Ans. d. Multivariate analysis (Ref: Park 23/e p73, 83, 21/e p68-70, 78-79, 84; en.wikipedia.org/wiki/Bias_(statistics); style="font-size: 1.04761904761905em; font-family: Times New Roman, Times, serif; margin: 0 0 0 8px; text-indent: 0">Multivariate analysis refers to the analysis of two or more dependent variables at a time. It is done after the study has been conducted, to analyze the study. While matching, binding and randomization all are done when the study is being conducted. (AIIMS November 2010 Repeat)."Confounding factors are an important source of bias. This can be removed by matching in case-control studies.""The selection bias can be eliminated by randomization.""Interviewer's bias can be eliminated by double-blinding."Groups to be compared are differentially susceptible to the outcome of interest, even before the experimental maneuver is performed is known as selection bias. In the given question, researcher is concerned that the smokers might differ from non-smokers in their diet, exercise, etc. This concern is known as selection bias.Some important Types of Biases in Epidemiological StudiesInterviewer bias* Interviewer devotes more time of interview with cases as compared to controlsQMemory/ Recall bias* Cases are more likely to remember exposure more correctly than controlsQSelection bias (Susceptibility bias)* Groups to be compared are differentially susceptible to the outcome of interest, even before the experimental manoeuvre is performedQInformation bias (Observer or misclassification bias)* Connective type of bias in which proportion of subjects are misclassified on exposure or diseases* Occurs due to people's curiosity and confusion of goals when trying to choose a course of action* It includes reporting, recall, interviewer & attention biasQBiasBias is any 'systematic error' in an epidemiological study, occurring during data collection, compilation, analysis and interpretation0Predominantly Biases are of 3 TypesSubject biasInvestigator biasAnalyzer bias* Error introduced by study subjects.* Examples:- Hawthorne effect- Recall biasQ* Error introduced by investigator* Selection biasQ* Error introduced by analyzer Some Important Types of Biases in Epidemiological StudiesApprehension bias* Certain levels (pulse, blood pressure) may alter systematically from their usual levels when the subject is apprehensiveAttention bias (Hawthorne effect)* Study subjects may systematically alter their behaviour when they know they are being observedQBerkesonian bias (Admission rate bias)* Bias due to hospital cases and controls being systematically different from each otherQInterviewer bias* Interviewer devotes more time of interview with cases as compared to controlsQLead time bias (Zero time shift bias)* Bias of over-estimation of survival time, due to backward shift in starting point, as by screening proceduresQMemory/ Recall bias* Cases are more likely to remember exposure more correctly than controlsQNeymann Bias (Prevalence-incidence bias)* Bias due to missing of fatal cases, mild/ silent cases and cases of short duration of episodes from the studyQSelection bias (Susceptibility bias)* Groups to be compared are differentially susceptible to the outcome of interest, even before the experimental manoeuvre is performedQ Minimization of Biases in Epidemiological Studies by BlindingTypeMethodMinimizesSingle blindingStudy subjects are not aware of the treatment they are receivingSubject biasDouble blindingStudy subjects as well as investigator are not aware of thetreatment study subjects are receivingSubject bias + Investigator biasTriple blindingStudy subjects, investigator as well as analyzer are not awareof the treatment study subjects are receivingSubject bias + Investigator bias + Analyser bias