Formative vs. Summative Assessment
Formative & Summative Evaluation
What are formative and summative assessments?
- Formative = improve as you go
- Summative = judge results at the end
- A difference in purpose, not rank
Once you tell formative and summative assessment apart, where a training manager sends the results changes. Summative scores go to HR and executives—they are the basis for deciding who completed the program and who earns certification. Formative results go to the instructor and the learner themselves. If half the class got the cash flow section of the financial statements wrong on the Day 3 quiz, that becomes the basis for revising the Day 4 lesson plan that same evening. Even the same question serves a different purpose depending on “who is looking at it, when, and why.”
In practice, three criteria separate the two. First, whether the score goes on an individual record. Formative assessments should be anonymous or unrecorded so learners can get things wrong honestly. Second, whether there is still time to intervene after the results come in. If no sessions remain, it is a summative assessment no matter how far into the course it was given. Third, the number of questions and their difficulty. Formative assessments work better short and frequent, with 5 to 10 questions in under 5 minutes, than a single 20-question test. It is easy to confuse this with Kirkpatrick's four levels, but Kirkpatrick is a layer of "what is measured" (reaction, learning, behavior, results), while formative/summative is an axis of "when and for what purpose it is measured." The two axes intersect: measuring Level 2 learning midway is formative, and measuring it at the end is summative.
The most common failure in practice is scoring formative assessments and feeding the scores into HR decisions. The moment that happens, learners stop revealing what they don’t know. That is how you get a 92-point average on mid-course quizzes and half the class failing the final exam: they compared answers before submitting. The opposite failure exists too: running formative assessments diligently but never applying the results to the next session. Learners start asking “why are we doing this?” and response rates fall to around 40% within three weeks. Formative assessment lasts only when learners can see something change because of their results.
Hwi KimCEO of HRDIST & Talent Development Systems Consultant
John KimJohn Kim, author of B2B Marketing Design and CEO of Nanum ConsultingThe distinction was established in 1967 by the educational evaluation scholar Michael Scriven, who set out formative and summative evaluation as separate concepts.
Consider a hypothetical case. Company A, an auto-parts manufacturer with about 620 employees, ran an 8-week job-orientation program for 34 newly hired production managers. Over three consecutive cohorts, the pass rate on the final certification exam stalled at 58%, and 11 retraining requests came in within three months of deployment. The training team saw the problem as “we didn’t know for 8 whole weeks that no one was keeping up.”
- When Manager Park on the training team reviewed an existing program, the only assessment was a single certification exam in Week 8. When asked, the instructor said, “It felt like everyone understood.” Manager Park replied, “Let’s go by data, not feelings,” and decided to add a 10-question, 5-minute quiz every Friday.
- Instead, they set conditions. Quiz scores would be anonymous and kept out of personnel records, with results going only to the instructor and the individual. At the first cohort’s orientation, Manager Park made it clear: “We are not grading you with these scores—we are grading the course.”
- In the Week 2 quiz, the correct-answer rate on process standard-cost calculation questions was 31%. Instructor Kim cut 30 minutes of the new-product case section from the Week 3 lecture plan and added three standard-cost practice problems. The re-measured correct-answer rate was 79%.
- In Week 5, instead of a quiz, six senior practitioners joined for 15-minute one-on-one feedback sessions. The same comment came up for 12 participants: “The report has numbers but no one-line cause.” A “probable cause” field was added to the Week 6 assignment template.
- The Week 8 certification exam stayed as it was: 60 questions, 70 to pass. Formative quiz response rates averaged 94% over the eight weeks. Manager Park attributed that to learners seeing each week’s results reflected in the next week’s lecture.
- The cohort’s pass rate rose from 58% to 85% (29 of 34 learners), and re-training requests three months after job placement fell from 11 to 2. Starting with the next cohort, the team is redesigning the weekly quiz questions as answer-rate data builds up.