CP 6691 - Educational Research Terms
Compiled by Gregory P. Snyder
Use this hyperlink alphabet to browse the glossary. Note: no terms appear under the gray letters. Use the Up Arrow beside each letter in the glossary to return to the alphabet. Return to your previous location when finished.
- Accessible population
- The population from which the sample was drawn.
- Analysis of Covariance (ANCOVA)
- A statistical procedure that allows the researcher to make adjustments to the posttest (dependent variable) mean scores to compensate for variance that occurs when the groups have scored differently on the pretest due to a stated variant.
- Analysis of Variance (ANOVA)
- A statistical procedure that allows the researcher to compare the variance/difference of the means due to error and determine which variances are statistically significant i.e. expected by chance.
- Arithmetic Mean
- Average score, the sum of all the scores divided by the number of scores.
- A systematic process which produces a difference between the sample selected and the population.
- Causal-comparative Research
- Research that compares two groups of individuals who differ in some important characteristic in order to identify possible causes of the phenomenon in which we are interested. (Since it is conducted after the fact it can only explore possible causes and effects.)
- Chi-square Test
- A statistical procedure used to compare categorical data. Chi-square test may be used to determine if the distribution of two groups of subjects, based on categorical variables, is statistically different. Most often is used to compare the distribution of individuals from two or more categorical groups on categorical variables such as social class.
- Cluster sampling
- Selecting naturally occurring groups within a population (e.g. classes within a school).
- Control group
- The group that consists of the subjects that are not given the treatment in the research.
- Contingency Coefficient
- The numerical value derived from a Chi-square Test to indicate the measure of relationship between two sets of categorical data.
- Continuous Data
- Comprised of Interval Data and Ratio/Scale Data.
- Concurrent validity
- Validity that is determined by relating the test scores of a group of subjects to a criterion measure administered at the same time or within a short period of time.
- Content validity
- The degree to which items on a test actually represent the content that the test is designed to measure.
- Construct validity
- The extent to which a particular test can be shown to measure a theoretical construct--some thing that is not directly observable but rather inferred on the basis of their effects on behavior.
- Correlation coefficient
- A statistical tool that permits us to compare the scores of a group of subjects on two variables and obtain a numerical estimate of the magnitude of the relationship between these variables. Correlation coefficient noted as: [r]. Values range from -1(perfect indirect/inverse correlation) through 0(no correlation) to +1(perfect direct correlation).
- Correlational research
- Research that seeks to either to determine the relationships between variables related to the science in question or to make predictions. Correlational research involves one or more correlation coefficients.
- Correlation matrix
- A matrix that depicts all of the possible intercorrelations between a list of variables.
- Descriptive Statistics
- Methods used to derive from raw data certain indices that characterize or summarize the entire set of data.
- Descriptive Research
- One of three broad categories of research (the other categories are correlation and experimental). Descriptive research is designed to describe the characteristics of the subjects of the science. Descriptive research can be roughly classified as either survey research or observational research.
- Dependent variable
- The variable that the researcher is trying to change in the research.
- Directional hypothesis
- A statement of the relationship between the variables being studied or a difference between experimental treatments that the researcher expects to emerge.
- Dissertation Abstracts International
- A monthly compilation of abstracts of dissertations submitted by more than 375 cooperating US and Canadian institutions.
- Experimental group
- The group that consists of the subjects who are given the treatment (program or experience) in the research that the researcher wants to study.
- Experimental (treatment) Research
- A particular experience that the experimental group undergoes that brings about changes in the behavior of the experimental group. The experimental treatment is the dependent variable.
- Extraneous variable
- A variable beyond the control of the researcher that influences the results of a research project (i.e. dependent variable) but has nothing to do with the experimental treatment (i.e. independent variable).
- Face validity
- The degree to which a test appears to measure what it purports to measure.
- A statistical procedure (also known as Analysis of Variance (ANOVA) used to compare two or more groups for significance of the statistical difference between/among them.
