Cochran W.G. Planning and analysis of observational studies

N.-Y.: Wiley, 1983. — 154 p.Treats studies, primarily in human populations, that show casual effects of certain agents, procedures, treatment or programs. Deals with the difficulties that comparative observational studies have because of bias in their design and analysis. Systematically considers the many sources of bias and discusses how care in matching or adjustment of results can reduce the effects of bias in these investigations.Variation, Control and Bias.Strategy in controlled experiments - sampled and target populations.
The principal sources of variation in the responses.
Methods of control.
Effects of bias.Statistical Introduction.
Drawing conclusions from data.
Tests of significance.
Confidence intervals.
Systematic differences between the populations.
The model when bias is present.Preliminary Aspects of Planning.The statement of objectives.
The treatments.
Measurement of treatment levels for individual persons and the effects of grouping.
Other points related to treatments.
Control treatments.
The responses.
Timing of measurements.Further Aspects of Planning.
Sample size in relation to tests of significance.
Sample size for estimation.
The effect of bias.
More complex comparisons.
Samples of clusters.
Plans for reducing nonresponse.
Relationship between sampled and target populations.
Pilot studies and pretests.
The devil’s advocate.Matching.
Confounding variables.
Matching.
The construction of matches.
Effect of within-class matching on x.
Effect of caliper matching on x.
Effect of nearest available matching on x.
Effects on bias of y1-y2.
Effect of matching on the variance of y1-y1.
Introduction to statistical analysis of pair-matched samples.Adjustment in Analysis.Y continuous: x’s classified.
Y binomial: x’s classified.
Treatment difference varying from cell to cell.
Y and x’s quantitative adjustments by regression(covariance).
Regression adjustments with some x’s classified.
Effect of regression adjustments on bias in y1-y2.
Effect of curvature on linear-regression adjustments.
Effectiveness of regression adjustments on precision.
Effect of errors in the measurement of x.
Matching and adjustment compared in experiments.
Matching and adjustment compared in observational studies.
A preliminary test of comparability.Simple Study Structures.A preliminary test of comparability.
The single group measured after treatment only.
The single group measured before and after treatment.
The single group series of measurements before and after references.