Science Final Exam 17 December 2013 Tuesday 18302030 In the Gymnasium 20 Questions All short answer 5 marks each Worth 20 of the course grade Deadlines Participation due 10 th of December ID: 432249
Download Presentation The PPT/PDF document "Review Part 3" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Review Part 3
ScienceSlide2
Final Exam
17 December 2013 (Tuesday)
18:30-20:30
In the Gymnasium
20 Questions
All short answer
5 marks each
Worth 20% of the course gradeSlide3
Deadlines
Participation due: 10
th
of December.Slide4
Causation ≠ Correlation
C
ausation does not imply correlation. If A and B are correlated there are several possibilities:
A causes B
B causes A
C causes A and C causes B
A and B are only accidentally correlatedSlide5
Common Cause!Slide6
From the Daily Mail
Lede
: “[new]
findings, published in the latest online edition of the journal Appetite, show the way we perceive tasty treats like chocolate cake is just as important as the calorie count when it comes to expanding waistlines
.”Slide7
From the Daily Mail
“They
recruited almost 300 volunteers, aged from 18 to 86, and quizzed them on their eating habits and whether they were trying to lose
weight. They
also asked them if eating chocolate cake made them feel happy or guilty.
The results showed 27 per cent associated it with guilt and 73 per cent with
celebration. When
the researchers looked at weight control 18 months later, they found those riddled with guilt had gained significantly more
.”Slide8
Probably Common Cause
Maybe people who eat unhealthily feel more guilty about eating chocolate. After all, they can see the harm they’re doing to themselves.
And maybe people who eat very well don’t feel guilty having chocolate. Slide9
Coincidence
In
1979, two researchers, Nancy Wertheimer and Ed
Leeper
, published an article
alleging
that the incidence of childhood leukemia was higher in Denver neighborhoods that were near electric power
lines.Slide10
The “Texas Sharp Shooter”
Suppose I stand in front of a barn. I have a machine gun with me, and I am blindfolded. I shoot wildly at the barn for several minutes.
Afterward, I walk up to the barn. I find a spot where three bullets are very close together, and I paint a target around them. “Look!” I say, “at what an excellent marksman I am!”Slide11
Power Lines and Cancer
The power lines study is just like this.
The researchers found places near power lines and looked at
all
the health problems anyone had in the area.
Of
all
the health problems what’s the chance that
one
is accidentally correlated with power lines?Slide12
Coincidence
High enough!
Later studies showed there was no relationship.Slide13
Example Study
“
A randomized controlled clinical trial of auricular acupuncture in smoking cessation
.”
Wu TP, Chen FP, Liu JY, Lin MH, Hwang SJ.Slide14
Example Study
“
A randomized controlled clinical trial of auricular acupuncture in smoking cessation
.”
Wu TP
, Chen FP, Liu JY, Lin MH, Hwang SJ.
Person who runs the lab,
Had the idea for the paper.Slide15
Example Study
“
A randomized controlled clinical trial of auricular acupuncture in smoking cessation
.”
Wu TP,
Chen FP
,
Liu JY
,
Lin MH
,
Hwang SJ
.
People who did the work
(Most work first).Slide16
RCT
“We
conducted a prospective, randomized, controlled trial using auricular acupuncture for smoking cessation in 131 adults who wanted to stop smoking. Thirteen subjects withdrew from the study and 118 subjects were included in the final analyses (mean age, 53.7 +/- 16.8 years; 100 males, 18 females
).”Slide17
RCT
We
conducted a prospective, randomized, controlled trial using auricular acupuncture for smoking cessation in 131 adults who wanted to stop smoking. Thirteen subjects withdrew from the study and 118 subjects were included in the final analyses (mean age, 53.7 +/- 16.8 years; 100 males, 18 females).Slide18Slide19
The Importance of Randomization
Improper randomization
procedures on average exaggerated effects by 41%.
This is an average result, so improper randomization often leads to exaggerations that are even larger than 41%.Slide20
“Garbage In, Garbage Out”
This means meta-analyses conducted on poor quality trials should not be trusted!Slide21
RCT
“The
treatment group (n = 59) received auricular acupuncture in Shen Men, Sympathetic, Mouth and Lung points for 8 weeks. The control group (n = 59) received sham acupuncture in non-smoking-cessation-related auricular
acupoints
(Knee, Elbow, Shoulder and Eye points). The enrolled subjects were then followed monthly for 6 months after stopping the acupuncture treatment
.”Slide22
Auricular AcupunctureSlide23
Sham AcupunctureSlide24
LOTR MemeSlide25
RCT
“
Between both groups before acupuncture treatment, there was no significant difference with regard to gender, mean age, education level, and mean values for the age at which smoking started, smoking duration, daily number of cigarettes smoked and nicotine dependent score
.”Slide26
Internet Polls
Internet
polls are not trustworthy
. They are biased toward people who have the internet, people who visit the site that the poll is on, and people who care enough to vote on a useless internet poll.Slide27
Representative Samples
The opposite of a biased sample is a representative sample.
A perfectly representative sample is one where if n% of the population is X, then n% of the sample is X, for every X.
