My goal when I chose this topic was to test a few of the "models" the media spout during the election. "The taller candidate always wins"; "Democrats are always elected when the economy is strong" and so on. I tried to seek out a database of such raw information (some conclusions would have been nice as well), but I couldn't find any such thing. I ran searches on several engines with several different approaches each (wild cards, etc.), but mostly I came up with out-of-date sites devoted to the most recent presidential election. Finally, I just ran a search for "candidate height weight" to try to get just exactly what I was looking for. I found something even better instead.
The first site that came up was from some
nut out there who had compiled in table format all of the information for
the Playboy Playmates going all the way back to the original herself, Marilyn
Monroe. In many of the earliest few years, a few items were missing
here and there, but for the most part I had bust sizes, waist, hip measurements,
age, height, and weight. From these variables plus the date of publication,
I could run an enormous number of statistical tests. For example,
given that this is a biological sample of a highly selected population,
I could check for correlations between various aspects of figure, age,
between height and weight, check for trends in measurements according to
year, etc. What follows is the report on those findings.
|Multiple Regression Models
On Playboy Models
*everything you ever wanted to know about Playboy Playmates, but never bothered to ask.
The Task: Find the story in the data.
From isolated (but frequent!) sampling of issues of Playboy Magazine, it is apparent that the overall "look" of the models featured as Playmates has changed over the years. It is my hypothesis that there is a relationship between the passage of time and the individual attributes (age, bust, waist, hips, weight, and height).
Biometric information about Playboy Playmates was extracted from a the web site "Playboy Playmate Data Statistics" and formatted in such a manner to perform statistical analyses. This information concerned the height, weight, age, bust, waist, and hips of each model (all of which were continuous variables). These data were recorded with the date of publication of each issue in which the model appeared.
Simple Linear Regression
Prior to developing a multivariate regression model, simple linear regression (SLR) was employed to determine the relationships between the year and the following variables age, bust, waist, hips, weight, and height.
|For age (p <0.0001, R2 =
year = 1952.95 + 1.24 (age, years)
|For bust (p <0.0001, R2
year = 2078.40 + -2.78 (bust, inches)
|For waist (p <0.0001, R2
year = 1919.24 + 2.60 (waist, inches)
|For hips (p <0.0001, R2
year = 2090.05 + -3.16 (hips, inches)
|For weight (p <0.0001, R2
year = 1980.38 + -0.004 (weight, pounds)
|For height (p <0.0001, R2
year = 1847.13 + 2.01 (height, inches)
|The Correlation Matrix
A correlation matrix was generated to determine the magnitude of the relationships between the predictors.
Both the Stepwise Forward and Backward and the Maximum R2 methods yielded the same highly significant (p < 0.0001, R2 = 0.4807) model.
These scores indicate the relative (not the absolute) contribution of each variable to the model.
One unexpected finding was the complete absence of a significant change (p = 0.9510) in the weight of the Playmates over the history of the publication. This can be explained by the highly statistically significant (p < 0.0001) contradicting influences of an increase in height and waist line with declining bust and hip measurements. In effect they grew taller and less shapely over the years.
Kachigan, Sam Kach. 1991. Multivariate Statistical Analysis: A Conceptual Introduction. Radius Press, New York.
Appendix: SAS Program and Data Set
The raw data set, SAS program, and SAS output for the analysis on can be found on this page.
|Back to the index|