SIMPSON’S

PARADOX

GROUP 3

Presenters:
Aakanksha Agarwal
Heena Chandak
Pavan Puranik
Priyam Chakraborty
Richa Jairaj
Rohit Prakash
WHAT IS IT ABOUT ?

  1972- Colin B Blyth introduced the concept by the

name Simpson's paradox 

 Edward H Simpson described this phenomenon

 Simpson’s paradox is also know as “Yule-Simpson Effect”

WHAT IS IT ABOUT ?

 Simpson’s Paradox- Basically occurs when an association between

two variables is reversed after observing a third variable

 Incorrectly combined groups may lead to an improper weighting of

results

 This mistake is most common when dissimilar groups are combined

 A lack of care in controlling for Simpson's Paradox can lead to

faulty conclusions and inaccurate decisions
EXAMPLE 1
 My friends team and my team play a volleyball game and
each serve 40 shots.
 Who is the better at it:

My Team My Friends
Team
Serve in 20
Serve in 16
Serve out 20 Serve out 24
Total 40 Total 40
EXAMPLE 1
 But, which is the better team if you control for the distance of
the shot? Which team is performing better?

MY TEAM
RUNNING STANDING TOTAL
SERVE SERVE
SERVE IN 5 15 20
SERVE OUT 8 12 20

TOTAL 13 27 40

MY FRIENDS TEAM
RUNNING STANDING TOTAL
SERVE SERVE
SERVE IN 10 6 16
SERVE OUT 6 18 24
TOTAL 16 24 40
EXAMPLE 2
o 1999- University of Calizona at Los Phoenix (UCLP), Arizona
o Four graduate programs-
o English, Physics, Psychology, and Materials Science.

 393 men applied

 294 were admitted- admission rate of 74.8%

 444 women applied

 135 were admitted -admission rate of 30.4%

o Suggested sex bias favoring men in graduate admissions

o To track down the source of the bias, administrators ordered the
individual graduate programs at UCLP to report their admission
rates for men and women in 1999
 EXAMPLE 2
Below is a table giving the results of the investigation.
Note carefully which of the four departments admit men at higher rates
than women.

Men Women
Accepted Applied Rate Accepted Applied Rate
English 12 43 27.9% 48 130 36.9%
Physics 119 124 96.0% 8 8 100.0%
Psychology 7 60 11.7% 59 285 20.7%
Materials Science 156 166 94.0% 20 21 95.2%
294 393 74.8% 135 444 30.4%
ANALYSIS – EXAMPLE 2

 In our example the problem is, that the two departments that
have low admission rates overall also have high numbers of
applications from women and low numbers from men.

 In the two departments with high admission rates overall the

situation is reversed. Thus a large percentage of men are
admitted and a large percentage of women are not, even
though every department admits women at a higher rate than
men.
 OUTCOME OF EXAMPLE 2
 This baffled administration at UCLP. It was at a loss to explain
the situation

 Every department admits women at a higher rate than men, but

overall the university admits men at a much higher rate than
women

 This counterintuitive situation is a concrete example of Simpson’s

Paradox

 This paradox asserts that event A may be positively relevant to

event B in every block of some partition of the population and yet
be negatively relevant in the population as a whole (e.g., being
female is positively relevant to being admitted in every program
yet negatively relevant to being admitted overall)
IMPLICATIONS IN DECISION MAKING

 The practical significance of Simpson's paradox surfaces in

decision making situations where it poses the following
dilemma

 Which data should we consult in choosing an action, the

aggregated or the partitioned?

 Given the same table, one should sometimes follow the partitioned
and sometimes the aggregated data, depending on the story behind
the data with each story dictating its own choice
 CONCLUSION

Simpson’s paradox is a rare phenomenon! It does

not occur often. Thus statisticians must be trained
academically & ethically well enough to make sure that if it
has occurred they will detect and correct it.
THANK YOU