2. Short Story
• Ahmed, A young man accused by robbing a supermarket.
arrested by the police and charges where pressed by the attorney
general.
• attorney general is claiming that Ahmed robbed the supermarket
• Ahmed heirs an attorney to defend him. and prove him innocent
• Two possibilities:
• he robbed it
• he didn’t
But whats the plan, How can we now the truth
3. In Court
• In order for the Attorney general to
lock him away he needs to prove
him “Guilty”
• The Attorney Genaral needs to
prove that Ahmed is not innocent
• In Order for Ahmed to be set free
his lawyer needs to Stop him from
doing so
• The lawyer doesn't need to prove
that Ahmed is innocent
4. Lets Think about it
• Hypothesis = Claim
• Attorney Genaral is claiming that
Ahmed robbed the supermarket
• His Hypothesis is “Ahmed Robbed the
Supermarket”
• He is Trying to prove it by Rejecting the
idea that “ Ahmed did not rob the
Supermarket
• if he dose succeed. we will by left with
one conclusion that Ahmed did rob the
supermarket
• If he doesn't succeed. The two
possibilities are still their. And we can’t
send Ahmed to jail because we are not
sure
5. Ahmed
This is what Robbed The
Supermarket
Did Not rob the
Supermarket
the Attorney
general
wants
6. Ahmed
This Is what Robbed The
Supermarket
Did Not rob the
Supermarket
Ahmed’s
Lawyer • You Can not convect a person unless your sure
wants • Its better to send a guilty man free, than or send an innocent
man to jail
• Benefit of the doubt
7. Inferential Statistics
• I’m Assuming that i will score a full mark in the
USMLE exam
• This assumption has a very rare probability
• Then this assumption is not true
8. Lets put it into scientific terms
• If you want to prove something, you
need to formulate a hypothesis
• Your assumption (hypothesis) called
the (Alternative Hypothesis) H1
Rare Event Rule for Inferential Statistics
• The opposite called the (Null
Hypothesis) H0
• You Try to prove the the null hypothesis
only happens very rarely
9. Hypothesis
This is what Alternative
Hypothesis (H1)
Null Hypothesis
H0
the Attorney
general
wants
10. Hypothesis
This is what Alternative
Hypothesis (H1)
Null Hypothesis
H0
the
Investigator
wants
11. Hypothesis
THA is Alternative
Hypothesis (H1)
Null Hypothesis
H0
Superior to
Bipolar in THA is
True
There Is No
NOF
Superior Difference
12. Example
• The Journal of Bone and Joint Surgery
2010;92:792.
• Assuming there is “deference Between
Gamma nail and DHS”
• H1 : No difference
• H0: there is difference
• Conclusion : “They are the Same”
13. Hypothesis
Alternativ Null
e Hypothesi
Hypothesi s (H0)
s (H1)
There is No There Is a
difference difference
Between between
Gamma and Gamma
DHS and DHS
14. Inferential Statistics
• You Can not prove something to be true, But you can disapprove something by
finding an exception
No, I
All swans found a
are white black
one
15. Example
• Wrong Hypothesis
• Wrong Results
• Wrong Conclusion
• Wasted 8 years of hard Work
This is the JBJS after all
16. Not Everything you
Read is Correct
• We ended up responding
to JBJS pointing out this
huge fault
17. • Rare Event Rule for Inferential Statistics: If,
under a given assumption, the probability
of a particular observed event is
exceptionally small, we conclude that the
assumption is probably not correct.
Inferential
Statistics
18. How rare is rare?
• The is a pre-determined value.
• The probability of the H0 being true
• It could be any number
• In clinical studies its 0.05
• in pharmaceutical studies its 0.01
• Some times its call the critical value
• its a cut-off for making a decision about the Null Hypothesis
19.
20. Probability Value
• What is the probability that of getting this result by chance?
• What is the probability of getting this Sample that show this result?
• You could be unlucky and have a very unfortunate sample?
• Or that is the real probability of getting this set of data
21. P-Value, Is That Enough
• How sure are you?
• If we repeat the experiment 100 times. Each time
taking a sample, will we have the same P-Value?
23. Confidence Interval at 95%
• Your a Resident sitting for the promotion exam
• Your score was 70%
• Take the test again for 100 times
• Scores ranged from 60-80%
• Now I can say That
• If you take the exam 100 times, I’m 95%
Confidant that you will score any mark between
60-80
• In other words 70 -10, +10
24. Confidence Interval at 95%
• An experiment has been done comparing 50
subjects
• Cemented THA vs non-Cemented THA
• Rate of Failure 20% vs 10% after 10 years
• P-Value 0.04
• Looks Good, Doesn't Are You ?
How Sure it ?
What is your confidence Interval?
25. Confidence Interval at 95%
• If Your repeat this experiment 100 times
• Will you have the same P-Value?
