LET Elementary Assessment of Learning — Principles of Assessment and the Table of SpecificationsSummary

Measurement is the quantitative process of assigning numbers or scores to indicate the level or extent of a characteristic or performance. It answers the question 'How much?' For example, when a Grade 4 student scores 42 out of 50 on a math quiz, that raw score (42) is the product of measurement. Measurement is purely numerical and objective in its execution; it does not involve judgment or interpretation about the value of the performance.

Assessment is the broader, ongoing process of systematically gathering evidence about student learning from multiple sources over time. It goes beyond a single test score. In your Grade 2 classroom, assessment might include quiz results, observation checklists during group work, student portfolios, performance tasks, and self-reflection journals. Assessment uses measurement (the scores) as one input but combines many types of evidence to develop a fuller picture of what learners know and can do.

Evaluation is the act of making a judgment about worth, quality, or achievement against a predetermined standard and then making a decision. It answers questions like 'How good is this work?' and 'Has the learner passed?' When you decide that a score of 42/50 qualifies as 'Very Satisfactory' under the DepEd grading scale, you are evaluating. Evaluation is inherently value-laden and requires professional judgment. It is the culmination of measurement and assessment.

Assessment FOR learning is formative assessment—ongoing, diagnostic evaluation that occurs during instruction to gather real-time feedback about student progress and learning difficulties. In your Grade 3 English class, you might pause mid-lesson to check students' understanding through think-pair-share, exit tickets, or quick quizzes. The purpose is not to grade but to identify gaps immediately so you can adjust teaching. The teacher is the primary user of the information, and results are often ungraded or low-stakes. This aligns with DepEd's emphasis on responsive, differentiated teaching.

Assessment OF learning is summative assessment—formal evaluation that occurs at the end of a unit, term, or course to determine and certify what learners have achieved overall. Your Grade 5 unit test on the Water Cycle, administered after three weeks of instruction, is summative. It answers whether the learners met the intended learning outcomes. Results are typically graded and reported to parents, administrators, or used for placement decisions. It is a high-stakes assessment in most classrooms.

Assessment AS learning develops learners' metacognition (awareness of their own thinking and learning processes) and self-regulation. Here, students actively monitor their own progress through self-assessment, peer feedback, reflection journals, and goal-setting. When a Grade 6 student writes in their learning journal, 'I find dividing fractions difficult because I forget to flip the second number,' they are engaging in AS learning. The student becomes the primary user of the assessment information to adjust their own study strategies and learning behaviors.

Placement assessment is administered before instruction begins to determine learners' entry-level knowledge, skills, and readiness and to place them appropriately in a learning sequence or program. For example, at the start of Grade 1, a teacher might use a simple oral and manipulative task to determine which children recognize numerals 1–10 and which need more exposure. Placement assessment asks 'Where should this learner start?' and 'What is their readiness level for the upcoming instruction?'

Diagnostic assessment occurs before or during instruction to pinpoint specific learning strengths, weaknesses, and the underlying causes of learning difficulties. It goes deeper than placement; it investigates 'What exactly is wrong and why?' If a Grade 4 student struggles with word problems in math, a diagnostic assessment might reveal that the issue is not arithmetic but reading comprehension or inability to extract relevant information. Diagnostic assessments often use interviews, think-aloud protocols, error analysis, and detailed observation to uncover root causes of misconceptions.

Formative assessment is administered during instruction to monitor progress, identify learning gaps, and provide immediate feedback to guide teaching and learning. Examples include daily quizzes, observation checklists, peer reviews, and student work samples reviewed mid-unit. Formative assessments are typically low-stakes and ungraded (or low-weight in the grade), allowing learners to take intellectual risks without fear of heavy penalty. Feedback is descriptive and actionable, not just a score.

Summative assessment is administered at the end of a unit, term, or course to judge and report overall achievement and certify competency. Examples include unit tests, final exams, major projects, and standardized tests like the National Assessment Test (NAT). Summative assessments are usually graded and high-stakes (significantly weighted in the overall grade). They provide evidence of whether learners have met the intended learning outcomes and determine grades, promotion, or certification.

Norm-referenced assessment interprets a learner's score by comparing it to the performance of a reference group (the norm group), typically a large, representative sample of peers at the same grade level or age. Results are reported as ranks, percentile ranks, or standard scores. For example, if a student's score falls at the 75th percentile on a norm-referenced test, it means they scored higher than 75% of the norm group. The meaning of the score is entirely relative to the group. Entrance examinations like UPCAT are norm-referenced: your usefulness as a result depends on how you rank among applicants.

Criterion-referenced assessment interprets a learner's score by comparing it to a fixed, predetermined standard (criterion) that defines mastery or competency. Results are reported as the degree of mastery of specific, well-defined learning objectives. For example, 'The student demonstrated mastery of 9 out of 10 fractions competencies' or 'Achieved 85% on the reading comprehension standards.' The criterion is set independent of how the group performs; theoretically, all learners could master all criteria, or none could. The LET itself is criterion-referenced: passing is defined by a fixed standard (a weighted average of at least 75% with no test score below 50%), regardless of how other test-takers perform.

