Term
|
Definition
| The quality of a test score that suggests it is 1) sufficiently consistent, and 2) relatively free of measurement error. |
|
|
Term
|
Definition
| Difference between observed score and true score. Factors that affect observed score are due to measurement error. |
|
|
Term
|
Definition
| An abstraction, used to unify or produce responses meausured in an assessment. Construct produces behavior |
|
|
Term
|
Definition
| Discrete events, observable to the self or other, an assessment of performance. |
|
|
Term
| Assumptions of Traditional Measurement Theory |
|
Definition
1. Important psychological processes CAN be measured.
2. Construct produces behavior on measure.
3. Variations in scores represent "true" differences among individuals.
4. All scores contain error.
5. Obtained score = true score + error
6. True score is constant.
7. Error is random.
8. Scores represent an underlying process (construct) that generates behavior.
9. What the measure assesses is more important than the measure used to assess it.
10. Construct cannot be fully operationalized by a measure, even in theory. |
|
|
Term
| Measurement problems common to psychological assessment |
|
Definition
1. No single approach to the measurement of any construct is universally accepted.
2. Psychological measurements are based on limited samples of behavior.
3. Measurement is always subject to error.
4. Lack of well defined units
5. Psychological constructs cannot be defined only in terms of operational differences. |
|
|
Term
| Different forms of reliabitity |
|
Definition
Test-restest
Alternate forms
Split-half
Internal Consistency
Inter-rater |
|
|
Term
|
Definition
Purpose: to determine whether scores on one test are comparable over time.
Source of Error: changes over time- life changing events, maturation, etc |
|
|
Term
|
Definition
Purpose- Are scores produced by different raters consistent?
Source of Error- scorer differences, human judgement |
|
|
Term
|
Definition
Purpose: Are scores produced by different versions of the same test consistent?
Source of Error: item sampling, changes over time |
|
|
Term
|
Definition
Attempts to assess for item inconsistency, concerned with the extent to which scores produced by half of the items are consistent with scores produced by the other half.
Source of Error: item sampling, nature of split |
|
|
Term
|
Definition
Tests whether each item correlates positively with all other items
Source of Error: item sampling, test heterogeneity |
|
|
Term
|
Definition
1. On long tests, alpha can be misleadingly high.
2. High alpha can mask bad items
3. Very high alpha could also mean that the operationalization of the construct is too narrow. |
|
|
Term
|
Definition
| Setting an upper limit on one variable. By setting an upper limit on Variable x, subjects can continue to have higher scores on variable y, but they are RESTRICTED on Variable x. The ceiling on variable x is causing a lower correlation than if subjects could continue to have high scores on variable x. |
|
|
Term
| Standard Error of Measurement |
|
Definition
Represents the standard deviation of a hypothetical distribution if a subject were to take a test an infinite number of times.
As the reliability coefficient of a test increases, SEM decreases. SEM gives a Range in which a true score is likely to fall within a given probability (confidence interval).
Multiply SEM by 1.96 to get a 95% confidence interval |
|
|
Term
|
Definition
| for internal consistency for DICHOTOMOUS variables, used less than Cronbach's alpha |
|
|
Term
|
Definition
internal consistency among ALL items
The method used depends on the PURPOSE of obtaining a measure, and the way the measure will be USED. |
|
|
Term
| Sources of Error Variance in Measurement |
|
Definition
Test construction (item selection, instruction)
test administration/environment
test scoring and interpretation
Mood, hunger, motivation, etc
|
|
|
Term
| Systematic Measurement Error |
|
Definition
| When a test systematically measures a trait besides the trait on a test |
|
|
Term
|
Definition
| Any fluctuation in scores that result from factors related to the measurement process that are irrelevant to what is being measured. |
|
|
Term
|
Definition
| the mean of a hypothetical distribution of scores that would be obtained if a person took the test an infinite number of times. You can used obtained score to estimate true score. |
|
|
Term
|
Definition
| index of the strength of the relationship between two variables. Correlation tells the degree and direction. |
|
|
Term
| What are the three conditions for conducting Pearson's R correlation? |
|
Definition
1. Relationship between the variables is LINEAR.
2. The two variables are CONTINUOUS.
3. The pairs of observation are INDEPENDENT of one another. |
|
|
Term
| Significance and what does it depend on? |
|
Definition
Is the relationship reliably different from zero??
