By Kelly Ryan (MSc 2018/19)

As a hiring decision-maker you might believe that you would only progress a candidate’s CV based on merit, yet research shows that unconscious biases are likely to impact the objectivity of your selection decisions (Hunt, Prince, Dixon-Fyle, & Yee, 2018). Maybe you see yourself in their shoes; or they went to the same University as you, or you both have the same name? These and many others are reasons that we might subconsciously choose one candidate over another.

All decision-makers have their own biases, because humans understand the world from their own perspectives, be this related to their age, gender, title, socio-economic status or similar. Therefore, some applicants may receive preferential treatment in a recruitment process if the decision-maker can relate to them in some way. Likewise, other applicants may be treated less favorably because the decision-maker cannot relate to them, or in other words cannot accurately interpret their career trajectory. In spite of fact, there has been a lack of research conducted to date exploring the impact of what is known as the similar-to-me effect on CV screening.

The similar-to-me effect suggests that people are most likely to be attracted to those who are like themselves. This effect has been observed across a plethora of domains including perceptions of beauty, financing decisions of venture capitalists and during employment interviews. The similar-to-me effect is underpinned by theories such as the similarity-attraction paradigm. See Byrne (1971) for more detail.

The research project:

For my thesis, I conducted an experiment to assess the impact of the similar-to-me effect on hiring-decision maker’s CV screening. One-hundred and seventy-two participants predominately from Ireland, the United Kingdom, India and the United States took part in the study. All participants had recently engaged/or were currently engaging in some form of recruitment.

What participants did:

Firstly, participants were asked to fill in a CV template – their perception of a desirable candidate (i.e. someone that they would like to hire). All participants completed a CV template which included: desirable candidate’s name (an implicit measure for gender), nationality, highest level of education, university attended, degree-classification and personal interests or hobbies. After completing this CV template, participants were asked to fill in a similar CV template which included their own variables on the next page. In this CV template participants were explicitly asked for their gender rather than their name so as to ensure that all participation remained anonymous.  Participants were then asked to provide rationale for the decisions they made for their desirable candidate’s CV template.

Findings

Several analyses were then conducted. Firstly, I assessed whether the variables contained in the participant’s own CV would predict the variables they selected for their desirable candidate’s CV template. The results revealed that the vast majority of variables contained in the participant’s own CV significantly predicted the selection of these same variables for their desirable candidate’s CV. To take some examples from the data-set…

• Irish participants were 129 times more likely to select an Irish candidate than participants who identified as different nationalities were to select an Irish candidate.
• Participants with a doctorate degree were 66 times more likely to select a candidate with a doctorate degree than participants who did not have a doctorate degree.
• Male participants were five times more likely to select a male candidate than a female participant.
• Female participants were seven times more likely to select a female candidate than a male participant.

It is important to note however that not all of the variables contained in the participants’ own CVs predicted their selection of these same variables for their desirable candidate’s CV. For example, citing one’s highest level of education as a foundation degree or degree classification as third-class honors did not predict the selection of these same variables for one’s desirable candidate.

Nonetheless, almost half of the participants who engaged in the experiment replicated more than half of their own CV in the CV of their desirable candidate. These findings are important because most of the variables that were explored have little/no evidence linking them to performance. Interestingly however, when asked to explain the underpinning rationale as to why participants chose these particular variables for their desirable candidate’s CV template, even participants who created their desirable candidate’s CV in the image of their own CV oftentimes did not acknowledge doing so. This may point to the often-unconscious nature of biases where decision-makers may not be aware of their tendency to hire in “their own image” which makes this tendency even harder to deal with. Other participants however were explicit about seeking a candidate similar to themselves for rapport purposes and in order to validate this candidate’s credentials.

Implications

Overall, these findings present a number of implications for recruitment practices. Firstly, it would appear that hiring decision-makers subconsciously or otherwise favour candidates with a CV that resembles their own. Thus, recruiters could benefit from unconscious-bias training and awareness building around these issues. These findings also suggest, much like previous literature (Franke, Gruber, Harhoff, & Henkel, 2006) that applicants should not be overtly discouraged by rejection at the CV screening stage, as this decision may have had to less to do with their suitability for the role and more to do with the hiring-decision maker who reviewed their profile. 

One way of counteracting the similar to me effect may be to have all CVs reviewed by at least two hiring decision-makers to control for the impact of individual differences on hiring decisions (Frank & Hackman, 1975). This however, would be resource intensive and perhaps adopting a standardized scoring system for CV screening (Carbonaro & Schwarz, 2018) or blinding CVs by removing any information pertaining to applicants’ biographical characteristics (Rand & Wexley, 1975) would be more viable alternatives to counteracting the similar-to-me effect. 

Can you think of any other ways to limit the impact of such biases in selection processes?

Feel free to comment below!