AssessmentA broad term used to describe a systematic process used to collect information.TestA systematic procedure for measuring a sample of behavior from a domain.MeasurementThe process of assigning numbers to some characteristic according to some specified rules.Item Difficulty for a test item scored incorrect (0) or correct (1) from a classical test theory perspectiveProportion of individuals out of the total number of individuals who answered the item correctly (40 students out of 100 answered the item correctly = .40; the bigger the number the "easier" the item).Item Discrimination for a test item scored incorrect (0) or correct (1) from a classical test theory perspectiveDegree to which an item differentiates people who score high on the total test from those who score low on the total test. Item-total correlation is one way of defining item discrimination. You want items to have positive values for item discrimination.Item Variance for a test item scored incorrect (0) or correct (1) from a classical test theory perspective<span>Item variance = (proportion of individuals who answered the item correctly) x (proportion of individuals who answered the item incorrectly)
p x q</span>Covariance between 2 variablesCovariance is a measure of the linear association between 2 variables with the 2 variables represented in their original units (not standardized). Unlike a correlation coefficient that goes from -1 to +1, the value of a covariance can go from - infinity to + infinity.Pearson product moment correlation (r) between 2 variablesPearson correlation (r) is a measure of the linear association between 2 variables with the 2 variables being standardized (mean = 0, SD = 1). Correlation coefficient that goes from -1 to +1Variance-Covariance MatrixIn a variance-covariance matrix the variances of the variables are represented in the diagonal of the matrix; the covariances between pairs of variables are represented in the off diagonalConstructAn unobserved or latent variable that is used to explain some observed phenomenon or experience. Latent variables are represented by a circle in measurement models.Reflective Measurement ModelA measurement model that holds that holds that a construct is viewed as the cause of the observed or measured variables.Formative Measurement ModelA measurement model that holds that holds that the observed or measured variables are the cause of the latent variableOperational Definition of a ConstructA definition that is clear and detailed that defines the construct by specifying the procedures used to measure a constructContent ValidityOne of five sources of measurement validity evidence that often involves experts who are knowledgeable about the construct making evaluative judgments about whether the items used to measure the construct adequately reflect the definition of the construct.Convergent ValidityA construct measured by 2 methods should produce a moderate to strong correlation between the 2 measures (e.g., a person's anxiety determined by a questionnaire and a person's anxiety determined by a structured interview).Discriminant ValidityScores on the measure are not related to other measures that are theoretically different (if Grit is theoretically different from Conscientiousness, the correlation between these measures should not be strong)Nomological NetworkAn approach used as part of the process of gathering evidence of construct validity of educational and psychological measures. Similar to the research process, relationships based on theory between measures that the new measure should relate to and measures it should not relate to are tested.<span>Reliability</span>
p x q</span>Covariance between 2 variablesCovariance is a measure of the linear association between 2 variables with the 2 variables represented in their original units (not standardized). Unlike a correlation coefficient that goes from -1 to +1, the value of a covariance can go from - infinity to + infinity.Pearson product moment correlation (r) between 2 variablesPearson correlation (r) is a measure of the linear association between 2 variables with the 2 variables being standardized (mean = 0, SD = 1). Correlation coefficient that goes from -1 to +1Variance-Covariance MatrixIn a variance-covariance matrix the variances of the variables are represented in the diagonal of the matrix; the covariances between pairs of variables are represented in the off diagonalConstructAn unobserved or latent variable that is used to explain some observed phenomenon or experience. Latent variables are represented by a circle in measurement models.Reflective Measurement ModelA measurement model that holds that holds that a construct is viewed as the cause of the observed or measured variables.Formative Measurement ModelA measurement model that holds that holds that the observed or measured variables are the cause of the latent variableOperational Definition of a ConstructA definition that is clear and detailed that defines the construct by specifying the procedures used to measure a constructContent ValidityOne of five sources of measurement validity evidence that often involves experts who are knowledgeable about the construct making evaluative judgments about whether the items used to measure the construct adequately reflect the definition of the construct.Convergent ValidityA construct measured by 2 methods should produce a moderate to strong correlation between the 2 measures (e.g., a person's anxiety determined by a questionnaire and a person's anxiety determined by a structured interview).Discriminant ValidityScores on the measure are not related to other measures that are theoretically different (if Grit is theoretically different from Conscientiousness, the correlation between these measures should not be strong)Nomological NetworkAn approach used as part of the process of gathering evidence of construct validity of educational and psychological measures. Similar to the research process, relationships based on theory between measures that the new measure should relate to and measures it should not relate to are tested.<span>Reliability</span>
8
0