- The degree to which we can generalize (draw a general conclusion; to give general applicability) the results of a research study to the population from which the researchers sample is drawn.
- High inference behavior
- Behavior that requires an observer to translate his general impression into a single judgment, and calls for a high level of inference or judgment.
- Hypothesis (Research Hypothesis)
- A statement that describes how the researcher expects the research to come out, the differences or relationships that the researcher expects to find.
- Independent variable
- The variable that is expected to bring about or account for a difference or a change in the dependent variable.
- Inferential Statistics
- Methods that allow the researcher to generalize characteristics from his set of data to a larger population.
- Interval Data
- Indicate the relative position of the subjects on a discernible scale. The scale does not, however, have a baseline/an absolute 0.
- Leading questions
- Those questions that are phrased in such a way as to lead the respondent to consider one reply more desirable than another.
- Low inference behaviors
- Specific, clearly defined kinds of behavior that an observer notes in which the observer is not required to interpret or draw inferences.
- The arithmetic average of all values in a distribution.
- The number which lies at the midpoint of the distribution and divides the distribution of scores into two equal halves.
- Measure of central tendency
- A numerical index of the average score of the distribution e.g. mean, median, and mode.
- Mental Measurements Yearbooks (MMYB)
- A manual that provides test reviews and references on test construction, use, and the limitations of specific tests.
- Translating the findings of related studies into effect sizes i.e.. the difference between the effect of the experimental and control treatments on the variable.
- The most frequently occurring score in a distribution.
- A learned treatise (systematic exposition or argument in writing including a methodical discussion of facts and principles involved and conclusions reached) on a small area of learning.
- Multiple Regression Analysis
- A statistical procedure that permits the researcher to simultaneously explore the relationships of several independent (predictor) variables on the dependent (criterion) variable.
- Negative correlation
- Low values of one variable are paired with high values of the other or high values of one variable are paired with low values of the other.
- Nominal/Categorical Data
- Data that indicate that the subject(s) fit in a particular class or category. Categorical groups indicate relative position i.e.: low, middle, or high. Nominal/categorical data are the least powerful data types.
- Nonparametric Statistics
- Statistics comprised of noncontinuous data that can make few if any assumptions about the characteristics of the population.
- Normal (Bell) Curve
- A frequency polygon/frequency curve for which the mean, median and mode all are located at the same point.
- Null hypothesis
- A statement that no relationship exists between the variables being studied or no difference will be found between the experimental treatments.
- The degree to which a measure may be influenced or distorted by the beliefs or biases of the individuals who administer or score it.
- Observational Research
- A type of descriptive that involves observation , which in this case refers to any objective procedure for recording the characteristics or behavior of the subjects. Tests are considered a form of observation.
- One-tailed Test of Significance
- A graphic comparison of the high tail of one error curve with the low tail of the other.; used with a directional hypothesis.
- Ordinal/Rank Order Data
- Indicates the relative position of the subjects within a group.
- Parametric Statistics
- Statistics comprised of continuous data that make various assumptions about the characteristics of the population from which a sample is drawn.
- Pearson Product-Moment Correlation
- A technique for determining the correlation coefficient using continuous data such as height and weight.
- The defined group to which the researcher plans to apply (generalize) the research results.
- Population validity
- The degree to which the sample of subjects is representative of the population from which they were selected.
- Positive correlation
- High scores on one variable are paired with high scores on the other variable or low scores on one are paired with low scores on the other.
- Predictive/criterion-related validity
- The degree predictions made by a test are confirmed by the later behavior of the subjects. e.g.. The correlation between an aptitude test and an achievement test.
- Preliminary source (of information)
- Indexes or abstracts used to locate books, articles, and other educational documents that relate to a problem being researched (e.g. RIE-Resources in Education; CIJE-Current Journals in Education; Education Index; Psychological Abstracts).
- Primary (First hand) source (information)
- Publications in which persons who carry out research report their findings (e.g. professional Journals, reports, scholarly books, and monographs).
