Newsroom math
Prof. Steve Doig
Cronkite School, ASU
Journalists hate math






Definition of journalist: A do-gooder who hates math.
“Word person, not a numbers person.”
1936 JQ article noting habitual numerical errors in
newspapers
Japanese 6th graders more accurate on math test
than applicants to Columbia’s Graduate School of
Journalism
20% of journalists got more than half wrong on 25question “math competency test” (Maier)
18% of 5,100 stories examined by Phil Meyer had
math errors
Bad examples abound





Paulos: 300% decrease in murders
Detroit Free Press (2006): Compared
ACS to Census data to get false drop in
median income
KC Star (2000): Priests dying of AIDS at
4 times the rate of all Americans
Delaware ZIP Code of infant death
NYT: 51% of women without spouses
Common problems






Numbers that don’t add up
Making the reader do the math
Failure to ask “Does this make sense?”
Over-precision
Ignoring sampling error margins
Implying that correlation equals
causation
Dangers of journalistic
innumeracy




Misleads math-challenged
readers/viewers
Hurts credibility among math-capable
readers/viewers
Leads to charges of bias, even when
cause is ignorance
Makes reporters vulnerable to being
used for the agendas of others
The bad news…

To be a good journalist, you
MUST
be able to do math
The good news!


…it’s grade school math!
None of this stuff:






Calculus
Geometry proofs
Base-12
Venn diagrams
Ballistics
Etc….
My office bookcase
Sarah Cohen’s tips





Keep the digits in a paragraph below 8
Memorize common numbers on your
beat
Round off – a lot
Learn to think in ratios
Envision your dream number, then
calculate it
Newsroom math crib sheet
Comparing numbers






Difference
Percent
Percent difference
Percentage change
Millage
Per capita
Difference

Okay, I won’t insult you….
Percents

To get X% of Y:



Turn X% into a decimal, then multiply by Y
20% of 90 = 0.2 * 90 = 18
130.5% of 45 = 1.305 * 45 = 58.7
Comparing X and Y




X is what percent of Y?
X is X/Y of Y
Then multiply X/Y by 100
5 and 8:



5 is 5/8 of 8
5 is .625 of 8 (or 62.5%)
8 and 5


8 is 8/5 of 5
8 is 1.60 of 5 (or 160%)
Comparing NEW and OLD




Percentage change!
(NEW/OLD – 1)
Or: (new – old)/old
€8 million this year, €5 million last year



(8/5 – 1) = 1.6 – 1 = 0.6 = 60%,
So the budget has increased 60%
€5 million this year, €8 million last year


(5/8 – 1) = 0.625 – 1 = - 0.375 = -37.5%
So the budget has decreased 37.5%
Remember PEMDAS!

Order of algebraic operations:






Parentheses
Exponents
Multiplication
Division
Addition
Subtraction
Compare X and Y (% difference)





X is (X/Y – 1) MORE/LESS than Y
Use MORE THAN if the answer is positive
Use LESS THAN if the answer is negative
8 & 5: 8/5 –1 = 1.6 – 1 = 0.6 = 60%, so 8
is 60% more than 5
5 & 8: 5/8 –1 = .625 – 1 = -0.375 = -37.5%,
so 5 is 37.5% less than 8
Beware of base changes



Newsroom budget of €1 million grows
by 10% one year to €1.1 million!
Next year, recession, so boss has to cut
10% from budget
Result: €1.1 million – 10% of €1.1
million = €990,000
Beware of small bases


Easy to get big percentage change
when you start with small values
Population 2000: 1,000



Population 2010: 1,500
Percentage change: +50%
Population 2000: 1,000,000


Population 2010: 1,100,000
Percentage change: +10%
Property tax millage




Mill = €0.001 = 1/10th of a cent per €1
Change “mills per dollar valuation” into
“euros per €1,000 valuation”
Calculate tax based on “typical” value,
like a €100,000 home
Example: Tax rate of 8 mills

€8 per €1,000, or tax of €800
Rates




Number of events per some standard
unit (per capita, per 100,000, etc.)
Crime rates, accident rates, etc.
RATE = (EVENTS / POPULATION ) *
(“PER” Unit)
Use to compare places of different size
Calculating rates




RATE = (EVENTS / POPULATION ) *
(“PER” Unit)
If there were 320 murders in a
population of 1,937,086, what is the
murder rate per 100,000
320 / 1937086 = 0.0001652…
0.0001652 * 100000 = 16.5 murders
per 100,000 population
Consumer Price Index


Used to correct for inflation
Get the CPI at http://www.ine.pt
Price Now
Price Then
=
CPI Now
CPI Then
Using the CPI







CPI in 2010 = 109 (base year 2005)
CPI in 1990 was 52,9
Gasoline in 1990 was €0,68 per liter
X / 0,68€ = 109 / 52,9
X = (109 / 52,9) * 0,68€
X = 2,06 * 0,68€ = 1,40€
Gas in 1990 cost the equivalent of €1,40 per
liter in 2010 euros
Newsroom Statistics



Mean (Average): Add the values, then
divide by number of values
Median: Sort the values, then find the
middle one
Mode (rarely used): The most common
value
American baseball salaries
in 1994 strike year



Mean (average): $1.2 million
Median: $350,000
Mode: $100,000
Weighted average


Don’t average averages
Example:




Teacher average: €27 000
Janitor average: €15 000
Principal average: €65 000
Simple average: €35 667
Weighted average (continued)





Teachers: 10 000 x €27 000 = €270,0m
Janitors: 2 000 x €15 000 = € 30,0m
Principals:
500 x €65 000 = € 32,5m
Sum:
12 500 people
Weighted average: €26 600
(not €35 667)
€332,5 million
Public opinion surveys



Census vs. survey
A random sample is necessary
Size of the population being sampled
doesn’t matter, only sample size
matters
Sampling error



Rule: The bigger the sample, the smaller the error
Sampling error = 1/N
 N=100
1 / 100 = 1/10 = +/-10 pts.
 N=400
1 / 400 = 1/20 = +/- 5 pts.
 N=900
1 / 900 = 1/30 = +/- 3.3 pts
Other sources of error
Estimating crowds


Beware the “official” estimate
Better method:




Estimate the area in sq meters (L x W)
1 person/meter in a loose crowd
Divide by 0,75 for a tighter crowd
Account for turnover?
Newsroom math bibliography




“Numbers in the Newsroom”,
by Sarah Cohen, IRE
“Precision Journalism (4th edition)”,
by Phil Meyer
“Innumeracy”, by John Allen Paulos
“A Mathematician Reads the
Newspaper,” by John Allen Paulos
Preguntas??