Week 3 Norms textbook Flashcards
Define norms
tables of the distribution of scores on a test for specified groups in a population that allow interpretation of any individual’s score on the test by comparison to the scores for a relevant group
what is a non linear transformation?
a transformation that preserves the order but not the equivalence of disturbance of the original scores. eg bunched together. Example is percentiles
What did Weschler do in 1955?
Transformed z scores so no negative numbers. In developing adult int scale (now WAIS-V) he expressed an ind’s score as a z score and transformed the distribution so the mean was 100 and the SD was 15. Gives us a deviation IQ, to compare metric with peers.
Standardised score
a score based on a z score, but with a particular mean and SD for a particular purpose
T score vs sten score
T = a score standardised so the mean is 50 and SD is 10 Sten = Mean is set at 5.5., SD at 2 (The 16PF Questionnaire uses this scoring method.)
Describe the different ways of expressing test scores
- Item total = when an item can be right or wrong, 1 for correct, 0 for incorrect.
- Using raw score totals (summing correct answers, need to be norm referenced or criterion referenced to give meaning)
Explain the difference between norm referencing and
criterion referencing in interpreting test scores
Criterion-referenced tests:
“…the task itself is a yardstick (criterion) to which performance is referred, raw score often has meaning in itself
e.g., reading test, driving test
Norm-referenced tests:
»Raw score is compared to “the average score (or nor
m) of a representative group of people similar to the person being tested”
»more common in psych testing, e.g. see how others performed in a school test
How to norm reference
express the raw score total in terms of its position in a distribution of raw scores. Top end of dist = person better than most.
Describe the different methods for transforming scores
Linear >z scores »standardised scores »T scores »stenscores
Non-linear transformation
»percentiles
How to convert Standard scores to z scores
Subtract the mean from the score
»Then divide the result by the SD
Normal Curve
A statistical distribution that is symmetrical about the mean –half the scores below, half the scores above
Percentile
Nonlinear transformation of the z score
»Don’t confuse it with the percentage correct!
»Percentile = % of cases that lie below it
IQ score of 130 is at the 98th percentile, thus most people score below that.
When is it useful for norms to be age based?
When characteristics you are measuring vary across the lifespan. Other tests might compare again clinical or
community samples of adults, for example
What happens if comparison is not made to appropriate group when trying to norm reference?
The transformation fails to covey meaning
Why is a percentile transformation not linear?
Scores that are an equal number of percentiles apart are not necessarily an equal distance apart in the raw score distribution. Need to bear in mind where the percentiles are on the percentile scale. (We could find a fixed 10 % of scores in the middle, but 10% at the tails might need to cover more distance.)
What does the test manual often have?
a table of percentile equivalents for all possible raw scores on the test
What happens when a distribution of test scores departs from the normal distribution?
Some test makers will force the distribution into a normal form. Easiest way to do this is use normalised standard scores. Enter the tables with percentiles (calculated from raw score distribution) and reads off z score equivalents. This will be nonlinear
What is the stanine? (4)
1) A variant of the percentile is the stanine scale.
2) Score on a nine-point scale with the points set in terms of percentiles.
3) Stanine distribution Mean = 5, SD ~2. !st stanine = up to 4, 2nd = 4-11th percentile, 3rd = 11-23rd percentile.
4) Stanines is non-linear transformation
If for children in grade 6 the median score is 17, what is the raw score of 17 mean?
The age and grade equivalents rule means that the norm is the median, so if a child scored 17, this would be equivalent to being a grade 6.
- critisiced because kids have difference types of learning etc
Considerations for norms
- Check source of norms
- Are norms relevant to this situation?
- If concerned about country of origin, what is known
about susceptibility of test to cultural differences? - How is the test result being used? (cut-off scores determining course of action?)
- Is it possible to check in another way against another test with appropriate norms? (beware small samples)
- Explain in your reports if there are norm issues and that results should be interpreted with caution