- Probability error
- A value produced as a result of an inferential statistical calculation. It indicates the probability that the computed value of the statistical test is a random occurrence - noted as [p].
- Quasi-experimental research
- Studies that are essentially experimental in nature where it is not possible to assign subjects randomly to the experimental and control groups.
- Random error
- Sampling errors that occur when a variable is exhibited in the sample at a greater rate than the population even though the selection was random.
- The difference between the lowest and highest scores (usually, but not always, 5 times as large the standard deviation).
- Range of Correlation Coefficients
- High Correlation (+ or - value) range ~.7-.1.00; Moderate Correlation: (+ or - value) range ~.4-.69; Low Correlation: (+ or - value) ~.1-.39
- Ratio Scale
- A data scale that has all the characteristics of nominal, ordinal, and interval scales and which contains a baseline/an absolute 0 - a scale value which signifies the absence of the attribute being measured.
- Response effect
- The tendency of the respondent to give incorrect responses e.g. the answer given by the respondent and the true answer.
- A defined, relatively small portion of the population the researcher selects to participate in the study.
- Secondary (second hand) source (of information.)
- Publications in which the author is reporting on research that someone else carried out.
- Semi-structured interview
- Interviewer asks listed questions but may ask unplanned questions that follow up respondent's answers.
- Simple random sampling
- Selection of a sample in which all of the individuals have an equal and independent (selection of one does not affect in any way the selection of any other) chance of being selected as a sample member. (Usually a table of random numbers is used.)
- Spearman's rho
- A statistical technique for determining the correlation coefficient using data in the form of ranks; often called the rank-order correlation technique.
- Standard deviation
- A measure of dispersion in a distribution. Calculated by taking the square root of the variance. See variance for its computational procedure.
- Statistical Significance
- A determination that the difference or relationship that the researcher finds is likely to occur again if the study is repeated using other samples drawn from the same population. A statistical indication that the results of a study could occur by chance less than a particular percentage of the time. Typical confidence levels are 95 and 99 percent.
- Stratified sampling
- Classifying the population into levels based on a particular characteristic to ensure that certain subgroups will be represented in proportion to the population itself.
- Structured interview
- Interview in which the interviewer asks specific pre-planned questions without deviation.
- The individual (or entity) that is studied in a research project. Subject is noted as [S].
- Survey research
- A type of descriptive research that typically employs questionnaires and interviews in order to determine the characteristics, opinions, preferences, and perceptions of persons of interest to the researcher.
- New insights based on research; impact of the research on educational practice and society as a whole.
- Systematic sampling
- Obtaining a sample from a defined population by placing the members on a list, beginning at a random position on the list, and selecting every Nth subject on the list until the desired sample size is reached.
- Target population
- The population to which the researcher wants to generalize the findings.
- Test Collection Bibliographies
- Text published by Educational Testing Services (ETS) that includes the name of the test, author, date published, age or grade level for which appropriate, name and address of publisher, and a brief description of the variables the test is designed to measure.
- Test reliability
- The degree to which test scores are free from measurement errors. Expressed as a correlation's coefficient that reflects the degree of the relationship between two sets of scores obtained from the same subjects under different conditions reflects the internal consistency of a measure and its stability over time.
- A parametric statistical test (procedure) designed to determine if the mean scores of two groups to determine if the difference is statistically significant.
- Two-tailed Test of Significance
- Used in conjunction with a null hypothesis when the researcher does not which group will have the higher mean.
- Unstructured interview
- Interviewer follows a general plan and asks questions and makes comments to meet interview objectives.
- The degree to which a test actually measures the variables it claims to measure.
- Anything in the research situation that varies/differs.
- The extent to which scores differ/vary, reflected by range, variance, and standard deviation.
- A measure of dispersion in a distribution. Determined by (1) calculating how much each score deviates from the mean, (2) squaring these deviations, and (3) adding the squared deviations for all points along the distribution.
- Zero correlation
- No systematic relationship can be found to exist between the two variables being measured.
- Scores of the values of interest that reflect distribution on a normal curve.