Flashcards in Chapter 5: Standardization, Scoring, and Norms Deck (13):
dichotomous responses. Only one of two outcomes is available for each observation or for each test item. Variance is represented differently for binary items – If test scores have no variance, then they cannot be correlated with any other set of scores. But there is a specific way to calculate variance for binary scores that allows for correlation.
are calculated from data in multiple variables in order to form reliable and valid measures of latent, theoretical constructs.
covariance is computed from variability among scores in two different distributions of scores. The covariance represents the degree of association between the variability in the two distributions of scores.
tells you the percentage of people in your sample whose scores fall at or below a given raw score
an ideal, symmetric distribution of scores that serves as the basis for many staitsical procedures and concepts.
Normalized standard scores
standard scores calculated after the distribution of scores has been normalized by stretching the skewed curve into a shape of a normal curve (i.e., a nonlinear transformation that changes the shape of the distribution)
Normative Data or Norms
- these serve as a frame of reference for interpreting raw scores; they allow us to indicate a person's standing on a test relative to the distribution of scores obtained by people of the same chronological age, grade, sex or other relevant demographic characteristicsThree types of norms: Percentiles, standard scores and normalized standard scores. means and standard deviations ARE NOT norms - common mistake
if you create norms using your standardization sample then the normative and standardization sample ARE the same group. If you have a normative sample, then you thus have a standardization sample (either the same one or a different one). BUT having a standardization sample doesn't mean you have a norming sample.
Percentile (or Percentile Ranking)
- a ranking that provides information about the relative position of a score within a distribution of scores; specifically, a percentile tells you the percentage of people in your sample whose scores fall below a given raw score
Standard Scores (e.g., z and T scores)
standard scores (e.g., z and T scores) tell you where a raw score sits in your distribution relative to the mean (they involve simple linear transformations of raw scores that do not affect the shape of the distribution)
the test is administered using the same materials, same directions, and same scoring procedures in the same manner to everyone
- standardization sample = large group of people for whom you follow the standardization procedures and then indicate how they perform on the test. you could have more than one standardization sample (ex. scoring method A vs. B, or administration method A vs. B), sometimes called a REFERENCE SAMPLE