Transcript Open Review Day, The First Hourly

Preparing for the First Hourly
Summer Term
Summer Term Course Structure
• Probability and Design Issues 
• Descriptive Statistics, Confidence
Intervals and Hypothesis Tests
Course Objectives: First Half, Summer Term
Probability
I.1. Conduct simple experiments in sampling and compute sample frequencies (counts
and proportions).
I.2. Understand the principles of frequentist probability theory.
I.3. Understand the relationship between sample proportions and probabilities.
I.4. Compute probabilities using the Additive, Multiplicative and Complementary rules.
I.5. Understand the concept of the random variable, and how probabilities are computed
for random variables.
I.6. Understand the implications of observing events with small probabilities (rare
events).
I.7. Compute and interpret conditional probabilities.
Design Issues
I.8. Sketch clinical trials.
I.9. Spot faults in sketches of clinical trials.
I.10. Spot faults in sketches of random sample surveys.
Problem Tasks: I
Bowls, Urns, Dice
Estimate Probabilities Using Samples from Bowls, Urns, Dice
Compute Probabilities from Probability Model for Bowls, Urns, Dice
Compare Probabilities and Estimates from Bowls, Urns, Dice
Random Variables
Compute the Values of Random Variables Using the Definition
Compute Probabilities of Outcomes Defined by Random Variables
Probabilities
Compute Probabilities from Scratch Using a Probability Model
Compute Probabilities Using Computation Rules
Interpret Probabilities Using Long Run Argument
Compute Conditional Probabilities Using the Definition
Random Samples and Probabilities
Estimate Probabilities from Samples
Compare Samples and Probability Models
Evaluate the Rare Event Assumption Using Samples
Six (6) cases will be presented on the First Hourly. They will be representative of cases worked
in class and of cases from sample tests.
Problem Tasks: II
Clinical Trial Methodology
Sketch a clinical trial
Spot faults in clinical trial sketches of designs
Random Sample Survey Methodology
Spot faults in clinical trial sketches of designs
Six (6) cases will be presented on the First Hourly. They will be representative of cases worked
in class and of cases from sample tests.
Will the problems count equally? Yes. There will be six (6) cases, at 25 points
Maximum per case, for a total of 150 points maximum. The raw test score will be
scaled to 100 percent
Will there be partial credit? Yes, but be aware that full work and detail is required
for full credit. Your work and details are the basis for scoring each case solution.
What about the testing protocol and tool-sheet? The hourly is not a memory test.
Hence, you are permitted the use of one (1) 8.5” by 11” sheet of paper. Put on this
sheet whatever it is that you deem useful. You alone will use this sheet. Sign and abide
by the test protocol that will accompany the hourly.
What about the calculator? You must provide your own working calculator, and
you must be able to use this calculator. Do not share calculators. Your calculator is
your individual responsibility.
The Essential Randomized, Controlled, Blinded Clinical
Trial
The clinical trial as discussed in this course is a very simplified
version of those designed and
executed in the Real World. The method essentially requires the
following:
A Well Defined Disease or Condition of Interest
An Appropriate Population at Risk
A Set of Well Defined Treatments
A Set of Well Defined End Points or Outcomes
A Random Method for Assigning Subjects to Treatments
A Method of Obtaining Informed Consent from Potential Subjects
Implementation of a Double Blind
We briefly sketched a variety of these simplified clinical trials. Examples
of these are available online.
A few notes of clarification.
The essential purpose of randomization is to assign subjects to
treatments in such a way that the only systematic differences between
treatment groups are random variations and the treatments themselves.
Blinding of both subjects and clinic workers is employed to avoid
differences in recorded responses due to either placebo effect or observer
bias. Blinding also helps to avoid excessive subject loss when a placebo is
employed.
A placebo is a medically inert mock treatment, intended to
resemble as closely as possible the active treatments in a study.
Broad classes of study end points and outcomes include treatment effect in
modifying disease or condition, survival/mortality, quality of life, side
effects and adverse events.
The Essential Random Sample Survey
The random sample survey was discussed in our class as a method for
scientifically obtaining information in a representative manner from a
population of interest. We presented some general principles on the design an
execution of credible scientific polling. Our simplified design for a random
sample survey included:
A Well Defined Population of Interest
A Sampling Frame Based on the Population of Interest
A Random Sample Obtained from the Sampling Frame
A Reasonable Set of Research Objectives
A Well Defined Survey Instrument Based on the Research Objectives
A Reasonable Protocol for Administering the Survey Instrument
Design Fault Spot
In this case type, brief descriptions of clinical trials or sample surveys are
presented.
Briefly identify problems with these designs.
Notes for Study / Preparation
Study for one case type at a time. Take notes as you
go along.
When you have finished study for all case types,
compile your notes into a single tool sheet.
Customize this tool sheet for your own personal use.
Some Advice
• Check your calculator’s health before the exam.
• Bring pens, and work the test in ink – or bring presharpened pencils.
• Be sure to study your weakest areas.
• Study more recent material earlier, then work
backwards.
• Tailor your tool-sheets to your individual needs.
• Recall that discussion and interpretation typically
comprise about 40% of earned credit on individual
cases.
• Be sure to use the methods and approaches specified in
the cases.