Understanding Why STAR Test Scores Fluctuate - Statistical Explanations and Actions

Sep 15, 2026

Uploaded by Aladdin

STAR test scores can decline between administrations for several reasons, including measurement error inherent to any educational test, natural fluctuations in student performance, and the statistical effect known as regression to the mean. The text explains how to judge whether a decline is meaningful by comparing individual score drops against the standard error of measurement (SEM), using a rule of thumb of about 1.5 times SEM. Guidance is provided for comparing averages and investigating larger declines, including considering retesting under monitored conditions.

LanguageEnglish

CategoryExam

FormatPDF

Pages2

Reading Time5 min

AccessFree to download

Understanding why STAR test scores fluctuate

Measurement error

Fluctuations in student performance

Regression to the mean

What to do when scores decline

Understanding why STAR test scores fluctuate Scores on any test may decline from one administration to the next for a variety of reasons, some of which are easy to understand, while others are difficult. Three of the reasons scores decline are the following: 1. Measurement error, which is present in any test 2. Fluctuations in student performance 3. A statistical phenomenon called "regression to the mean" Measurement Error Educational tests are all imperfect; as a consequence, every educational test score contains some degree of error. As a result, if a group of students were to take a test twice, some of their scores would be expected to decline, even if their skills have improved during the interval between tests. To determine whether a student's score has declined more than would be expected, one must compare the amount of the score decline to a statistic called the "standard error of measurement" or"SEM." The SEMs for STAR tests vary somewhat by Year and by score level; their typical magnitudes are tabulated in the technical manuals for each of the STAR tests. As a rule of thumb, a score decline should not be considered meaningful unless it is more than about 1.5 times the SEM. For example, imagine that a student obtained a scale score of 600 on STAR Maths™ in September, and 550 when she took it again in October; if the SEM for STAR Maths is 40 points, the 50-point decline should not be considered significant since it is less than 1.5 times 40 (the SEM), or 60 points. If there has been no learning at all between the first and second test administration, about half of all scores will increase, and the other half will decrease, solely because of measurement error. Although individual students’ scores may fluctuate substantially, the average score of all students will be nearly identical on both the first and second tests. If there HAS been learning growth between test administrations, some scores may still decline because of measurement error, but the percentage of students with declining scores should be less than half, and more than half of all scores will normally increase. The longer the interval of time between successive test administrations, the smaller the proportion of score declines will be. When some scores decline, there is a natural tendency to distrust the validity of the test. This is inappropriate, however, if measurement error alone would explain the decline – as is often the case in educational testing. As a check on the validity of a test where there are some score declines, a teacher should compare the average scores and the spread of scores on the first and second test administrations. Has the group's average and spread of scores increased? Is the increase close to what is expected, given the time interval between tests and the typical amount of growth expected over that interval? If the answer to each question is "yes", there is no reason to be concerned about the validity of the test scores. Fluctuations in Student Performance A test score is a measure of a student's performance on a given day. Human performance is not always consistent from one occasion to another. A student may perform at her best during one administration of a test, and somewhat less than her best on another occasion. These fluctuations may be related to such things as illness, distractions, anxiety, attention or motivation. We are not surprised by such fluctuations when they affect physical performance – such as running speed or jumping distance in athletic contests. Similarly, we should not be surprised that inconsistent test performance may result in declining scores on occasion. Regression to the Mean This is an inevitable phenomenon that occurs whenever there are multiple observations of an attribute that is not perfectly consistent from one occasion to another. Educational test scores are a classic instance of that. When students are measured twice, there is a tendency for those with the highest s

FAQs

What are the main reasons STAR test scores fluctuate between administrations?

Scores may decline due to measurement error, genuine fluctuations in student performance, and regression to the mean.

How can a teacher tell whether a score decline is meaningful?

Compare the individual decline to the standard error of measurement (SEM); a decline should be treated as unlikely to matter unless it is more than about 1.5 times the SEM.

What steps should be taken when scores decline?

First compare average scores over time for the class or larger group. Then compare each individual decline to the SEM; for large declines, investigate possible factors and consider closely monitored retesting or technical assessment guidance.

Sep 15, 2026

Uploaded by Aladdin