110



                                                                                                          SOURCES OF ERROR AND BIAS IN 
                                                                                                                              EPIDEMIOLOGIC  STUDIES
                                              


                                                                                                         
                                                    EXTERNAL                                                      RANDOM                   SYSTEMATIC
                             POPULATION
                                                                                                                                                   non-random
                                                                                                                                                   sampling

                                                                                                                                                   selection bias
                                                                                                                                                   (bad choice of
                                                                                                                                                   
                                                                                                                                                       populations)
                                 TARGET
                                POPULATION


                                                                                                                                            

                                                     


                                                                                                 
                                                         sampling                                    sampling                                   non-random  sampling
                                                                                                                                                              
                                                                                                                                                        




                                                                                                                error
                                                                                                                                                   selection bias
                                                                                                                                                   
                                                                                                                                                       (e.g.,  survivor bias)
                                                                STUDY
                                POPULATION


                                                                                                                                                   
                                                                                                                                                       selection bias
       request  participation                                                                                                                      (e.g., non-participation)


                                       
                                                   STUDY
                      PARTICIPANTS

                                                                                       random variability                                          misclassification
                                                                                                                                                   

                                                                                        
                                                                                            (within subject)                                           (subject bias)
                                    measurement                                                                                                    
                                                                                        
                                                                                            (within observer)                                          (observer bias)
                classification                                                                                                                     
                                                                                        (between observers)                                            (measurement bias)
                                                                                        (test)


                                                     COMPARISON
                                                     

                                                                                                                                                               
                                                              GROUPS                                                                                               confounding