The Table of Specifications is a two-dimensional test blueprint that maps the content domain (topics or standards taught) against cognitive levels of learning objectives (usually Bloom's revised taxonomy: remember, understand, apply, analyze, evaluate, create) and specifies the number of test items for each intersection. A TOS typically shows content topics in rows and cognitive levels in columns; cells contain the count of items allocated to that content-level pairing. For a 50-item Grade 5 science unit test on the solar system, the TOS might allocate 8 items to remembering facts, 12 to understanding concepts, 15 to applying knowledge to scenarios, 10 to analyzing planetary data, and 5 to creating models—with these distributed across the four celestial-body topics taught. The TOS operationalizes content validity.

Content validity is the degree to which a test adequately samples the subject-matter domain it purports to measure and the appropriateness of using the test results to make inferences about learner competency in that domain. A Grade 3 spelling test that includes words from the lesson list, at appropriate difficulty levels, and in sufficient number to represent the lesson content is content-valid. Content validity is established through expert judgment—typically, curriculum specialists or experienced teachers review the test against a clearly defined Table of Specifications and the curriculum standards. It is not computed statistically like other validity types.

Criterion-related validity is the degree to which scores on a test correlate with an external criterion—a separate, independent measure of the same trait—and the appropriateness of using the test to predict or estimate performance on that criterion. It splits into two types. Concurrent validity examines correlation between the test and a criterion measured at the same time (e.g., how well a classroom math quiz correlates with a standardized math assessment given in the same week). Predictive validity examines whether the test forecasts future performance (e.g., whether a Grade 6 readiness test predicts success in Grade 7 mathematics). Both are expressed as a correlation coefficient; the higher the coefficient, the stronger the relationship. Entrance tests like UPCAT aim for high predictive validity to forecast college success.

Construct validity is the degree to which a test measures the theoretical psychological construct (trait, ability, or quality) it claims to measure, such as critical thinking, reading comprehension, anxiety, or self-esteem. Establishing construct validity is complex: it requires converging evidence from multiple sources, such as factor analysis, convergent validity (the test correlates with other measures of the same construct), divergent validity (the test does not correlate highly with measures of different constructs), and logical analysis of the theory underlying the construct. For example, a critical-thinking test would have construct validity if it correlates with other measures of critical thinking, does not simply measure vocabulary knowledge, and items logically reflect the definition of critical thinking used.

Face validity is the superficial appearance that a test looks like it measures what it claims, regardless of whether it actually does. It is the weakest and least rigorous form of validity and is sometimes called 'apparent validity.' A Grade 4 'critical thinking' test that consists of 50 vocabulary-matching items might have high face validity (it looks like it is testing thinking) but low actual construct validity (it is mostly testing word knowledge). Face validity matters for test acceptance—if a test does not look relevant to stakeholders, they may resist using it—but it provides no guarantee of actual measurement validity.

Test-retest reliability is measured by administering the same test twice to the same group of learners under similar conditions, then correlating the two sets of scores. The resulting correlation coefficient (often denoted as r or rtt) indicates stability of scores over time. A high correlation (close to 1.0) suggests the test produces stable, consistent results; learners who score high on the first administration score high on the second, and vice versa. A limitation is that performance on the second administration may be affected by memory of the first test or practice effects, which can artificially inflate the correlation. Test-retest reliability is most appropriate for measuring stable traits like aptitude or personality.

Parallel forms reliability is measured by creating two equivalent (parallel) tests that measure the same construct with similar content, difficulty, and format, then administering both to the same group and correlating the scores. If both forms are truly parallel, the correlation should be high, indicating that either form produces equivalent results. This method avoids memory and practice effects that confound test-retest reliability. The challenge is constructing truly equivalent forms, which is time-consuming and requires careful item matching. If you develop two versions of a Grade 2 phonics screening test, parallel forms reliability ensures both versions measure phonics skill equally.

Split-half reliability is measured by dividing a single test into two equivalent halves (usually odd-numbered items vs. even-numbered items), calculating a score for each half, and correlating the two halves. Because a shorter test is inherently less reliable, the Spearman-Brown prophecy formula is applied to adjust the correlation upward to estimate the reliability of the full test. A high split-half correlation (after correction) indicates internal consistency—the two halves measure the same trait similarly. This method requires only a single test administration, making it practical for classroom use. For example, you might correlate items 1, 3, 5, 7... with items 2, 4, 6, 8... on a 30-item Grade 5 reading comprehension test.

