Slide Number 1Class DetailsPrevious LectureToday’s LectureSlide Number 5Research The Scientific MethodThe Scientific Method (cont..)Empirical DataVariable and DataResearch Methods1. Case Studyhttp://www.cdc.gov/mmwr/preview/mmwrhtml/june_5.htm2. Naturalistic Research3. Archival RecordsSlide Number 164. Survey Research5. Experiment6. Quasi-experiment6. Quasi-experimentStudy Designs for Observational Studies Extending the Concept of “Quasi-Experiments”Cross-sectional (Survey)Prospective Design (cohort)Retrospective Design (case-control)Very simplified criteria used to infer causation:Expanded Criteria of Causation (Hill)Expanded Criteria of Causation (Hill)Expanded Criteria of Causation (Hill)Expanded Criteria of Causation (Hill)Expanded Criteria of Causation (Hill)Sufficient Cause and ComponentsDifficulty of Causal InferenceAssessing the Validity of the StudyTypes of Data or Scales of MeasurementMeasurementScales of MeasurementNominalSlide Number 38Slide Number 39OrdinalExamplesIntervalExample: Fahrenheit and Celsius scales of temperatureRatioExamplesOther Measurement ClassificationsQuantitative Data Can Be Discrete Or ContinuousReal Limits of a MeasurementThe Scientific MethodThe Scientific Method (cont..)Health Behavior Statistical Methods HP340 Lecture 2 January 12, 2017 Statistics in Scientific Research (Chapter 2)Class Details • Syllabus and key dates • Thursday 03/02 – Midterm • Tuesday 03/14 – No class (Spring break) • Thursday 03/16-No class (Spring break) • Thursday 05/09 - FinalPrevious Lecture • We covered material in Kiess & Green Chapter 1 • Key topics: – Variables and data – Statistic/descriptive statistic – Population and sample • Random sample, stratified sample, convenience sample – Parameters – Dependent and independent variables – ExperimentToday’s Lecture • We will cover material in Kiess & Green Chapter 2 • Key Topics – The Scientific Method • Research hypothesis, empirical data, research design/method – Research methods • Case study, observational study, archival records, surveys, experiments, quasi-experiments • Study Design for observational experiments. – Scales of measurement • Nominal, ordinal, interval, ratio – Discrete and continuous variablesScience is a method of gaining knowledge.Research Scientific question Reach an inference Hypothesis Analyze the data Define variables Describe the data Research method Collect empirical dataThe Scientific Method 1. Identify a question. 2. Propose an answer (research hypothesis). 3. Identify variables, propose how they are expected to be related. 4. Plan a research design (method) to test the hypothesis.The Scientific Method (cont..) 5. Collect empirical data. 6. Analyze data. 7. Based on analyzed data, confirm or denounce hypothesis. 8. Interpret results.Empirical Data • Empirical Data are scores that can be measured based on observation or sensory experiences.Variable and Data Variable: a characteristic that varies. Value of a variable constitute data 1. Dependent Variables: outcomes, endpoints, response or effect variables 2. Independent Variables: treatment, classification, risk factors, predictor variablesResearch Methods (Observational /non-intervention) Studies • Case study • Naturalistic observation • Archival records • Survey research (Intervention Studies ) • Experiment • Quasi-experiment1. Case Study • Descriptive study: careful examination of a single case (or several similar cases) • Used for rare or new conditions • Useful for generating hypotheses Example: 1981 report of Pneumocystis carnii pneumonia in 5 healthy young men in Los Angeleshttp://www.cdc.gov/mmwr/preview/mmwrhtml/june_5.htm2. Naturalistic Research • Unobtrusive observation of behavior taking place in natural settings. Example: Observe frequency of aggressive behavior among boys and girls on a playground. • Natural experiment - naturally occurring incidents (e.g. earthquake, famine, influenza epidemic). Examples: – Traffic related air pollution and asthma in children; – Dutch Hunger winter 1944-1945 & schizophrenia3. Archival Records • Research using existing records collected for some other purpose. (e.g. medical records, census data). • Examples: – Data from National Center for Health Statistics on Homicide. – HCUPnet: hospital and emergency department utilization (hthcupnet.ahrq.gov/) – National historical census data in GIS form (www.nhgis.org/ )4. Survey Research • Oral or written interviews. Collect data from a representative sample of interest. – Example: ask people for reaction toward an event and tabulate responses based on race/ethnicity, region, or occupation. • In this type of survey data, explanations of why people hold certain views is not attempted – i.e. direction of effect of the independent variable on the dependent variable is uncertain.5. Experiment • A controlled situation in which one or more variables are manipulated to investigate the effect on the dependent variable. • Experiments allow scientists to explore causal effects; the effect of the independent variable on the dependent variable.6. Quasi-experiment • Independent variable cannot be randomly assigned: – Not under control of the investigator. – Often an intrinsic characteristic of the subject. • Resembles an experiment but groups are not created by random assignment. – The independent variable grouping is determined before the experiment begins • Example: Having a particular disease or not and wanting to look at the frequency of a certain factor in association in diseased and non-diseased (Case-control sampling).6. Quasi-experiment • VERY IMPORTANT: – Because independent variable cannot be randomly assigned, – Can’t determine true causal relationship BUT can elude to one (common in epidemiology).Study Designs for Observational Studies Extending the Concept of “Quasi-Experiments” This type of study design attempts to detect and interpret relationships (associations). • Cross-sectional • Prospective • RetrospectiveCross-sectional (Survey) Factor B “Snap-shot” (Single Time Point) Factor A present absent absent present A & B A & not B not A & B not A & not B Example: TV viewing and obesityStudy population free of disease Factor present Factor absent disease no disease disease no disease Present time Future time Prospective Design (cohort) Time: both study