Transcript Standardized Testing (1)

Standardized Testing (1)
EDU 330: Educational Psychology
Daniel Moos
Standardized testing: Warm-up
activity
Think about a time when you took a standardized test
(e.g., SAT, ACT) and reflect on the following
questions:
1)
2)
3)
4)
When you received your score for this test, did you
know how to interpret the score?
Did you feel as though the test was an accurate
assessment?
How should standardized tests be used?
Is there a place for standardized tests in education?
Why or why not?
Validity, continued
Type
Description
Content validity
Ability to appropriately sample the
content taught and accurately
measure the extent to which students
understand the content; i.e. the
extent to which test content is
consistent with curriculum objectives
Predictive validity
Ability to gauge a learner’s future
performance; determined by
comparing test results with other
measures (i.e. GPA)
Construct validity
Logical connection with the area it is
designed to measure; determined by
professional assessment
Classroom Assessment:
Validity & reliability
Statistics and standard scores
(1)
Evaluating results of standardized tests through descriptive
statistics (Measures of central tendency):
(1) Mean:
Average score
(2) Median:
Middle score in the distribution
(3) Mode:
Most frequent score
(4) Range:
Distance between top and bottom score in distribution
Statistics and standard scores
(2)
How should you use mean, median, mode, and range to interpret
the results?
70, 80, 80, 80, 90
Mean:
(400/5) = 80
Median:
80
Mode:
80
Range:
(90-70) = 20
Statistics and standard scores
(3)
How should you use mean, median, mode, and range to interpret
the results? (continued)
20, 80, 80, 80, 90
Mean:
(350/5) = 70
Median:
80
Mode:
80
Range:
(90-20) = 70
Statistics and standard scores
(4)
How should you use mean, median, mode, and range to interpret
the results? (continued)
60, 60, 80, 100, 100
Mean:
(400/5) = 80
Median:
80
Mode:
60, 100
Range:
(100 – 60) = 40
Statistics and standard scores

(5)
Standard Deviation (SD):
Statistical measure of spread of scores
 If data points are close to mean, then SD ≈ 0
 If many data points are not close to mean, then SD >
Formula Example on board
Example: Mean of scores is 40 and standard deviation of 4

28 32 36 40 44 48 52
0% 2% 16%
84% 98% 100%
50%
Statistics and standard scores
(6)
Evaluating standardized tests through descriptive statistics,
continued

Standard Error of Measurement



Due to measurement error, scores represent an
approximation of student’s “true” score
Example: Dan obtains a score of 54 on a standardized test. If
the standard error of measurement is 3, then the range of
Dan’s true score would be 51 to 57.
Normal Distribution:

Mean, median and mode are (roughly) equal; scores
distribute themselves symmetrically in a bell-shaped curve
Statistics and standard scores

Raw Scores:


What do they really mean?
Percentiles (also call Percentile Ranks):


≠ percentages
Rankings may not be equal; example:




(7)
raw score of 58 = 90th percentile
raw score of 56 = 80th percentile
Raw score of 54 = 60th percentile
Stanines:




• Grade equivalents: Compares
scores with particular age group
(1st digit = grade, 2nd digit =
month); Grade equivalent for
total reading of 8.4 means __ ?
Range: 1 to 9
Stanine 5 = center of distribution
One Stanine above/below = +/- 0.5 SD
Stanine 6 = +0.5 SD; Stanine 4 = -0.5 SD; Stanine 8 = __SD
Statistics and standard scores
(8):
Ending Exercises (1)
A superintendent from New Hampshire says:
“Standardized testing has its place -- but
standardized testing should only be one piece of
the picture of a child's performance or a
school/district's performance at a given time. It is
essential to educate ourselves and our
communities about what standardized testing
measures -- and what it does not measure….
I like the analogy that standardized testing is a
snapshot that is one entry into an album about a
child or school”
Ending Exercises (2)
A high school science teacher from Washington, DC
says:
“I think they can limit the teaching environment because
you end up teaching to the test. I think they can
ultimately narrow the curriculum….
They do provide a basis, are consistent because the scorer
cannot be biased since there is only one answer (that is,
if it is multiple choice!)….
I have always created my own tests, never have I used a
‘book test’...I feel this way my students have the
opportunity to demonstrate what they know through
short answer, longer essays, etc.”
Ending Exercises (3)
A high school senior from Washington, DC says:
“ I don’t know. I guess they are important, but I don’t think
one test can really tell you how good a student I am…or
even tell you how much I have learned. I just get real
nervous when I take them because they are important, but
it is kind of a one shot deal. I don’t think that is very fair.”
Ending Exercises (4)
(1)

(2)
In your opinion, what are the pros and cons in using
student performance on standardized tests to evaluate
teachers?
Please reference at least one concept we have talked
about today (e.g., functions such as assessments and
evaluation; statistical issues such as central tendency,
validity)
The correlation between SAT and college grades is .42
(Shepard, 1993). Why do you think there is not a stronger
positive relationship between these two?