Assessio continuously updates its norm groups in order to reflect the current state of normal, work-related behavior and be able to make predictions based on this information. As of January 2024, a new global norm group has been implemented on the Assessio personality assessment, MAP. All new recruitment processes started after January will use this new norm. Scores that were already in the Assessio platform remain unchanged.
Overview
The MAP norm group consists of 20,348 people. It replaces the 2019 norm and is based on data collected from 2019 - 2023. Data collection took place in high-stake setting, meaning Selection (73%) and Development (27%). People in the norm group are aged 18-70 (Mean age is 34) and 49.8% identified as female, 49.8% as male and 0.4% other. The group contains 162 nationalities with none of these nationalities exceeding 2.54% of the group. Education level ranges from Elementary school (2.5%) to PhD (2.5%), with the largest group having completed 3 or more years of post-secondary education (56.7%). More than 25 job families are included, such as Business and financial operations (9.4%), Office and administrative support (7.0%), Sales and related (6.9%), Management (5.5%), Computer and mathematical (5.5%), Transportation and material moving (5.0%), Production (4.8%) and Architecture and Engineering (4.4%).
Criteria
The norm group was updated based on quality standards derived from various international standards, including EFPA, COTAN, and ITC guidelines. These guidelines set out criteria for various aspects of the norm group, such as time since the last update, size of the norm group, and composition. A discussion of these can be found below.
Update
Over time, what is considered normal behavior changes. Major events and crises have an impact on the way people in general behave and new generations may also challenge the existing standards. Therefore, with respect to assessments, it is highly important to update norm groups at a regular basis and make sure that all candidates and people assessed are evaluated with a norm group representing the current state and what is currently considered normal behavior, since that will provide the most valid assessment. In addition, updating the norm group keeps scores balanced and avoid too many candidates getting either high or low scores. In other words, norm updates allow for better differentiation of candidates, which in turn leads to better recruitment decisions. According to EFPA and COTAN guidelines, a norm of the highest quality should not be older than 10 or 15 years, respectively.
At Assessio we are committed to checking if updates are needed at least every 2 years and update our norm groups frequently. The current update is performed 5 years after the previous update (2019).
Sample size
A good norm group consists of many people, as a high number provides greater representation and statistical certainty. The prevailing view is that the larger the sample, the better the norm group. While that is true, it very much depends on sampling procedures as well as composition with respect to different demographic characteristics. In general, norm groups that are too small run the risk of underrepresentation (e.g., too few people with a certain occupation or education level), whereas too large norm groups risk overrepresentation (e.g., too many people of a certain age or nationality). According to EFPA, a sample size of at least 1,000 constitutes an excellent norm group (in some cases, smaller norm groups may also be sufficient depending on composition, target groups, and intended applications).
The MAP norm group consists of 20,348 people who were selected through stratified randomization from a total of 227,779 people aged 18-70 who completed the assessment in a high-stake setting. Statistical analyses confirmed that the norm group does not represent a biased sample, as score differences between different samples were only small or negligible across scales (Cohen’s d ranging from 0.00-0.42 with an average of 0.15).
Composition
To ensure that a norm group is representative of all target groups and is appropriate for all intended applications, key demographic characteristics must be carefully weighted and balanced, especially those that can lead to potential score differences between subgroups.
To construct a proper global norm, the sample was stratified for gender at the nationality level, hence making each nationality contributing an equal number of main genders (M and F). Then, nationalities were stratified such that each nationality constituted a maximum of 2.5 % of the total norm group. Next, other genders were added to the norm group as well with aim of having a representation of roughly 1 % but with the restriction that this addition did not cause any nationality to be largely overrepresented. Although the final age distribution was slightly skewed to the left (with a median of 32), statistical analyses revealed no significant relationships between age and any of the scores as evidenced by very low correlations ranging from -.16 to .09, with absolute values averaging just .08. Therefore, it was deemed unnecessary to stratify for age, as this would only reduce the sample size without impacting overall scores across age groups. As the final sample comprised a proper range of education levels and occupations (job families), and there were no major score differences, the sample was not further stratified for any of these demographic variables.