Introduction: Research designThe Biggest Problem in Research: Establishing CausalitySlide 3Slide 4Slide 5The problemReview of internal and external validityInternal validity: the two problems The two primary threats to internal validity.External validitySlide 10ClarificationWhat is gold standard research design?Next class: STATAIf you want to play around with StataIntroduction: Research design17.871Spring 20121The Biggest Problem in Research: Establishing CausalityReturn to the case of voting machine problems in FloridaAfter the election, we wanted to know: are some machines “better” than others?For the policy choice, we want to know if2The Biggest Problem in Research: Establishing CausalityReturn to the case of voting machine problems in FloridaAfter the election, we wanted to know: are some machines “better” than others?For the policy choice, we want to know if3The Biggest Problem in Research: Establishing CausalityReturn to the case of voting machine problems in FloridaAfter the election, we wanted to know: are some machines “better” than others?For the policy choice, we want to know if4The Biggest Problem in Research: Establishing CausalityReturn to the case of voting machine problems in FloridaAfter the election, we wanted to know: are some machines “better” than others?For the policy choice, we want to know if5The problemHow do we make sure that quality differences observed among machine types are due to machine types per seThis is an issue of causalityWe attend to “internal validity” so that when we observe differences between groups, we can assure ourselves that this is because of the “treatments” of interest6Review of internal and external validityInternal validity: the two problemsThe two primary threats to internal validity.1. Nonrandom selection into the treatment group (confounding variables)Comparing apples with apples or apples with oranges?Random assignment ensures apple to apple comparisonsRegression, matching, difference-in-differences also attempt to compare apples with apples2. Reverse causationThe chicken and egg problem, which came first?Is your dependent variable influencing your treatment (your explanatory variable)?If you can address these problems, you almost always have an internally valid studyRandomly assigned experiments address bothExternal validityIs your sample representative of the population?Make sure your study population is relevant to the general populationAddress by randomly samplingGood research is about addressingInternal validityExternal validityClarificationRandomly sampling cases gets you?External validityRandomly assigning to treatment group?Internal validityControlling for variables with regression addresses?Internal validityWhat study design addresses both internal and external validity?Field experimentsWhat is gold standard research design?Field experiment, e.g., Connecticut voting turnoutWhy? AddressesInternal validityNonrandom selection into the treatmentReverse causationExternal validityWhat aspects of our lives are governed by gold standard research?In this class, we mostly do observational studies, But the key to a successful observational research is always keep in mind how one study differs from a field experiment12Next class: STATA Kohler & Kreuter, Data analysis (2nd edition)Chapter 1Skip section 1.3.19 (linear regression)Chapter 3Only read section 3.1 Chapter 5Read section 5.1 but skip 5.1.3 and 5.1.4Read section 5.2Handout: “How to use the STATA infile and infix commands” (course website)13If you want to play around with StataVisit http://ist.mit.edu/services/software/athena/numerical for basic info about accessing Stata in AthenaLook on pp. xxi-xxii of Kohler & Kreuter for info about download example
View Full Document