Stat. Lec: Lesson 5-9

• Internal Validity: Association between exposure and outcome is true if the study rules out the 3 alternative explanations
• Internal Validity: the extent to which a research study establishes a cause-and-effect relationship
• Bias: Systematic error in the investigator’s design or conduct of the study that leads to false association between exposure and outcome
• Bias: This can be done due to undercoverage, nonresponse, behavior of the interviewer, or poorly worded questions
• Random error: Probability that the observed result is due to chance
• Confounding: Third / extraneous / nuisance variable that distorts the relationship between exposure and outcome
• Confounding: Can be an independent risk factor for the outcome
• Qualitative Confounding: It inverses the direction of the association
• Adjusted Odds Ratio: determine the correct association between given variable and exposure
• Adjusted Odds Ratio: To control the confounder we adjust the association based on the category of your confounder
• Adjusted Odds Ratio: We stratify the crude based on the category of our confounder
• Positive Confounding: The confound pulls the observed association from true association
• Positive Confounding: It exaggerates true association
• Negative Confounding: Underestimates the interaction exposure and outcome
• Negative Confounding: It hides true association
• Cochran-Mantel-Haenszel method: A technique that generates an estimate of an association between an exposure and an outcome after adjusting for or taking into account confounding
• Cochran-Mantel-Haenszel method: The method is used with a dichotomous outcome variable and a dichotomous risk factor.
• Non-probability sampling designs: The probability of each member of the population to be included in the study is hard to determine
• Purposive / Judgment: Investigator chooses the samples subjectively
• Haphazard / Accidental: Researcher may use in his study whatever items come at hand / whoever is available
• Quota sampling: Data collectors have to obtain samples up to a given quota
• Snowball technique: Used in collecting data from hidden populations
• Probability sampling designs: The rules and procedures for selecting the sample and estimating the parameters are clearly and strictly specified
• Simple random sampling (SRS): Most basic type, small populations, Requires a sampling frame
• Systematic sampling: Variant of SRS, Used when sampling frame is unavailable or when the sampling units are too numerous to number
• Stratified sampling: Selected when the investigator wants every subgroup of the target population to be adequately represented. More precise than SRS
• Cluster sampling: Used when sampling frame is unavailable, Cheaper, requires a higher sample size
• Multi-stage sampling: Used when the target population is widespread, Population is divided intro primary (first stage), Secondary
• Bar Graph: Quali or Quanti, compare frequencies among the different categories of a variable
• Pie Chart: Qualitative, breakdown of a group / total that has few categories
• Component Bar Graph: Qualitative, similar with Pie chart (shows components or breakdown of a group)
• Histogram: Continuous quantitative, shows frequency distribution of a continuous variable
• Frequency Polygon: Quantitative, Shows 2 or more distributions in a single graph, similar with histogram
• Line graph: Shows trends or changes in a particular variable overtime, Time series
• Scatter Plot: Quantitative, Shows the relationship between 2 quantitative variables