Significance depends primarily on SAMPLE SIZE. |
|
|
Term
| Magnitude and what does it depend on? |
|
Definition
The STRENGTH of the relationship between x and y.
Depends on the HETEROGENEITY of a sample, not the N.
Look at significance first, then magnitude! |
|
|
Term
| Three Factors that Affect Alpha |
|
Definition
1. Content homogeneity of items
2. Nature of construct being assessed: unidimensional vs. multidimensional
3. Number of items. More items = more reliable score. |
|
|
Term
|
Definition
| When a variable is measured that has a ceiling or a floor, we obtain a lower correlation coefficient than if the variable were measured without a floor or ceiling. |
|
|
Term
| How to Detect Restricted Range |
|
Definition
| Look at the frequency distribution of the variables, Standard deviation of your scores compared with those of other researchers. |
|
|
Term
| What factors can contribute to range restriction? |
|
Definition
| Bad luck recruiting, sample selection criteria |
|
|
Term
| Properties of Normal Curve |
|
Definition
Bell-shaped
bilaterally symmetrical
Tails never touch
Baselines stretch to infinity
Uni-modal (single point of maximum height)
Has a mean, mode, median that coincide at the center
Standard deviations are positioned at equal distances along the x axis. |
|
|
Term
|
Definition
Descriptive
Inferential
Estimating population parameters |
|
|
Term
|
Definition
| Name, numbers are used instead of words, identity. SSN, football players jerseys, numerical codes for sex or psychiatric diagnosis |
|
|
Term
|
Definition
| Rank, percentages. Does not assume equal intervals and does not specify distance between intervals |
|
|
Term
|
Definition
| Assumes equal intervals, but does not have absolute zero. Farenheit, Celcius |
|
|
Term
|
Definition
| Assumes absolute zero. Numbers can be +, -, x, /. |
|
|
Term
| Measures of Central Tendency |
|
Definition
Mode- most frequently occurring value
Median- the middle value
Mean - arithmetic average |
|
|
Term
|
Definition
| Describe how much dispersion or scatter there is in a set of data. |
|
|
Term
|
Definition
| Distance between the highest and lowest value |
|
|
Term
|
Definition
| 1/2 of the interquartile range |
|
|
Term
|
Definition
| tops of 1st and 3rd quarters of a distribution. Range between Q1 and Q3 - middle 50% of the distribution |
|
|
Term
|
Definition
| The average of the sum of squares |
|
|
Term
|
Definition
| Square root of the variance. Provides a single value that represents individual differences in a data set. |
|
|
Term
| Importance of Variability |
|
Definition
Without individual differences there would be no variability and test would be useless in helping us make determinations or decisions about people. The greater the amount of variability among individuals, the more accurately we can make distinctions that need to be made among them.
Variability provides a portion of test score interpretation. |
|
|
Term
| Standard Error of the Difference |
|
Definition
Comparing the difference between individual scores.
Comparing two scores obtained by one person on two different PARTS of a test; comparing the scores of two different people on the same test.
Standard Error of the Difference is BIGGER than SEM for either individual score because it's affected by measurement error from BOTH scores. |
|
|
Term
| What is the Spearman-Brown Coefficient used to test? |
|
Definition
| Split-half reliability. Allows you to estimate the internal consistent reliability for the WHOLE test. |
|
|
Term
| Spearman-Brown (split-half reliability) steps |
|
Definition
1. Divide the test into two equivalent halves
2. Compute a Pearson's R between the scores on the two halves of the test.
3. ADJUST the reliability coefficient using Spearman-Brown formula.
Rsb = 2rxy/1+rxy |
|
|
