We have a pleasure to invite you to participate in a training workshop on “Logistic Regression and Measurement Quality” organized by the Hanoi School of Public Health (HSPH)
Workshop on Logistic Regression and Measurement Quality
Sunday, 21st – Tuesday, 23rd December, 2008
at the Hanoi School of Public Health
The principal instructor of this course will be Professor Do Van Dung, Vice-Dean of Public Health Faculty, University of Medicine and Pharmacy (UMP) in Ho Chi Minh city. The course is designed in the practical way requiring a lot of computer practices.
This workshop is supported by the Health Research for Development Initiative (HRDI)* program with the aims of providing practical training to public health and social science researchers on advanced data analysis skills.
Specific Objectives: At the end of the workshop, the participants will be able to:
- Present the concept of maximum likelihood estimation
- Estimate the coefficient of logistic regression equation
- Interpret the results from logistic regression
- Differentiate between accuracy and precision
- Select and calculate the appropriate measure for accuracy and precision
Workshop contents:
1. Simple logistic regression
- Reason of using logistic regression: Logistic regression curve
- Odds and log(odds)
- The binomial distribution
- Simple logistic regression model
- Generalized linear model
- Maximum likelihood estimation
- Statistical tests and confident intervals: LR tests, Quadratic approximation of LR, score test; wald test an confident intervals
- OR and logistic regression model
- 95% CI for the OR associated with the unit increase in x
- 95% CI with grouped response data
- 95% CI for π[x]
- 95% CI for proportions
- Case control study: classical case control theory; 95% CI for the OR; test of the hypothesis that OR=1
- Logistic regression model for 2 x2 contigency tables
- Regresd disease against exposure
2. Multiple logistic regression
- Mantel-Haenszel estimate of an adjusted OR
- Mantel-Haenszel χ2 statistic for multiple 2 x2 tables
- 95%CI for the adjusted OR
- Breslow and Day's test for homogeneity
- Multipe logistic regression model
- Handling categorical variables in Stata
- Calculate Odds Ratios from multiple parameters
- The standard error of a weighted sum of regression sum coefficients
- Multiplicative model of two risk factors – Fitting Multiplicative model
- Fitting a model of two risk factors and interaction
- Nested models and model deviance
- Effect modifiers and confounding variables
- Goodness-of-Fit test:
- The Pearson χ2 Goodness-of-Fit test
- The Hosmer-Lemeshow Goodness-of-Fit
- Residual and influence analysis: Standardized Pearson residual and Δβˆ influence statistic
- Frequency matched case control studies
- Conditional logistic regression
3. Meassurment quality: precision and accuracy of direct variables
- the concept of direct and indirect measurement of variables
- terminology that reflects precision (reproducibility, repeatability, reliability)
- terminology that reflects accuracy (validity of many types - 'gold standard', criterion, concurrent, convergent, divergent, face, content, construct)
- that measurement 'quality' is reflected by showing evidence of both good precision and accuracy
- an introduction to the etiquette of drawing path diagrams to reflect direct and indirect measurements of variables and their inter-relationships
- the concept of agreement as distinct from correlation
- the widespread mis-use of correlation coefficients as measures of agreement
- the Bland-Altman approach to describing precision: the richness of interpretation therein
- coefficients of variation, intraclass correlation coefficients, and kappa coefficients as alternative summaries
Eligible Participants: All public health and social sciences researchers who already obtained a master degree are encouraged to apply. The priority will be given for researchers who are return fellows of the HRDI program, faculty from University of Medicine and Pharmacy (UMP) in Ho Chi Minh city, and Hanoi School of Public Health (HSPH).
Course requirements: Participants are required to bring their own laptop computers and are encouraged to bring their data sets to the training workshop.
Please fill in the application form and send it back to Mr. Nguyen Minh Hoang by 12 December 2008 to the following address:
Mr. Nguyen Minh Hoang, HRDI Program assistant,
Address: Hanoi School of Public Health, 138 Giang Vo Street Ba Dinh Hanoi
Email: nmh2@hsph.edu.vn.
Phone: 04-62662349. Mobile: 0914.341.349
Note:
(*) HRDI (Health Research for Development Initiative) is a partnership program among Hanoi School of Public Health (HSPH), University of Medicine and Pharmacy (UMP) in Ho Chi Minh city, and The Population Council in Vietnam (PCVN). The program support returned fellows who have studied overseas and obtained master’s degrees in public health or health social sciences under two fellowship programs managed by the Population Council and funded by the Ford Foundation and the Buffett Foundation to improve their research, teaching and professional skills. Total hits:
3
-
Last modified:
14/07/2009 01:42:36 PM | 