November 16, 2009
.Professor will be sending out on WebCT a list of content for Quiz 2 -
everything from the beginning of term until quiz 2 - see it in the e-
.Faculty of Health sent out a survey they have asked that you complete
.If you are interested in having things change do the survey.
Measurement and Data Collection Concepts
>Validity-does the test (study) measure what it is supposed to measure?
-e.g. chin-ups to test lower body strength
-Internal and External Validity
>Internal validity-is it correct to infer a cause and effect
relationship? Has researcher controlled all variables?
>Face [logical] validity-test appears correct
.This means on the surface does it appear that this is correct.
>Content validity rationale for each test item.
.When you conduct an experiment the experiment has test items.
.Do each one of those activities makes sense is there a rationale for
>Construct validity-test two extreme groups
.Honesty is an example of something that cannot be measured correctly.
.Sportsmanship is another concept that cannot be measured.
.If we take the example of honesty one would expect that people who are
in prison are dishonest and perhaps religious leaders might be very
.Therefore we test the two groups - one would expect the score for one
group to be higher than the other.
-Concurrent validity-compare established test to new one
(sometimes called criterion validity)
.You are getting the same type of measurement using a different type of
-Predictive validity-test score predicts future performance
.Example someone looked at your marks previously and decided that you
will be successful in university.
.Are these kinds of tests accurate? NO.
MCAP, LSAT, GRA etc.
Page 1 of5 >External validity-can the results be generalized to other settings?
-Was the sample representative of the population?
Two broad classes of External Validity-
-Generalize from sample subjects to similar population
.Professor said for example he takes approx 30-40 students and compare.
-Generalize from experimental conditions to other conditions.
>Reliability-refers to the consistency or repeatability of the data
-e.g. measure a board today and then one week later
.This would be considered good reliability.
>anytime a measurement is made some error introduced,
>any obtained score is composed of two elements, the true score component
and an error component:
>Xo = Xt + Xe
>where Xo is the observed score, Xt is the true score, and Xe is
>A test is administered twice to the same group of subjects.
>A correlation coefficient computed between the resulting two sets of
scores will serve as an indicator of the test's reliability.
.Example measuring the heights of various people in the class.
.Suppose we are measuring honesty and you score very high on it.
.Suppose you do another test and you score poorly - maybe this is not a
>The rest is split into two parts. Each part is scored and the resulting
two sets of scores are correlated.
>A knowledge test consisting of 50 items.
>The sum of the scores for the odd-numbered trials (items) and the scores
for the even-numbered.....(Professor changed slides)
>The even-odd split is usually adopted so that the effects of learning
from being tested and fatigue are equally represented in both sums of the
score for each individual.
>Thus two scores (an odd total and even total) are available....
(Professor changed slides)
Page 2 of5 .Suppose you are shooting a pistol - two targets shown.
Target 1 - Valid and Reliable
-shots shown are all close to the bull’s eye
Reliable BUT not Valid
.it is away from the bull’s-eye therefore it is an invalid instrument
that you are using.
Common Source of Error
>These are threats to validity
>Maturation-subjects change over time