You are on page 1of 16

test

reliability on
NRTs
What is NRT?
 Testthat determines a
student’s placement on a
(normal distribution) curve.

 'grading on a curve'.

(Kelly, Melissa)
Educators use norm-reference tests to
evaluate the effectiveness of teaching
programs, to help determine students'
preparedness for programs.

(Longsdon, Ann)
Statistical methods are used to
determine how raw scores will be
interpreted and what performance
levels are assigned to each score.
Many tests yield standard scores,
which allow comparison of the
student's scores to other tests.

(Longsdon,
Ann)
Reliability of
NRTs
Extent to which the results can be
considered consistent or stable.
 Measurement
Error or
Error
 Meaningful Variance
variance

 Variance  From
related to extraneous
the purposes sources
of the test
Measurement Error
Literature
 Variance due to environment
(location, amount of space, noise level, ventilation, weather, lighting)

 Variance due to administration procedures


(testing mechanics, directions, attitudes)

 Variance attributable to examinees


(physical characteristics, psychological, experience)

 Variance due to scoring procedures


(subjective nature= biases)

 Variance attributable to the test and test items


(technical part: clarity of print outs, format, number of items)
Reliability coefficient
(reliability estimates)
 Percent of systematic,or consistent,
or reliable variance in the scores on
a test
Three Strategies to
Estimate Reliability
 Test- retest reliability
 Estimate reliability of a test overtime

 Equivalent forms (parallel forms reliability)


 two different but equivalent tests

 Internal-consistency strategies
 Split-half method
 Split the test on odd-even numbers

 Spearman-Brown formula
 Cronbach α

 Kuder-Richardson Formulas
 K-R20
 K-R21
other types of
reliability

Reliability of Rater
Judgments

raters make judgments and


give scores (speaking &
writing)

Interrater Reliability
(with Spearman-Brown formula)
Intrarater Reliability
(with Spearman-Brown formula)

Getting two sets of scores


produced by the same rater for
the same group of students
and calculating s correlation
coefficient between those two
sets of scores
Standard error of
measurement
Estimate (a sort of) average of the
distribution of error deviations across all the students wh

You might also like