Cognitive bias
When you bend your interpretation of an observation to fit the way you already think. One example is confirmation bias. Learn more from the Conceptually site. To explore the concept in full, take a look at this Wikipedia entry.

Confidence interval (uncertainty)
This is a range of values that you expect a test result or any other measurement of something to fall within. It is usually expressed together with a percentage (called the confidence level) indicating how frequently the expected results would fall within that range. Learn more from the Biostatistics for Dummies site. To explore the concept in greater depth, take a look at this Wikipedia entry about uncertainty.

Cross-sectional v. longitudinal views
These are two ways to view growth. Cross-sectional views look at test results for the same grade level over successive years. Different kids’ results are analyzed at the same grade level each year. Longitudinal views look at the same kids’ test results over successive years. Each answers a very different question. Learn more from the Institute for Work and Health site. To explore the concept in full, take a look at this Wikipedia entry about cross-sectional studies in medicine, or this useful entry from the Learning Hub.

Effect size
A measure of the degree to which one thing impacts another. John Hattie’s book, Visible Learning, is about the comparative effect sizes of things educators do to improve student outcomes. Learn more from this clear explanation from the Statistics by Jim website. To explore the concept deeper, take a look at this Wikipedia entry.

Entity
This refers to the object you’re analyzing. It could be one student, a class, a school, a district, you-name-it. It could be a graduating class cohort or a subgroup of students (e.g., girls).

Fuzzy factor
An easier way to describe the uncertainty and imprecision inherent in test results, survey results, or any other measurement that relies on sampling. The term “fuzzy” describes the inexact nature of the thing being measured, and the imperfect method of measuring it.

Margin of error, standard error (imprecision)
A measure of the accuracy or precision of an estimate (like a test result). In other words, it’s a measure of how spread out data values are around the mean, defined as the square root of the variance. Learn more about how to account for standard error when interpreting test results by reading this essay by Nate Jensen, research director at the Northwest Evaluation Association. To explore the concept in full, take a look at this Wikipedia entry.

Measurement error
An error that occurs when gauging the quantity or quality of observed behavior, attitudes, actions, or other factual events. The error is the difference between the observed value and the true value. Mismeasurement leads to misunderstanding the size or meaning of something, which in turn may lead to errors in judgment. To learn more, you may find this excellent essay in the magazine Science helpful. To explore this big concept more fully, take a look at this Wikipedia entry about observational error.

“N” size
A count of the number of things you’re analyzing. If it’s the number of students in a large middle school classroom, it might be 36. If you’re in a smallish district, it could be 500. The larger the “n” size, the smaller the imprecision or standard error in your test results. Learn more about its meaning from the Sciencing website. To explore the concept in full, take a look at this Wikipedia entry about “n” size and sample size determination.

Noise and signal
Noise is false or irrelevant information that is commingled with true information (signal) in the data. The early stage of any analysis includes distinguishing noise from signal. Learn more from the Conceptually site. To explore the concept in full, take a look at this Wikipedia entry about Nate Silver’s book, The Signal and the Noise.

Spread or dispersion or variability
A measure of how spread out results are. It is measured in standard deviations. (Normal distributions usually have about 68 percent of results within one standard deviation of the midpoint.) It is an often neglected attribute of analysis of test results, because most people ask only about averages or medians. Learn more from the Math is Fun website. To explore the concept in full, take a look at this Wikipedia entry

Statistical significance, or p-value
A concept intended to measure the probability of results occurring by chance. It is now in such disfavor that the American Statistical Association banned its use from their journal in March 2019. It was intended to express the probability that a result did not occur due to chance events. But this became so confused with real-world significance, that it was criticized and buried once-and-for-all in a special issue of their journal, The American Statistician.