrected for the prime and formed less favourable impressions following the positive than the negative prime. In the current research researchers gave some of the participants a secondary task to impede their ability to correct, while they were reviewing the applicants during the job application phase. They thought that participants under cognitive load would be less able to correct. The goal of the present research was to investigate whether non-stereotypic association training would be effective in reducing sex discrimination and to explore the potential moderating role of correction processes that might limit the effectiveness. The writers thought that participants in the training condition would show levels of discrimination equivalent to that of the no-training control group when hiring decision directly followed non-stereotypic training. In this condition and the no-training condition, participants were expected to prefer male over female candidates. But, when the tasks were separated in time by an unrelated intermediate task or when participants were unable to correct because they were impeded by cognitive load, the training was expected to be effective in ameliorating sex discrimination. In these last two conditions, bias against female candidates was expected to be reduced in the latter two conditions and participants were expected to select women and men in the same extent.

The study

In this experiment, the main independent variable, Training, included four levels: no training, non-stereotypic association training, non-stereotypic association training plus filler task and on-stereotypic association training plus probe reaction task. Participants were seated behind a computer and they were presented with a series of photographs of men and women. The participants were told that underneath each photograph there would be two traits, one on the left and one on the right. They had to select the trait that was not culturally associated with the gender category in the photograph and to press the appropriate button on the button box. All participants were presented with the job application task. Participants in the no-training condition performed only the job application task. Participants in the non-stereotypic association training condition first completed the training task and then immediately after that the job application task. Participants in the non-stereotypic association training plus filler task condition, performed an intermediate task before proceeding to the job application task. Participants in the non-stereotypic association training plus probe reaction task condition received a probe task between the training and job application task. This probe task was designed to utilize some of the participants’ cognitive capacity and deplete resources for conscious control during the candidate selection phase.

The results indicated that with practice participants showed improvement in learning new, non-stereotypic associations. However, the selection of male or female candidates for a managerial position showed no difference between non-stereotypic association training and no-training conditions. The writers counterbalanced the sex of the candidate and the specific credentials and given the same documented qualifications, women were recommended for the position less often than men. This shows there is a prevalence of sex discrimination. In the non-stereotypic association training plus filler task condition and the non-stereotypic association training plus probe reaction task condition, participants showed less sex discrimination.

This research shows that researchers interested in interventions to stop/reduce bias should consider the effects of strategies aimed at changing stereotypes on different levels. Strategies aimed at changing the automatic activation of stereotypes will not necessary result in reduced discrimination. When the intervention is obvious, its effectiveness may be compromised. Therefore, the target task need not to come immediately after the first task. Also, when the conditions allow conscious control of responses and deliberation, the effectiveness of the training will not be so great. High cognitive load is needed to stop this conscious control and deliberation. Future research needs to examine ways to stop the stereotype habit under both automatic and controlled processing conditions.