Accessing Holistic Review Scores

What is holistic review?

A holistic review approach considers multiple dimensions such as personal attributes, experiences, skills, and academic performance to identify individuals whose unique strengths align with a program’s goals. This approach focuses on the whole applicant rather than on just one or a few factors, for example, test scores. For more information about understanding holistic review scores, models, and access, explore the Predictive and Holistic Insights Guide. 

What is the holistic score?

To assist in the holistic review process, WebAdMIT automatically calculates a holistic score for each applicant. This model was developed by Othot, Liaison's data science team, which has over a decade of experience in the field, in conjunction with the CAS association client. The model is:

• Research-based: developed through rigorous research, incorporating feedback from both the community and experts.
• Explainable: unlike some data science models that function as “black boxes,” where it’s challenging to interpret how they arrive at their decisions, this model’s algorithm is designed to be comprehensible.
• Customizable: flexible and can be tailored to align with your program’s mission, goals, and values.

How can I use the score?

You can choose to use the holistic score at your discretion. For example, you may choose to use the List Manager to find and sort applicants by score and prioritize reviewing those with higher scores. You may find that the holistic nature of the tool brings to your attention to applications that align with your mission, goals, and values, that might have otherwise gone unnoticed.

When using the score, note the following:

• This is a "surfacing tool." It helps bring applicants who may be a good fit for your program to the surface of your review.
  • This tool helps identify an initial set of applications to prioritize for review.
  • The List Manager is useful for finding specific subpopulations within the applicant pool and then reviewing the holistic score.
• You can customize the weighting of the scoring components to align with your program’s values.
  • The weights represent the relative priorities of different elements of the application to your program and how applicants compare to other applicants within your cohort.
  • Note that while you can adjust overall weightings, you cannot add or create custom scoring elements, adjust the weights for individual applicants, or adjust any final calculated scores.
• This is not meant to be used as a precision instrument.
  • You should look at the score in relation to the other scores; the absolute score is less relevant. For example, an 88.1 score is not meant to be far better than an 87.9 score.
• This is not meant to replace reviews.
  • This enhances your current reviewing process. Continue using Assignments or other review tools with your review team.
• This model will continue to evolve as we collect program feedback.

What are the key scoring concepts?

• The scoring model uses direct data to calculate scores.
  • Scores are derived from real data elements provided in the application or in official supplementary materials. The model does not attempt to infer things, such as the meaning of the word count of a personal statement.
• Each data element is calculated as a percentile score relative to other applicants to the program.
  • All data element scores are scaled from 0 to 100 (including binary variables).
• The scoring model organizes application data elements into domains and subdomains.
  • Domains are overarching categories, while subdomains provide further granularity within a domain.
• All scores and percentiles in the model are calculated within your program’s cohort of applicants, not against the general applicant pool.
  • Each applicant receives a unique score for each program, reflecting their relative standing among other applicants in that program.
  • Individual scores may shift over the course of the cycle as more applicants apply.

How are scores derived?

When you click a score, a breakdown appears in a four-column table.

1. On the left are the domains (categories of data elements), some of which can be expanded to show subdomains (such as Undergrad Background).
2. The Original Score column lists a set of scores assigned to the applicant, ranging from 0 to 100. These scores are derived using a percentile calculation method.
   • For each data element included in a domain and subdomain, WebAdMIT assesses how each applicant's performance compares to that of others applying to the same program. Once these percentiles are determined, they undergo a rescaling process, converting them from percentile values to a standardized 0 to 100 scale. This ensures a fair and consistent evaluation of applicants' standings within the applicant pool.
   • If a data element is missing, how it's handled depends on its importance. Missing values for high-priority elements are treated as zero. For non-critical elements, the weight is redistributed across the remaining domains and subdomains to avoid unfair penalization.
3. The Weight column displays the customized weightings assigned to each domain and subdomain.
4. The Weighted Score column displays the product of the original score and its weight.
5. The Total Weighted Score is the sum of the weighted scores for all domains. 

Removing Your Manually Distributed Holistic Review Weights

Click Edit, then click the red x next to a manually distributed weight. The applicable domain and/or subdomain percentages will automatically readjust to be evenly distributed.