Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
56 Cards in this Set
- Front
- Back
|
What is psychological measurement> |
Measuring an attribute of a person's mind. |
|
|
How can you obtain meaning in Psychological Measurement? |
By comparing a person's score to other individuals. |
|
|
What are norms? |
Comparison data are called norms and should– for most psychological tests – have beenobtained as part of the process of ‘testconstruction’
|
|
|
What are norm referenced tests? |
Tests or measures that require norms for the scores to be interpretated. |
|
|
What type of testing doesn't need norms? |
Criterion-referenced testing |
|
|
What is criterion-measured testing? |
AKA Mastery tests. When is it possible to specify exactly what we are measuring the test is meaningful in its self (like an exam)
|
|
|
When are norms not needed |
1. When there is a 'standard' against which the measurecan be calibrated (e.g., metres, weight?)2. When there is an outside criterion against which themeasure can be calibrated (e.g., skills such as driving).
|
|
|
What are samples of norms called? |
Normative or standardization samples
|
|
|
What do norms have to be for a norm-referenced test> |
Appropriate and good. |
|
|
What are good norms? |
Are based on an appropriate standardization(normative) sample for the testing purpose
|
|
|
What are appropriate norms? |
Relevant to testing purpose
• obtained from a large enough sample • obtained using the same testing conditions across the sample |
|
|
What are some cautions for interpreting norms? |
-The normative sample is important. Representative of whole population, for specific sample. Can be out dated quickly. Same samples can have a high error. |
|
|
How do you sample norms> |
Systematic or stratified sampling to represent the whole population. |
|
|
How does sample size effect error? |
The larger the sample the smaller the error because error=SD/√N |
|
|
What is a good sample size? |
1000 good- 2000 excellent |
|
|
How do you decide on an appropriate normative sample? |
Consider testing purpose Cultural implications Is a cut-off score being used Norms should be relevant Report any issues with norms |
|
|
What are the scales of measurement? |
Nominal Ordinal Interval Ratio |
|
|
What is transforming data? |
Rescaling data to z-score, T-scores or Standard Deviation |
|
|
What is a z-score |
The transformation of a percentage or apercentile ranks
|
|
|
What are linear transformations? |
Transformations which keep the distribution of the scores the same (ie: adding a constant to each score). |
|
|
What are percentiles and stanines? |
Different ways of representing a normal distribution. |
|
|
What is reliability? |
The consistency of tests scores (not a general property). |
|
|
What is reliability about? |
Consistency
Dependability Replicability |
|
|
What are the types of correlation? |
Strong, moderate, weak Positive, Negative, Curvelinear |
|
|
What is a correlation coefficient? |
Anassessment of the degree to which onemeasure co‐varies with another measureof the same thing/person/etc
|
|
|
What is pearson's r? |
The average sum of the products of a pair of z-sores.
|
|
|
What is the equation for Pearson's r |
r = Σ(zx zy)/N
• meaning multiply each X z score by its corresponding Y z score – then add up results ‐ then divide by the number of pairs |
|
|
What type of correlation should be used for ordinal data? |
Spearman’s rho (ρ)
|
|
|
What is the Classical Test/true Score Theory> |
• One way to think of reliability is that a test score gives only an ESTIMATE of the “truth”
• Each time the test gives a “measure” it missesthe “truth” or has error |
|
|
What is the equation for error? |
Observed Score=Test Score + Error • X = T + e |
|
|
What are some factors which affect reliability? |
Test construction - item sampling
• Test administration – person, place andadministrator factors • Test scoring factors |
|
|
What human factors affect reliability? |
• Health• Fatigue• Motivation• Emotional strain• Test‐wiseness• External conditions of heat, etc
|
|
|
What are some admin factors which affect reliability? |
-Bias in grading/performance -Conditions of testing (adhering to limits etc) -Interaction with examiner that facilitates orinhibits performance |
|
|
What is Test‐Retest Reliability? |
Give Test A and Time 1 then give Test A at Time 2 tothe same people and correlate results.
• Correlation gives the measure of reliability |
|
|
What is a possible issue of Test-Retest? |
Carry-over effects • May not always be appropriate for psychological characteristics as… • characteristic may change with time |
|
|
What does test-retest measure? |
Gives a measure of temporal stability
|
|
|
What is Alternate or Parallel Forms Reliability? |
Give Test A – give Test B (an alternate form) to thesame people and correlate.
• Correlation gives the measure of reliability |
|
|
When can the administration of Parallel Forms be close in time? |
When carryover from one for to the other is not a problem when needing to minimise effect of time
|
|
|
What does Parallel/Alternative Reliability measure? |
Gives a measure of content consistency and (whendelayed) temporal stability
• Not always possible or easy to have two forms of atest |
|
|
What is Split-half reliability? |
• Correlate halves of Test A ‐ such as odd/even etcgiven to one group of people
• Gives measure of internal consistency• Many ways to split test in half • Result will reflect the degree of consistency of the twohalves of the test |
|
|
What are some issues with Split-half reliability? |
T test is half its length – scores will besmaller than for full test and so range of scores willbe smaller and so correlation will be smaller (range restriction)
|
|
|
What are the types of Internal‐consistency measures for reliability?
|
Kuder‐Richardson Formula 20 (KR20) and Cronbach’sAlpha.
|
|
|
What score does Kuder and Cronbach need to be consistent? |
Around .80 |
|
|
What are some issues with Kuder and Cronbach tests? |
-Only can test one construct
-provide only an overall score and gives noinformation on individual items. -Sensitive to test’s length, with longer testsgetting higher scores |
|
|
What are some facts about KR20Kuder‐Richardson Formula 20?
|
Devised 1937 Measures inter-item consistency Standard of tests with dichotomous items A split-half on all combinations of items |
|
|
What are some facts about Cronbach's Alpha?
|
Describes how well a group of items focuses on asingle idea or construct
Finds average correlation of items |
|
|
What does a high Cronbach's a tell us? |
hiIndicates that theitems are focusing on one construct, but doesn’tsay how well that construct is covered.
|
|
|
What is Interscorer Reliability?
|
Scores ‐ judges – observers ‐ raters• various measures are used
‐ depending on thedata set ‐ such as ordinary correlations, othertypes of correlations, and percentage agreement |
|
|
What is the Item‐Response Theory? |
This examines the responses to the items interms of their ability to discriminate betweenaspects of test‐takers …
|
|
|
How can you measure the reliability of individual scores? |
Standard error of measurement (SEM) gives an idea ofthe “test precision” for individual scores• The smaller the standard error the more precise the test
|
|
|
What is the equation for the Standard Error of Measurement? |
For z-score SEM = √(1‐α)
• For raw scores SEM = SD raw scores√(1‐ α) • Where α = reliability coefficient |
|
|
How can you work out if the score is meaningful? |
Standard error of thedifference between two scores
|
|
|
Testing reliability of compositescores
|
• Overall test scores are sometimes created byadding sub‐test scores
• The more highly correlated the individual subteststhe more reliable the composite score(i.e., combining tests is rather like lengthening |
|
|
What are the sources of error? |
• Test‐retest ‐
• Alternate form • Split‐half • Internal‐consistency |
|
|
How reliable should a test be? |
• This depends on many things ‐ use of test,expense, and so on.
• High reliability required when precisionimportant. |
|
|
What are the actors affecting reliability? |
1. Characteristics of people sitting test 2. Characteristics of test – including the length – Spearman‐Brown formula predicts the effect of lengthening the test 3. Intended use of scores 4. The method of estimation
|
