We’ve all heard aphorisms
about the misuse of statistics. And we’ve all seen those pithy
sayings come to life as individuals or groups twist statistics to
their advantage. In contrast, one hopes that those in the media
would report accurate statistics and would be able to discern
accuracy from manipulation.
The basic problem is that
any statistic is merely a proxy for the state of the world it
attempts to measure. This is a necessary evil. In order to convey
information about a sophisticated topic, one must reduce a complex
world into some easily digestible form. The same process takes place
in a variety of realms — from economic models to scientific
experiments. The key in each context is whether what remains after
the simplification is still relevant to the more complex reality.
Many statistics are close
representations of the state of the world they attempt to measure.
Others fall short. For example, the poverty rate is the percentage
of people living in households whose income falls below the "poverty
line." The poverty line was chosen in the mid-1960s, varies with
family size and is annually adjusted for inflation. But what if the
poverty line was not chosen well initially? And to revisit a recent
debate, what if inflation is consistently exaggerated? These factors
could lead the poverty rate to mistate the level of true poverty.
Four other problems arise
since the poverty line is a measurement of reported cash income in a
given year. First, the Census Bureau can only measure whatever
income is reported. Underground economic activity leads to an
exaggerated poverty rate. Second, the Census Bureau only measures
cash income, omitting any non-cash government benefits (about 75% of
redistribution to the poor). Thus, the poverty rate is, to some
extent, a measure of dependency rather than low standards of living
per se. Third, the poverty rate only deals with income and ignores
wealth. As a result, some with low incomes but significant assets,
especially the elderly, are measured as poor.
Fourth and most important,
the poverty rate is only a snapshot of the economy in a given year.
(Ideally, one would have a motion picture — following people for a
number of years. Such data exist but are relatively costly to
calculate.) As a result, many are measured as "poor" early in their
lives — for example, graduate students — who will do well
financially later in life. Are the same people poor each year or is
there significant movement into and out of poverty? Unfortunately,
the poverty rate and income- distribution statistics tell us nothing
about the more vital question of income dynamics.
All of this explains some
otherwise anomalous statistics. For example, the poorest fifth of
all Americans consume almost two times more than their "income" and
consume as much as the average family in the early 1970s (adjusted
for inflation). And 41% of poor households own their own homes,
including 750,000 poor people who own homes worth more than
$150,000. The bottom line is that the poverty rate should be taken
with a shaker of salt.
Other statistics are
inadequate since they compare apples and oranges. Statistics on
income differentials between race or gender groups typically fail to
hold important variables constant - for example, levels and types of
education, age/experience, region, and family structure. In
addition, newspapers often report income comparisons between all men
and all women — even though the job market preparation and
experience of women has changed dramatically over the last 30 years.
Another example is the
tired comparison between the average worker’s real wages and average
CEO compensation. This one should be obvious — comparing one group’s
wages with another group’s compensation. (With a bad year in the
stock market, look for statistics manipulators to make this
transition by themselves!) In fact, looking at real wages over the
last 25 years is itself a dubious exercise since the proportion of
compensation devoted to fringe benefits has risen so dramatically.
(Higher marginal tax rates have made non-taxed benefits relatively
more attractive than taxed wages.)
All in all, statistics have
much to offer. The point is not to dismiss the poverty rate, but to
be well aware of its inherent flaws and to seek out data on income
dynamics. The point is not to downplay the issues of discrimination
and worker compensation, but to use honest comparisons. The media
have a responsibility to use statistics with discernment and
D. Eric Schansberg is
associate professor of economics at Indiana University (New Albany).