"How many subjects do I
need?" is a common question from researchers. "It depends" is the common,
unsatisfactory, response. Learn what the
answer depends on in Thomas Ryan's online course, “Sample Size and Power
Determination,” at Statistics.com.
"Sample Size and Power
Determination" covers how to plan the appropriate sample size for a study,
striking the optimal balance of feasible sample size, reasonable assumptions,
and acceptable power. The power of a study (the study's ability to prove a
treatment effect exists) is determined by such factors as the magnitude of the
treatment effect, the sample size, alpha (the level of statistical significance
required), and (for survival studies) the study duration.
Since some of these factors are
under the researcher's control while others are not, the goal of power analysis
is to balance them as a series of "What if's." For example "What
sample size would we need if the treatment reduces the risk of death by 10%,
and what sample size would we need if the treatment reduces the risk of death
by 20%?" Or, "How would power be affected if the study followed
patients for two years rather than three?" This process of finding a
balance among factors is done most effectively with graphs that allow the
researcher to grasp (and communicate) a range of options in a single picture,
and find the one that strikes the optimal balance of feasible sample size,
reasonable assumptions, and acceptable power.
Illustrations include examples from means, proportions, correlations,
and survival analysis, and possibly from other procedures as well.
Course Program:
Course outline: The course is structured as follows
SESSION 1: Introduction to Sample Size Determination and Power, Including
Useful Software
- Hypothesis tests and
confidence intervals
- Factors that determine
sample size
- Sample size for estimating a
population mean
- Examples, including a study
from the literature
- External and internal pilot
studies
- Ways to estimate sigma
- What should be avoided: Retrospective power and standardized
effect sizes
- Ethical issues in power
analysis
- Recommended references
- Software
SESSION 2: Tests on Population Means (continued)
- T-Test or Z-Test for
population mean?
- Testing the normality
assumption
- Confidence Intervals on
Power and/or Sample Size?
- Two-sample study from the
literature with unequal sample sizes
- Sample sizes determined by
scientist in two stages without software
- Illustration of more
efficient sample determination using software
- Using coefficient of
variation
- Paired data
- Additional examples
SESSION 3: Tests on Proportions and Variances
- One
proportion
- Software disagreement and
rectification
- Two proportions
- Options, including
transformations built into software, for tests of proportions
- One variance and two
variances
- Examples
SESSION 4: Regression and Design
- Simple linear regression
- Complexity caused by what
must be inputted
- Multiple linear regression
- Optional material:
Repeated measures designs, Logrank test for survival analysis
- Literature references for
sample size determination with more advanced
statistical methods and some information on corresponding software
capability
HOMEWORK:
Homework
in this course consists of short answer questions to test concepts and guided
data analysis problems using software.
In
addition to assigned readings, this course also has supplemental readings
available online
Who Should Take This Course?
Anyone responsible for the planning of a
study, or its subsequent analysis. Investigators writing grant applications or
other proposals in which sample size must be specified.
The instructor, Dr. Thomas P. Ryan, is the author of "Sample Size and Power Determination" (Wiley, 2103), a number of other books, plus numerous papers in peer-reviewed journals. He is an elected Fellow of the American Statistical Association, American Society for Quality, and Royal Statistical Society. Participants can ask questions and exchange comments with Dr. Ryan via a private discussion board throughout the period of the course.
You will be able to ask questions
and exchange comments with the instructors via a private discussion board
throughout the course. The courses take place online at
statistics.com in a series of 4 weekly lessons and assignments, and require
about 15 hours/week. Participate at your own convenience; there are no
set times when you must be online. You have the flexibility to work a bit every
day, if that is your preference, or concentrate your work in just a couple of
days.
We, the Center for eLearning and Training
(C-eLT), Pune, partner with Statistics.com and offer these courses to Indian participants
at special prices payable in INR.
Call: 020 6680 0300 / 322
Websites:
No comments:
Post a Comment