Causal Diagrams for Policy Analysis

1
Causal Diagrams for Policy Analysis

• Experiment in Spring 2006 on CMU students
  Task Take 1-page policy brief understand and
  analyze implications (esp. causal) Conditions
  - Text Only No diagrams - Diagram
  Correct diagram provided, then used - Tool
  Construct diagram, then use it

2
Text
Childhood obesity is now a major national health
epidemic. A number of facts are widely agreed
upon by the public and scientific community
doing exercise decreases obesity, and eating junk
food increases obesity. It's also clear that
people who watch more TV are exposed to more junk
food commercials. Parents for Healthy Schools
(PHS), an advocacy group which fought
successfully to remove vending machines from
Northern Californian schools, claims that
junk-food commercials on children's television
programming have a definite effect on the amount
of junk food children eat. In a recent press
conference, Susan Watters, the president of PHS
stated that "...if the food companies aren't
willing to act responsibly, then the parents need
to fight to get junk food advertising off the
air." A prominent Washington lobbyist Samuel
Berman, who runs the Center for Consumer Choice
(CCC), a nonprofit advocacy group financed by the
food and restaurant industries, argues that junk
food commercials only "influence the brand of
food consumers choose and do not not affect the
amount of food consumed". While Mr. Berman
acknowledges that watching more TV may cause
people to see more junk food commercials, he
remains strongly opposed to any governmental
regulation of food product advertising. Recent
studies by scientists at the National Health
Institute have shown that watching more TV does
cause people to exercise less.
3
Task
Question According the the NHI, will making
children watch less TV decrease childhood obesity?
4
Question According the the NHI, will making
children watch less TV decrease childhood obesity?
Childhood obesity is now a major national health
epidemic. A number of facts are widely agreed
upon by the public and scientific community
doing exercise decreases obesity, and eating junk
food increases obesity. It's also clear that
people who watch more TV are exposed to more junk
food commercials. Parents for Healthy Schools
(PHS), an advocacy group which fought
successfully to remove vending machines from
Northern Californian schools, claims that
junk-food commercials on children's television
programming have a definite effect on the amount
of junk food children eat. In a recent press
conference, Susan Watters, the president of PHS
stated that "...if the food companies aren't
willing to act responsibly, then the parents need
to fight to get junk food advertising off the
air." A prominent Washington lobbyist Samuel
Berman, who runs the Center for Consumer Choice
(CCC), a nonprofit advocacy group financed by the
food and restaurant industries, argues that junk
food commercials only "influence the brand of
food consumers choose and do not not affect the
amount of food consumed". While Mr. Berman
acknowledges that watching more TV may cause
people to see more junk food commercials, he
remains strongly opposed to any governmental
regulation of food product advertising. Recent
studies by scientists at the National Health
Institute have shown that watching more TV does
cause people to exercise less.
5
Diagram
- Larkin, Simon (1987) computational offloading
(via search) - Mayer, Moreno (2002) and
distributed processing - Zhang, Norman (1994)
Re-representation - Stenning, Oberlander (1995)
Graphical constraining - Cox (1999), Ainsworth
(2003) Promoting self-explanation
?
Question According the the NHI, will making
children watch less TV decrease childhood obesity?
6
Tool
iLogos Harrell (2005) Easterday, Kanarek,
Harrell (in press)
6
7
translation strategy
default strategy
English / high cost ER text gt diagram,
tool HCII / complement diagram tool gt
text Teachers /(Strong) Construction tool gt
diagram gt text
answer
policy text
Comprehension
8
Sample - 63 CMU undergraduates enrolled in HSS
courses - 30 minute on-line reasoning lesson
9
Training
10
Experimental Setup
Midtest (obesity)
Training
Pretest (environment)
Posttest (crime)
10
Performance
Learning
11
?
12
Tool (36)
Diagram (35)
Text (34)
13
?
14
Diagram (49)
Text (41)
Tool (40)
Diagram gt Text (p0.04)
15
?
16
Tool (67)
Diagram (62)
Text (56)
Diagram gt Text (p0.04) Tool gt Text (p0.04)
17
Conclusions

• 30 minutes of training with diagrams helps policy
  analysis

1-month of training helps statistical causal
discovery
4 years of training ? Thomas Richardson!