Internal consistency reliability measures the degree to which all items in a test measure the same trait or construct homogeneously. It examines how items interrelate. Cronbach's alpha is used for scaled items (e.g., Likert-scale responses with multiple points); the Kuder-Richardson formulas (KR-20 for heterogeneous items, KR-21 for homogeneous items of equal difficulty) are used for dichotomous items (right/wrong, yes/no). A high coefficient (typically above 0.70) indicates good internal consistency; items cohere and measure a single construct. A coefficient below 0.60 may indicate the test measures multiple unrelated constructs or contains poorly written items. Internal consistency can be estimated from a single test administration using statistical analysis.

A test can be reliable without being valid, but a valid test must be reliable. This relationship is often illustrated with the dartboard analogy. If arrows are clustered tightly on the bullseye, the test is both valid and reliable (measuring the right thing consistently). If arrows are clustered tightly but far from the center, the test is reliable but not valid (consistent but wrong target). If arrows are scattered all over, the test is neither reliable nor valid. Reliability is a necessary but not sufficient condition for validity. A test must first produce consistent scores (reliable) before it can measure the right thing (valid).

Bloom's revised taxonomy organizes learning outcomes into six cognitive levels in increasing complexity: Remember (recall facts and basic concepts); Understand (explain ideas or concepts); Apply (use information in new situations); Analyze (draw connections among ideas); Evaluate (justify a decision or choice); and Create (produce new or original work). Each level represents a different cognitive demand. The taxonomy is used to classify learning objectives and to ensure assessments target the cognitive levels intended. The DepEd K-12 BEC refers to similar hierarchies. A test with only Remember and Understand items does not assess higher-order thinking; distributing items across all six levels creates cognitive balance aligned with deeper learning goals.

These three concepts form a sequential chain. Measurement produces the raw numbers (scores). Assessment gathers those numbers along with other evidence. Evaluation interprets the evidence and makes a judgment. You cannot evaluate without assessing, and you cannot assess without measuring. This hierarchy helps you distinguish the three terms precisely—a frequent source of LET questions.

Assessment FOR learning (formative, feedback-focused) typically occurs during instruction. Assessment OF learning (summative, certification-focused) occurs at the end. Assessment AS learning (metacognitive, student-centered) can occur throughout but is most powerful when integrated into all phases. While not identical, formative assessments usually serve FOR learning purposes, and summative assessments serve OF learning purposes.

These standards of comparison determine how to interpret and use test results. Norm-referenced results are meaningful only in relation to a norm group; they answer 'How does this student rank?' Criterion-referenced results are meaningful in relation to a fixed standard; they answer 'Has this student mastered the competency?' Choosing the right standard depends on the assessment purpose: entrance tests typically use norm-referenced; mastery checks and DepEd report cards typically use criterion-referenced.

A well-constructed TOS operationalizes content validity, ensuring the test fairly samples the content domain. This supports fairness (all learners are assessed on what was taught) and defensibility (the test blueprint is documented and rational). The TOS is the practical link between validity principles and classroom test design. Without a TOS, content validity becomes difficult to claim or defend.

Learning objectives stated at specific cognitive levels (e.g., 'Students will apply the water cycle to local ecosystems') drive the TOS. The TOS allocates items to match these levels. Item writing then operationalizes the TOS by crafting questions at the correct level and about the correct content. This backwards design from objectives through TOS to items ensures coherence and alignment.

Validity is not a single property but a constellation of evidence. Content validity (sampling the domain), criterion-related validity (relating to external criteria), and construct validity (measuring the intended trait) all contribute to an overall judgment about whether test results can be appropriately interpreted and used. A comprehensive validation argument draws on multiple types of evidence.

Reliability is estimated through multiple methods, each addressing a different aspect of consistency: test-retest measures stability over time, parallel forms measure equivalence, split-half and internal consistency measure item homogeneity and coherence. Multiple reliability estimates strengthen confidence that a test produces consistent scores. A test with high test-retest but low internal consistency suggests fluctuating performance or inconsistent item quality.

This is the foundational conceptual relationship: a test must first be reliable (produce consistent scores) before it can be valid (measure what it should). Unreliable scores are 'noise' that prevents validity. Conversely, a reliable test can measure the wrong thing perfectly consistently (reliable but not valid). Therefore, reliability is necessary but not sufficient for validity. Always ensure reliability first, then validate.

RA 7836 (Code of Ethics) obligates teachers to employ fair, just, and non-discriminatory assessment. RA 7610 (Child Protection) mandates that assessments never harm or exploit learners. Valid, reliable, confidential assessment practices directly uphold these ethical and legal duties. Poorly designed assessments (invalid, unreliable, biased) violate these codes and potentially harm learner welfare and academic standing.

Assessment FOR learning (formative) benefits from easier items to quickly identify gaps; feedback should be immediate, descriptive, and actionable. Assessment OF learning (summative) typically includes a broader mix of difficulties to discriminate learners; feedback is formal and often grades-focused. Assessment AS learning includes reflective prompts and self-monitoring tools; feedback is internal. The assessment purpose shapes the item difficulty distribution and type of feedback provided.