Quiz 1

Question 1

Challenges to accurate measurement

1) participant reactivity

Answer 1

Participant reactivity: the act of measurement can influence the psychological state of process being measured.

Demand Characteristics: changing ones behavior to align with what is believed the the researcher is studying.

Social Desirability: changing ones behavior to impress the person doing the measurement

Malingering: changing ones behavior to convey a poor impression to the person doing the measurement.

Question 2

Challenges to accurate measurement cont

2) Observer bias and expectation

Answer 2

The observer or person measuring can bring biases that distort their observations. Even subtle biases or expectations can have an effect

Question 3

Challenges to accurate measurement

3) Composite scores

Answer 3

Taking the results from many questions and combining it into one score as a measure of a certain attribute.

Question 4

Challenges to accurate measurement cont

4) Score Sensitivity

Answer 4

Sensitivity is the ability to discriminate between units or measures. Using a measure that’s asks if your feeling good or bad may not be sensitive enough to capture a meaningful response. Thus, likert scales are used in these cases

Question 5

Challenges to accurate measurement cont.

5) Lack of awareness of important psychometric information

Answer 5

Some tests are administered without any knowledge of their psychometric properties. The creation of tests may lack psychometric knowledge which would lead to poorly constructed tests.

Question 6

Who coined psychometry?

Answer 6

Sir Francis Galton

Defined as the scientific study of the attributes of tests

Question 7

Definition of psychological tests

Answer 7

A systematic procedure for comparing the behavior of two or more people

Includes 3 components:

a) tests involve behavioral samples
b) samples are collected in a systematic way
c) the purpose of the test is to compare behavior of two or more people

3 attributes:

a) type of data (scores)
b) reliability
c) validity

Question 8

Purpose of Psychological Tests

Answer 8

To compare the behavior of different people (interindividual differences) or the behavior of the same individuals at different points in time or different circumstances (intraindividual differences).

Question 9

The Goal of Psychological Tests

Answer 9

To measure the psychological attributes of people (e.g. extroversion/introversion, openness to new experiences, readiness for change)

Question 10

Types of tests

Answer 10

Criterion-referenced: there is a predetermined cutoff test score to sort people into (1) those whose performance exceeds the criterion, (2) and those whose performance does not

Norm-referenced: comparing one’s scores with scores from a reference sample or normative sample to understand how they compare with others.

Question 11

Scales

Answer 11

Nominal: only holds identity (e.g. biological sex, depression or no depression)

Ordinal: rank order and identity (e.g. ranking children according to interest in learning, ranking clients in severity of depression)

Interval: identity, order, and quantity (e.g. temperature)

Ratio: identity, order, quantity, and an absolute zero (e.g. distance)

Question 12

Calculations

Mean, standard deviation, z and T scores

Answer 12

See actual card

Question 13

Percentile Rank

Answer 13

Indicates the percentage of scores that are below a specific test score. (i.e. 85% percentile: he or she or they scored higher than 85% of the other people who took this test)

Question 14

Correlation

Answer 14

The measure of association between two variables (x and y). The magnitude and direction of which is measured by the correlation coefficient falling between 0 and 1 (positive correlation) and 0 and -1 (negative correlation)

Question 15

When to use Pearson vs Spearman correlation?

Answer 15

Pearson Correlation (Pearson’s r) is a statistic that measures the LINEAR CORRELATION between two variables X and Y. It can evaluate ONLY a linear relationship between two continuous variables.

Spearman correlation: is a nonparametric measure of rank correlation (dependent on the ranking of two variables) to assess how well the relationship between two variables can be described using a monotonic function (as one variable increases the other increases or decreases BUT not at a constant rate):
These variables are Continuous or Ordinal and based on ranked values for each rather than raw data.

Question 16

Covariance

Answer 16

Is the deviation of each subject’s score on the X variable from the mean of X scores to the corresponding deviation for that subject on the Y variable.

How these two variables move together

Question 17

Test Dimensions

Answer 17

Unidimensional tests: tests that have items reflect only one psychological dimension. Examples: geometry test, a test of courage.

Multidimensional tests with correlated dimensions: A test with multiple dimensions that are correlated with each other. Examples: WAIS/WISC.

Each subtest has its own sub-test score and is considered unidimensional. These multidimensional tests with correlated dimensions are scored to produce a total score across combined subtests. Psychometric quality can vary depending on the subtest.

Multidimensional tests with uncorrelated dimensions: a test with multiple dimensions that are not, or are only weakly, associated with each other. NO TOTAL TEST SCORE IS COMPUTED. Each dimension’s scores are evaluated for psychometric quality.

Question 18

Factor Analysis

Answer 18

Factors: are a set of highly correlated items that represent a psychological dimension (assertive, talking, outgoing) (imaginative, creative, intellectual) each grouping is a factor.

Factor analysis is a statistical procedure often used to evaluate a test’s dimensionality. It involves the examination of correlations between factors.

Exploratory Factor Analysis processes large sets of correlations to analyze dimensionality.

Question 19

Eigenvalues

Answer 19

Ways to use
1) Examine the relative sizes of the eigenvalues. The number of rows in the output is equal to the number of items on the test. The location of the large difference between eigenvalues in the row determines the number of dimensions of the test (e.g. a large difference between rows 2 and 3, indicates that there are 2 dimensions to the test.)

2) Second way is the “eigenvalue greater than 1” rule. The number of eigenvalues in the output that are greater than 1 indicates how many dimension are in the test - this has been criticized
3) Third common way is the scree plot. Scree plot is a graphical presentation of eigenvalues. Where the plot flattens out suggests that the number of factors is one less than the factor number of the flattening out point.

Question 20

Factor Loading

Answer 20

The association between items and factors, each items loading on each factor. Identifying which items are most strongly linked to each factor.

Factor loadings range between -1 and +1, they are interpreted as correlations.

Interpreting factor loadings have 2 important pieces of info:
a) the size of the loading: the degree of associations between an item and a factor (0.3 or 0.4 as reasonably strong, 0.7 or 0.8 as very strong)

b) the direction of the loading (positive or negative). Positive: high score on item = high level of that factor
Negative: high score on item = low level of that underlying factor

Question 21

How are variance and standard deviation related?

Answer 21

Variance represents the average degree to which people differ from each other

Standard deviation: brings the variation back to it’s original measurement scale by getting the square root of the variance.

Four factors to consider

1) variance can never be 0
2) no way to interpret variance by itself
3) requires context for interpretation
4) variance and SD are fundamental concepts for psychometrics