Double-blind study

Download Report

Transcript Double-blind study

An argument for global warming.
9th grade
Some high school
High school grad
Some college(no degree)
Associate's degree
Bachelor's degree
Master's degree
Professional degree
PhD
$16,322
$19,095
$25,081
$29,903
$31,358
$41,361
$50,704
$76,659
$71,541
X = driving credit score
Y = average incurred loss per policy
(1-worst 10%, 2-next lowest 10%, … , 10-best 10%)
1
$918
2
$846
3
$791
4
$707
5
$703
6
$681
7
$631
8
$584
9
$568
10
$558
Team X = payroll (millions of $) Y = wins in 2005 regular season
NYY
208.3
95
BOS
123.5
95
NYM
101.3
83
LAA
97.7
95
PHI
95.5
88
STL
92.1
100
SF
90.1
75
SEA
87.7
69
CHC
87.0
79
ATL
86.5
90
LAD
83.0
71
HOU
76.8
89
CHW
75.2
99
BAL
73.9
74
DET
69.1
71
SD
63.3
82
AZ
62.3
77
CIN
61.9
73
FL
60.4
83
MIN
56.2
83
TEX
55.8
79
OAK
55.4
88
WAS
48.6
81
COL
48.2
67
TOR
45.7
80
CLE
41.5
93
MIL
39.9
81
PIT
38.1
67
KC
36.9
56
TB
29.7
67
Cleveland best value for the money
How much $$ per win?
Yankess = $2.18 million
Tampa Bay = $440,000
Phila = $1.08 million
Nation X = TV's per capita Y = life exp. (years)
Angola
Australia
Cambodia
Canada
China
Haiti
Japan
Russia
UK
USA
Yemen
1/200
1/2
1/177
1/1.7
1/15
1/234
1/1.8
1/3.2
1/3
1/1.3
1/38
44
76.5
49.5
76.5
70
53.5
79
69
76
75.5
50
State X = avg. spending per student (in $) Y = mean SAT
Math score
AL
5,885
554
AR
5,278
550
FL
6,170
499
GA
6,929
489
LA
6,037
562
MS
5,175
551
NC
6,346
499
SC
6,631
488
TN
5,687
553
Graph this to determine if a correlation exists.
Size of sample is important for accuracy…
More data = better results!
Correlation; describes the degree of relatedness
between two variables
•Positive correlation; both variables move in
the same direction (both up OR both down)
Ex. IQ and test scores
Attendance and grades
School cell phone use and detentions
Negative correlation; variables move in
opposite directions.
Ex. Brushing teeth and number of cavities.
Life expectancy and smoking.
Children raised in homes with more
appliances tend to perform better in
school. Therefore, appliances improve
intelligence.
Correlation does not imply causation
There is a correlation between watching violence
on TV and violence in adolescence.
Can you conclude that violence on TV causes violent
kids?
NO. It could be a direct result of a third factor,
such as growing up in a violent home.
…but it can be used to make predictions.
=
education
Lifetime earning
potential
Height
(inches)
Head Size
(inches)
Gathering Data:
•Experiments
•Observations
•Case Studies
•Surveys
•Longitudinal Studies
•Cross-cultural studies
Samples:
•Small group of the tested population
•Must be representative of the larger population
•To avoid bias…
•Random Sample; individuals picked purely by
chance (every individual has an equal chance of
being selected)
•Stratified Sample; individuals deliberately chosen
from various sub-groups of the population.
Ethnicity, age, sex, political/religious affiliation, etc
1.Experiments
• Control the situation, environment
• Decrease the influence of outside factors
• Hypothesis; educated guess…establish the
foundation of the experiment
• Independent variable; the factor the
experimenter changes to see its effect
• Dependent variable; the factor that is measured
to see how it was affected by the independent
variable
A + B + C + D = X
Water
Light
Temperature
Soil
Growth
Hypothesis: If I allow administer more Light, then the plant will grow faster.
Independent variable:
Light (manipulated by researcher)
Dependent variable:
Growth (“result” of independent variable)
Constant: Temperature, water, Soil
Control Group: 12 hours of light
Experimental Group: 24 hours of light
Controlled experiment:
Hypothesis: Red light helps plants grow better than blue light
Independent variable:
Color of light
12 hrs/day
Dependent variable:
Height of plant
24 hrs/day
liter of water/day
degrees Celcius
soil
Every factor other than the independent variable should remain CONSTANT!
Experimental group: gets the independent variable
Control group: gets treated exactly the same as the experimental group
EXCEPT for the independent variable
Experimental results must be replicable. Results published in a journal
then reviewed and critiqued by peers. If results can be replicated and
