# lect07 - Categorical Data Analysis Lei Sun 1 CHL 5210...

This preview shows pages 1–5. Sign up to view the full content.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Categorical Data Analysis - Lei Sun 1 CHL 5210 - Statistical Analysis of Qualitative Data Topic: Stratified Samples and Matched Samples Outline • Odds ratio inference from stratified samples. • Models for matched pairs. Categorical Data Analysis - Lei Sun 2 Odds ratio inference for stratified samples. • One of the common problems in applied statistics is that of comparing two proportions from stratified samples (i.e. sets of 2 × 2 tables: K × 2 × 2 tables ). • Such data may be obtained, for example, – in cross-sectional studies where subjects are sampled separately by region of a city (i.e. stratified sampling). – in randomized trials when subjects are assigned to treatment groups separately by institution (stratified random assignment), and – when comparing a binary exposure (e.g. smoking status) across sam- ples created by modelling confounders (e.g. race) using dummy vari- ables, – in matched-pair case-control studies. • Summary estimates of association can be constructed using odds ratios, risk ratios or risk differences. • We will limit attention to inferences about odds ratios. Categorical Data Analysis - Lei Sun 3 • Inferences need to distinguish between two asymptotic cases: – Fixed-number-of-strata (e.g. the effect of smoking on the risk of hav- ing a low weight infant stratifying by mother’s race). – Increasing-number-of-strata (e.g. data from matched-pair case-control studies). • We will begin by considering methods of analysis applicable when there are a fixed number of strata. • Is mother’s smoking status associated with her risk of having a low weight baby after stratifying by mother’s race? Conditional independence: when X (SMOKE) and Y are independent at every level of Z (RACE). (Conditional independence does not imply marginal independence.) Is the odds ratio of having a low weight baby for smokers relative to non-smokers the same regardless of a mother’s race? Homogeneous association: when the effect of X on Y is the same at each level of Z . (Conditional independence is a special case of homogeneous association.) Categorical Data Analysis - Lei Sun 4 ## race = 1 low smoke | 0| 1| Total---------+--------+--------+ 0 | 40 | 33 | 73---------------------------+ 1 | 4 | 19 | 23---------+--------+--------+ Total 44 52 96 Type of Study Value 95% Confidence Limits----------------------------------------------------------------- Case-Control (Odds Ratio) 5.7576 1.7823 18.5992 ## race = 2 low smoke | 0| 1| Total---------+--------+--------+ 0 | 11 | 4 | 15---------------------------+ 1 | 5 | 6 | 11---------+--------+--------+ Total 16 10 26 Type of Study Value 95% Confidence Limits----------------------------------------------------------------- Case-Control (Odds Ratio) 3.3000 0.6346 17.1602 ## race = 3 low smoke | 0| 1| Total---------+--------+--------+ 0 | 35 | 7 | 42---------------------------+ 1 | 20 | 5 | 25---------+--------+--------+ Total 55 12 67 Type of Study Value 95% Confidence Limits----------------------------------------------------------------- Case-Control (Odds Ratio) 1.2500 0.3502 4.4616 Categorical Data Analysis - Lei Sun...
View Full Document

{[ snackBarMessage ]}

### Page1 / 32

lect07 - Categorical Data Analysis Lei Sun 1 CHL 5210...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online