Convergent Validity And Discriminant Validity.ppt

Convergent Validity And Discriminant Validity.ppt

1. Ph.D. Course Work Topic:
Convergent Validity
And
Discriminant Validity
Presented By: Rupesh Solanki
Presented to: Dr.Jay BadiyaniSir
Department of Business Administration,
M. K. Bhavnagar University, Bhavnagar, Gujarat
1

2. Different Types of Measurement
Tests
•Inaresearch,soundmeasurementmustmeet
thetestsofvalidity,reliabilityandpracticality.
•Infact,thesearethethreemajor
considerationsoneshoulduseinevaluatinga
measurementtool.
•“Validityreferstotheextenttowhichatest
measureswhatweactuallywishtomeasure.”
2

3. 3
•Validityisthemostcriticalcriterionandindicatesthedegreeto
whichaninstrumentmeasureswhatitissupposedtomeasure.
Validitycanalsobethoughtofasutility.
•Inotherwords,validityistheextenttowhichdifferencesfound
withameasuringinstrumentreflecttruedifferencesamongthose
beingtested.Butthequestionarises:howcanonedetermine
validitywithoutdirectconfirmingknowledge?Theanswermay
bethatweseekotherrelevantevidencethatconfirmsthe
answerswehavefoundwithourmeasuringtool.
Construct Validityis an indication of how well a test measures the
concept it was designed to measure. There are two types of
construct validity:
(a)Convergent validity(b) Discriminant validity
Test of Validity

4. What Is Convergent Validity? Definition and
Examples
4
•Convergent validity is a subtype ofconstruct validity.
•Convergent validityrefers to how closely a test is
related to other tests that measure the same or similar
constructs. Here, aconstructis a behaviour, attitude, or
concept, particularly one that is not directly observable.
•Ideally, two tests measuring the same construct, such as
stress, should have a moderate to high correlation.
•High correlation is evidence of convergent validity,
which, in turn, is an indication ofconstruct validity.

5. •Example 1:In Convergent validity one is supposed to use
two different methods tocollect dataabout
happiness:observationand a self-reportquestionnaire. If
the scores of the two methods are similar, this suggests
that they indeed measure the same construct. A
highcorrelationbetween the two test scores suggests
convergent validity.
•Convergent validity is a bit more thin line, in that it
measures whether constructs that theoreticallyshouldbe
related to each other are, in fact,related to each other.
•For example 2, the scores of two tests, one measuring self-
esteem and the other measuring extroversion, are likely to
be correlated—individuals scoring high in self-esteem are
more likely to score high in extroversion. These two tests
would then have high convergent validity.
5

6. Convergent Validity
Vs.
Discriminant validity
•Together, convergent and discriminant validity
help us to establish construct validity. In research,
they are evaluated together because both must
be assessed in order to demonstrate construct
validity.
•Neither alone is sufficient, but it’s important to
remember that they are not the same thing.
•In short, while convergent validity focuses on
similarities, discriminant validity focuses on
differences.
6

7. •Convergent validityshows you whether two tests
that should be highly related to each other are
indeed related.
•Discriminant validityshows you whether two tests
that shouldnotbe highly related to each other are,
indeed, unrelated
•The idea here is that not only should a test correlate
with a similar test (i.e., measuring the same or a
related construct), but it should alsonotcorrelate
with dissimilar or unrelated tests (i.e., measuring
different constructs).
7

8. •For example, if there is no (or weak)
correlation between the scores of a test
measuring honesty and a test measuring the
favourite colour of participants, the test can
be said to have high discriminant validity. This
means that it only measures the construct it is
supposed to measure, and not other
constructs.
•Whenbothconditions (convergent and
divergent validity) are met, We can conclude
that a test shows evidence of construct
validity.
8

9. To establish the convergent validity of the test, we must do one of the
following:
•Compare your test scores against the results of at least one more
test measuring the same or a similar construct. For example, you
can measure your self-report questionnaire on introversion against
an existing one also measuring introversion.
•Compare the results of two different methods measuring the same
construct. For example, you can compare the results of an
observationand a questionnaireboth measuring introversion.
•In academic literature, individuals who exhibit narcissistic behavior
are also more inclined towards conspicuous consumption. The
expectation is that the scores of the two tests should converge.
•If the scores of the questionnaire correlate with the scores of the
other test, then your questionnaire has high convergent validity.
Conversely, if the scores of your questionnaire don’t correlate with
the scores of the other test, your questionnaire has low convergent
validity.
9
Examples of convergent validity and
discriminant validity

10. •Another Way, It can also test convergent validity by using two differentdata
collection methodsin your research design. To measure narcissism, you could also
design a personality testandastructured interview.Since the two methods are
measuring the same concept, their results should also correlate.
How to measure convergent validity?
•To measure the convergent validity of your test, you must demonstrate that there
is a positive correlation between measures of related constructs. In other words, if
you have two related scales, people who score high on one scale should score high
on the other as well.
•Correlation is estimated by acorrelation coefficient, such asPearson’sr, which is a
number that ranges between 1 and −1. This coefficient shows you the strength and
direction of the relationship between variables.
Correlation coefficient values can be interpreted as follows:
•r= 1: There is perfect positive correlation
•r= 0: There is no correlation at all.
•r= −1: There is perfect negative correlation
We can automatically calculatePearson’srin Excel,R, SPSS, or other statistical
software.
10

11. •Example,suppose you are interested in rumination as a personality
trait. Rumination is defined as obsessional thinking involving
excessive, repetitive thoughts that interfere with other forms of
mental activity.You have developed a new scale to measure that
trait.
•There is evidence in academic literature that people who tend to
ruminate also tend to show signs of neuroticism and anxiety. This
suggests that the construct of rumination is related to the
constructs of neuroticism and anxiety.
•To establish the convergent validity of your rumination scale, you
decide to compare it against scales measuring neuroticism and
anxiety.
•The researcher recruits asampleof 60 respondents and ask them
to fill in the three different scales. The researcher calculate the
correlation coefficients between the results of the rumination scale
and those of the other two assessments.
•The find out that your rumination scale correlates with the
neuroticism scale (r= 0.45) and with the anxiety scale (r= 0.60).
These values show that there is a strong positive correlation
between the three scales.
11

12. To Conclude:
•Keep in mind that the researcher will also need to
establish the discriminant validity of the scale before
drawing conclusions about construct validity. In other
words, researcher will also need to demonstrate that
there is either low correlation or no correlation at all
between suggested scale and scales of other unrelated
constructs.
•The researcher should first establish convergent
validity before testing for discriminant validity.
Together, these two tests allow you to assess construct
validity. Investigating relationships between constructs
helps us ensure a high correlation for convergent
validityand a low (or non-existent) correlation for
discriminant validity.
12

13. ****Thank You****
13