When looking for employment, people are often required to take a number of tests, whose results are used by employers to select the most suitable candidate. However, applicants do not always perform as well as they could during the selection process, which may lead to potential «<false negative» considerations. That is why professional associations have been encouraging policymakers to introduce adequate retesting policies. Previous research has demonstrated that when a firm gives unsuccessful candidates the opportunity to re-take previously failed tests, candidates are likely to perceive it as a fair organisation and may even recommend it to others (Arvey & Sackett, 1993; Gilliland, 1993).
As a result of the growing popularity of applicant retesting, researchers have been trying to identify the dynamics and consequences of these practices.
Past studies have revealed that when an applicant is allowed to retake a test (especially a cognitive ability test), their score is highly likely to improve; specifically (Hausknecht, Trevor, & Farr, 2002; Reeve & Lam, 2005, 2007). Therefore, available data suggests that thanks to applicant retesting, initially unsuccessful candidates could actually perform well enough to be selected.
In their study, Schleicher et al. (2010, p. 604) attempt to determine whether age, gender and race affect retest results. In order to collect a sufficient amount of data, they focus on eight different selection tests (some of which had never been studies before), namely three written tests, three types of interviews and two assessment centre exercises.
Hypotheses: The authors hypothesise that 1) white applicants will benefit from resting opportunities more than black and hispanic ones, 2) white applicants will obtain higher scores when retaking written tests than performance tests, and 3) younger applicants will benefit from resting opportunities more than older ones.
Sampling: The authors identified an agency of the U.S. Government which attracts thousands of job applicants every year. The agency requires applicants to pass a number of tests, which tend to change every year. Schleicher et al. (2010) identified numerous job applicants who had failed the agency’s written tests (which represent the first stage of the selection process) and had retaken them the following year. A total of 7031 applicants took part in Schleicher et al.’s (2010) study.
Even though some participants retook some of the tests twice, the authors chose to investigate score differences between the initial test and the first retest. In order to compare changes in test scores, applicants were classified by race, age and gender. To be more precise, the authors focussed on the following demographic characteristics: White, Black, Hispanic, Asian (Race-related categories) Male, Female (Gender-related categories Under 40, 40 and Over (Age-related categories)
After ensuring that both administration periods were characterised by equally difficult tests, the authors collected applicants’ composite scores (related to their written and performance tests) and examined their frequency distributions to see how many of them had obtained different scores when re-taking the tests. Correlation analysis was used to compare initial and retest scores and multiple regression analyses were conducted to test the hypotheses.
Results indicate that participants performed better when they retook the tests. With regards to the first hypothesis, regression analysis revealed that Whites benefit from larger score gains than Blacks and Hispanics when retaking written, job knowledge and biodata tests. As for the performance tests, the authors found that Whites performed better than Hispanics when they retook the case analysis test. As for the three interviews, Blacks exhibited larger gains than Whites and Hispanics.
On the other hand, Asians performed better than Whites when they were allowed to repeat the experience and interests interview. Overall, all three hypotheses turned out to be correct. Moreover, the authors found that while gender does not have a statistically significant impact on test result, women tend to perform better on performance tests.
Results suggest that retesting gains depend greatly on one’s race and age. However, they also indicate that different test types are associated with different retest effects. To be more precise, it appears that certain tests such as the biodata test and the leaderless group discussion – are linked with larger retesting gains than other tests – such as verbal ability. In view of these considerations, the authors asked themselves why candidates perform better only when retaking certain tests and why each demographic group seems to have a retesting advantage on a limited number of test types.
For instance, Schleicher et al.’s (2010) study revealed that when Whites are allowed to retake written tests, they tend to perform better during the second administration. The authors attempted to justify these unexpected patterns by identifying the reasons why retest effects vary across different test types and demographic groups. They found that different retest effects may be due to the following features: G-loaded: since g-loaded tests examine applicants’ cognitive abilities, which tend to remain stable over time, score changes should be nearly unnoticeable. Novel: when taking the first test, applicants usually acquire more knowledge which can help them obtain a higher score on more novel tests. Fakeable: fakeable tests are usually associated with larger retesting gains as score gains could result from response distortion.
Identical vs parallel tests: when applicants take the same test twice, it is highly likely that they will obtain a higher score on the second test. When they take parallel tests, the retest will probably reveal moderate gains. Sign-oriented vs sample-oriented: it is possible that Whites’ large retesting gains may depend on whether applicants are asked to retake a sign or sample-oriented test. Sample-oriented tests are similar to job performance tests and, apparently, whites tend to score less on these tests.
Numerous scholars have explored the psychological effects of selection practices on candidates (Cable, 2013, p. 42). As Cable (2013, p. 42) pointed out, selection procedures and decisions can have a profound impact on candidates’ wellbeing, self- confidence and self-esteem; that being said, research has mainly focussed on candidates experiencing standardised assessments for the first time. Since data have revealed that a significant number of unsuccessful candidates tend to repeat the testing process, more attention is being paid to the dynamics and implications of retesting (Cable, 2013, p. 42).
For example, a study by Langevin & Hausknecht (2009) showed that organisations that employ permissive retesting practices are likely to trigger positive reactions among job applicants. While candidates’ reactions may appear to be unrelated to score results, if researchers succeeded in demonstrating that there is a correlation between these two variables, organisations would have to rethink their selection procedures.
By exploring the impact of age, race and gender on retest scores, Schleicher et al. (2010) approached a highly controversial issue from a practical perspective. Should employers allow unsuccessful candidates to retest? If a candidate performs well during the first selection process, how should they react to the idea that unsuccessful candidates may be given a second or even a third chance? Would they perceive retesting as a fair practice, or would they lose confidence in organisations’ selection procedures?
These are perfectly reasonable questions that anyone who has ever had to complete a test for an employment agency or a direct employer may rightfully ask. The data collected by Schleicher et al. (2010) clearly suggest that some demographic categories are favoured when retaking certain types of tests, which means that retesting can minimise the risk of adverse impact against certain groups. Similarly to Schleicher et al. (2010), Van Iddekinge et al. (2011) tried to determine whether age and gender affect retest scores among internal candidates looking to be promoted.
Their findings revealed that women and younger applicants tend to perform better upon retesting than men and older candidates; this raises questions about the fairness of retesting, as it is obvious that while retesting opportunities benefit certain subgroups, they have an adverse impact on other subgroups.