Results of Poor Design

1
Results of Poor Design

• Inaccurate assessment of the prevalence of
  homosexuality based on human sexuality research
  by Kinsey, et al.

2
Complete Details inKinsey, Sex, and Fraudby
Reisman and Eichel
3
In 1988, health officials in New York reduced the
estimated number of male homosexuals in New York
City from 500,000 to 100,000. The larger number
consistently led to large overestimates of the
number of HIV-infected males in the city.
4
Where did the erroneous number of 500,000 come
from? It was based on erroneous estimates of
the frequency of homosexuality from the original
1948 Kinsey report on male sexual behavior.
5
Kinseys estimates for white males 10 are
homosexual for 3 or more years between
ages 16 and 55 4 are exclusively homosexual
throughout life (post adolescence) Better
estimates 2 and less than 1.
6
How did the Kinsey research arrive at estimates
that were 5 times too high? - Volunteer bias -
Disproportionate sampling from some groups
(strata) with high rates - Inappropriate
analytical methods - Misleading definitions -
Possibly, intentional fraud due to personal
bias.
7
Volunteer Bias

• Certain personality types are more likely to
  volunteer than others
• People with strong opinions are more likely to
  take surveys than those who are indifferent
• Those who agree with the views of a surveys
  designer are more likely to take part than are
  those who disagree

8
Disproportionate Sampling

• Kinsey seems to have sought out the
  unconventional
• 20 to 25 of the sample had prison experience
  and 44 of these had homosexual contacts while in
  prison
• may have sought out sex offenders
• 5 or more of his sample may have been male
  prostitutes

9
Disproportionate Sampling

• Goal not to estimate frequency of
  unconventional behavior, but to document the
  existence of unconventional behavior
• Sometimes asked interesting people if they knew
  of others that the team should interview

10
Is Fraud a Serious Possibility?

• Problems with Kinseys methods were reported to
  him by eminent statisticians through peer review
  processes
• Kinsey persisted in reporting biased rates of
  homosexuality based on seriously flawed methods

11
Social Consequences

• So what? Why would inaccurate data be a problem?
• Did overestimates of the AIDS epidemic do any
  real damage?
• Some sex education programs in schools actively
  promoted the false 10 number (e.g., Project 10 -
  L.A. Unified School District)

12
A Basic Question

• Should people seek truth, or should they seek
  results that please them?
• Should public policy be based on solid science?
• If not, who gets to decide?