• CI 95% turned to be from 0.1 - 0.003
26. Confidence Interval at 95%
• I am 95% sure that your score in the promotion exam will be from
0-100
• With a P-Value of 0.05
• I’m 95% sure that your score will be from -10 to 50
• With a P-Value of P 0.000000001
• I’m 95% sure that your score will be from 70-75
With a P-Value of 0.1
27. Confidence Interval at 95%
• In any statistical test there is a C.I. for any value
• The P-Value has no meaning without a CI
• If there is no CI someone is hiding something
28. Confidence Interval at 95%
• There is a CI for everything
• Gluco check reader has a CI -5 +5
• You got a reading of 7
• That means if you repeat the reading again it could be 2 or 12
• repeat it 100 times results will fall between 2-12 95% of the time
29. Power
• The ability of the sample to detect a deference when its there
• Power of 80% is accepted
• That means:
• That this sample can detect the intended difference if it exists
80% of the time
30.
31. Heard About Type1, Type II Error
• Type I:
• You Reject the Null. when in fact its true
• Probability of Having this Error is 0.05
• The call it the Alfa value
• Sending an Innocent guy to Jail
• Type II
• Failing to Reject the Null. when in fact false
• Probability of having it is 20%
• Called Beta Error
• Letting a guilty man lose
33. Why Do we Need One?
• Medical Research Needs approval
• Research costs money
• Clear plan on:
• What you want to do?
• Why You want to do?
• How you plan to do it?
34.
35. Before You Write
• Start with a novel idea
• What research is funded by institute and program
of interest?
• passes the “So what ?” Test
• Gather your software
36. Before You Write
• Get to know the review criteria
• Significance: Scientific & practical importance; Impact
• Investigators: Expertise, training, accomplishments
• Innovation: Challenge/shift current research/practice
• Approach: Soundness of overall strategy, methods, analyses.
anticipate problems, address risks
• Environment: Institutional resources, equipment, access to
special populations
• Reviewers: Who are they?
38. What Is Known?
Gap
Goal / Propose
Guiding Hypothesis
Specific aims
IMPACT/EXPECTED OUTCOMES
39. Skeleton
• Getting Started
• Aims and Expected Outcomes
• Background
• Significance and Innovation
• Research Plan, Methodology, Timing
• Budget and its Justification
• Benefits of Your Research
40. Introduction
• What is this proposal about and what is its
relevance for health .
• What do we know about this issue?
• What is the important gap in knowledge that you
need to fill?
• Why do you need to fill this gap?
Why is it a significant problem worthy of funding to address? E.g., Why
is it a barrier to progress in the field? )
41. Introduction
• Critical evaluation of literature as it relates to proposed work and
gap to be filled.
• Cite key articles, update
• Discuss any relevant controversies and how project will resolve
• Provides rationale for the gap(s) you have identified, leads into your
main hypothesis & preliminary studies
• Significance: Expand on importance and impact mentioned in
Specific Aims
• Why is work significant? Specific benefits of the knowledge to be
obtained?
• How will these benefits:
• fit with institutes mission?
42. Aims
• Main hypothesis:
• What is it? Why? (Is it derived from preliminary
findings? Is it the best fit given the literature?)
• Justification for the Proposed Research: Why is it
important to do this particular research? How will
it allow you to move to the next step?
43. Specific Aims
• Lay out each specific aim & hypothesis for each.
• Should follow closely from your overarching
hypothesis.
• Hypothesis-driven
• Mechanism/process-oriented rather than
descriptive
44. Specific Aims
• What will your work buy us?
• Innovation
• Expected outcomes
• What impact will your work have on the health-
related problem and the field?
46. • Create a writing
timeline.
• Be realistic!
• 4-6 months (!!!)
47. Design & Methods
• Describe how you will carry out
the research
• Must relate closely to your
specific aims
• Include details for specific
methodology and why the
chosen method is the best to
accomplish your goals.
• Organize by aim.
• Troubleshoot: How will you
avoid or handle potential
problems? Alternatives?
48. Design & Methods
• Statistical strategy (justify use
of specific analytic techniques,
power analysis)
• Remember to include
timetable for project
• Key point: Justify everything!
49. Budget
• Direct Costs
• can be specifically documented, eg:
• salaries;
• operating expenses (printing, consumables);
• travel (kilometerage, fares, lodging, sustenance);
• equipment, ect
50. Budget
• Indirect Costs
• real costs that cannot be easily identified as specific to a
particular project, eg:
• utilities;
• maintenance of space and equipment;
• security;
• computer services;
• legal services;
• accounting services;
• payroll services.
51. Budget Justification
• Justify every item. Do not merely restate proposed
expenditure.
• Explain why the project could not proceed/would
not be successful without these items
53. Authorship
• Criteria for authorship:
• Substantial contributions to conception and design, acquisition
of data, or analysis and interpretation of data.
• Drafting the article or revising it critically for important intellectual
content.
• Final approval of the version to be published.
54. Publishing
• Its your proposal combined with your results
• You should of had an Journal or conference in mind while in the
designing phase
• You will be rejected
• Dose a fewer Authors mean better paper?
• Target High Impact journals
55. Be Correct (in content & details)
• Instructions
(formatting, page
limits, font, margins)
• Sections in prescribed
order
• Proofread – eliminate
typos, grammatical
errors, etc.
• Proofread again…and
again