the methods and conclusions accepted by the scientific community the
results are accepted as either supporting or not supporting the
hypothesis. You cannot really PROVE things in science!
If you increase the number of cigarettes you smoke, then you will increase your risk
for lung cancer.
Independent : ________________________________________________
Dependent: __________________________________________________
If you increase the amount of blood in the water, then you will increase the number of
sharks in the area.
Independent : ________________________________________________
Dependent: __________________________________________________
If you increase the amount of milk you drink, then you will increase the strength of
your bones.
Independent : ________________________________________________
Dependent: __________________________________________________
If you increase the number of hours you spend in practice, then you will increase the
number of free throw shots you will make.
Independent : ________________________________________________
Dependent: __________________________________________________
One trap to avoid in research:
Self-fulfilling prophecy
The researcher finds what they expect to find and ignore or
overlook other important facts, data, etc.
The researcher may do this unknowingly, which is why we
publish results
Does a researcher’s beliefs, attitudes, or feelings affect the
way he or she will interpret the results of an experiment?
If I work for an oil company or a utility company that uses coal,
will I be as likely to believe in global warming, or will I only
look at data that support what I already believe? (cherry
picking the data)
4 Steps to Self-Fulfilling Prophecy
1. Perceiver has expectations about how target will behave
2. Perceiver then behaves in a way that is likely to elicit the
expected target behavior
3. Target indeed behaves in a way that confirms perceiver�s
expectations
4. Perceiver (Objective Perceiver) sees predicted behavior
The Placebo effect
Solution:
The Double-Blind Study
vs
Single-blind study: the test subjects are “blind”, but the
researcher knows the Coke and Pepsi cups and could influence the
results with body language, voice inflection, eye movements, etc.
(intentional or not)
Double-blind study: both the test subjects and the researcher
are “blind”. If neither know the products in the cups, no improper
influence could affect the results.
2. Naturalistic Observation:
•To understand how the subjects
behave naturally
•Avoid disturbing or interacting
with the subjects
3. Case Studies:
•Intensive investigation of an individual
or group
•Usually focus on a particular problem
or experience
•Combine long-term observations
(psychological tests, behavior reports,
self-reports, etc)
•Lacks a control group for comparison
4. Surveys
• Most practical way to gather data on
attitudes, beliefs, and experiences from
a large group of individuals quickly and
easily.
1. Interviews; allow modification of the
question if needed
2.Questionnaires;
• Less time
• Results more uniform
• Eliminates influence of researcher
5. Longitudinal Studies
•Covers long periods of time
•Study and restudy the same group at
regular intervals to determine changes
in behavior
•Time consuming…good results may
take years to obtain and publish
6. Cross cultural studies
•Comparisons of the way people in
different cultures behave, think, feel,
etc
•Certain values appear to be universal,
but cultural differences do affect the
development of moral reasoning.
Coca-Cola spends more money on advertising ($2.9billion)
than Microsoft and Apple combined
Coke uses 300,000 tons of aluminum for its cans every year just for its
US operations.
That's equal to 17.4% of what the entire US aluminum industry
produces
The Coca-Cola brand is worth an estimated $74 billion: more than Budweiser,
Pepsi, Starbucks and Red Bull combined
Coke makes so many different beverages that if you drank one per day, it
would take you over 9 years to try them all
Coca-Cola sells more than 1000 kinds of juice drinks, including: Simply,
Minute Maid, Fruitopia, Hi-C, Fuze and Odwalla
Pepsi owns non-drink brands like Frito-Lays and Quaker Oats
The average American consumes 10lbs of sugar from drinking Coke every year