For example, if 10% of the population smokes, 10% of the sample smokes.Slide28
Random Sampling
One way to get a representative sample is to randomly select people from the population, so that each has a fair and equal chance of ending up in the sample.Slide29
Results
“
At the end of treatment, cigarette consumption had significantly decreased in both groups, but only the treatment group showed a significant decrease in the nicotine withdrawal symptom score
.”Slide30
Fagerstrom Test
https://
outcometracker.org/library/FTND.pdf
Slide31
Statistically SignificantSlide32
P-Values
One way to characterize the significance of an observed correlation is with a p-value.
The p-value is the probability that we would observe our data on the assumption that the null hypothesis is true.
p = P(observations/ null hypothesis = true)Slide33
P-Values
Obviously lower p-values are better, that means your observed correlation is more likely to be true.
In science we have an arbitrary cut-off point, 5%. We say that an experimental result with p < .05 is
statistically significant
.Slide34
Statistical Significance
What does p < .05 mean?
It means that the probability that our experimental results would happen
if the null hypothesis is true
is less than 5%.
According to the null hypothesis, there is less than a 1 in 20 chance that we would obtain these results.Slide35
Only 5% chance it’s one of these.Slide36
Results
“
Smoking cessation rate showed
no
significant difference
between the treatment group (27.1%) and the control
group (20.3
%) at the end of treatment. There was also
no
significant difference
in the smoking cessation rate between
the treatment
group (16.6%) and the control group (12.1%) at the end of follow-up
.”Slide37
Statistically SignificantSlide38
Results
The decrease was real, the difference was not!Slide39
Conclusion
“
Our results showed that auricular acupuncture did not have a better efficacy in smoking cessation
compared to
sham acupuncture
.”
*Slide40
*P-Value Fallacy
Actually, the researchers found that
if
there was no relation between acupuncture and smoking cessation
,
then
the results they observed were typical (95% likely).Slide41
The Fallacy Fallacy
If you show that an argument is fallacious, you have not shown that it’s conclusion is false.
If you show that a relationship is not statistically significant in your experiment, you have not shown that it does not exist!!!Slide42Slide43
Effect Size
One NAEP analysis
of 100,000
American students found that
science
test
scores for men
were
higher
than
the test scores for women
, and this effect was
statistically significant
These results are unlikely if the null hypothesis, that gender plays no role in science scores, were true.Slide44
Effect Size
However, the average difference between men and women on the test was just 4 points out of 300, or 1.3% of the total score.
Yes, there was a real (statistically significant) difference. It was just a very, very small difference.Slide45
Effect Size
One way to put the point might be: “p-values tell you when to reject the null hypothesis. But they do not tell you when to care about the results.”Slide46
The Evidential Heirarchy
Systematic
reviews and meta-analyses
Randomized
controlled trials
Cohort
studies
Case-control studies
Cross sectional surveys
AnecdotesSlide47
Meta-Analysis
A meta-analysis is an analysis of analyses.
In clearer terms, it is a study that looks at lots of different experiments that have been conducted on the same problem, and tries to “put together” all of the findings.Slide48
BlobbogramSlide49
Jadad Scores
1. Do the researchers say that the study is randomized? Yes: 1 point, No: 0 points.
If yes, was the method described and is it appropriate? Yes: 1 point, No: -1 point.
2. Do they say it’s double-blind? Y: +1, N: 0.
If yes, described and appropriate? Y: +1, N: -1.
3. Did it describe the people who dropped out of the study? Y: +1, N: 0.Slide50Slide51
Churnalism
“19% of newspaper stories and 17% of broadcast stories were verifiably derived mainly or wholly from PR material, while less than half the stories we looked at appeared to be entirely independent of traceable PR.” – Lewis et. Al (on the course website)Slide52
Sadly…
Most science journalists don’t know anything about science (they have journalism degrees), and cannot tell an odds ratio from a risk ratio or a real scientific journal from a joke.
You
are better critical thinkers than the people who write the news!Slide53
Journalistic Embargoes
Even scientists are not above manipulating newspapers for undeserved fame and sometimes even money.
There is a practice whereby a scientist(s) will release a forthcoming article to the press, but not allow them to talk about it with anyone before it is published, and there is a press conference. This is called an “embargo.”Slide54
The Point of Embargoes
The idea is that when the press conference happens, the story will be big news: everyone will want to publish newspaper articles about it. So before the press conference, journalists will use the academic paper or study to write an article. But they can’t get quotes or other information from other experts, because they can’t talk about it. They can only present one side of the story, the researcher’s side.Slide55
The Link
One famous case involved researchers who had a fossil they wanted to present as “the Missing Link,” even though that makes no sense in evolutionary theory. They had a book deal and a TV special all lined up!Slide56
Not a Link
Of course, that sounds really exciting: “They discovered the missing link!!!”
But when other scientists actually had a chance to read the academic article, they found that the evidence that the fossil was a direct ancestor of humans not supported at all. Slide57
Self-Censorship
Another way reporters get the news wrong is that they don’t report it. Sometimes news angers those in authority, and journalists bow to their wishes.