Help
Options
Didn't know it?
click below
click below
Knew it?
click below
click below
Don't Know
Remaining cards (0)
Know
retry
shuffle
restart
0:00
Embed Code - If you would like this activity on your web page, copy the script below and paste it into your web page.
Normal Size Small Size show me how
Normal Size Small Size show me how
ECO 418 FINAL
| Question | Answer |
|---|---|
| _____ is used for general relationship analyses | OLS |
| _____ and _____ are used for regressions with binary outcomes | Probit, logit |
| _____ is used for regressions with censored data | Tobit |
| _____ is used for regressions with count data | Poisson |
| In probit and logit regressions, we are trying to find the _____ that the dependent variable equals _____ | Probability, 1 |
| _____ _____ and _____ _____ are used for regressions with categorical, ordered independent variables | Ordered probit, ordered logit |
| We need _____ dummy variables in an ordered regression, since we always exclude the _____ category | K-1, baseline |
| Probit, logit, and poisson use _____ _____ to quantify the magnitude of the changes | Marginal effects |
| A variable that impacts the dependent variable through another independent variable | Instrumental variable |
| _____: an independent variable that is correlated with the error term _____: an independent variable that is not correlated with the error term | Endogeneous Exogeneous |
| If data is quite dispersed, it is better to use a _____ _____ model instead of a poisson model | Negative binomial |
| _____ _____ are used to observe characteristics specific to each entity or time period and examine how that specifically impacts the dependent variable | Fixed effects |
| When regressing with a binary or ordered dependent variable, we don't need a _____ _____ of _____ _____. We instead need to estimate the _____ of each outcome | Predicted line, continuous values, probability |
| Marginal effects are computed after probit, logit, and poisson regression to act as a measure of _____, and they are measured in _____ | Magnitude, percentages |
| In probit, logit, and poisson regressions, the coefficients tell us _____ about the magnitude at which the independent variable affects the dependent variable. The _____ and _____ are what we care about | Nothing, sign, significance |
| Greek letter denoting year fixed effects | γ |
| Subscript for entity fixed effects in regression models | i |
| Subscript for year fixed effects in regression models | t |
| Greek letter denoting the error term in regression models | ε |
| Greek letter denoting the second stage coefficients in regression models | α |
| Greek letter denoting the first stage coefficients in regression models | β |
| In difference in differences regression: The _____ group is the group that receives the program or policy The _____ group is the group that does not receive the program or policy | Treatment Control |
| In difference in differences regression, any difference in outcome between the two groups can be interpreted as a _____ _____ | Causal effect |
| In difference in differences regression, we need _____ _____, meaning that in the absence of treatment, we would expect no _____ in the trends of the _____ | Parallel trends, difference, groups |
| Main question of an interaction term: Is the gap between the groups _____ _____ for both categories? | The same |
| Stata will automatically drop a variable if it is _____ _____ with another variable | Perfectly collinear |
| _____-_____: a one unit change in X leads to a 100*Beta percent change in Y _____-_____: a one percent change in X leads to a one percent change in Y _____-_____: a one percent change in X leads to a Beta/100 unit change in Y | Log-linear Log-log Linear-log |
| These are the two criteria for creating a valid instrumental variable | 1. Instrumental relevance (the instrumental variable needs to be correlated with the endogeneous variable) 2. instrumental exogeneity (the instrumental variable needs to be uncorrelated with the dependent variable) |
| We add this to the endogeneous variable to indicate that it has been taken from the first stage regression equation and is now placed into the second stage | Hat symbol ^ |
| Stata code to regress using two stage least squares: _____ _____ _____ (_____ = _____) _____, _____ | ivregress 2sls dependent_var (endogeneous = instrumental) independent_vars, robust |
| When we have a categorical independent variable, we interpret coefficients relative to the _____ _____, and we begin interpretations with "_____ _____" | Base group, on average |
| Testparm F-tests are used to test for _____ _____ (a _____ _____ impact on dependent for all groups at once) | Joint significance, jointly significant |
| F-tests with var1 = var2 are used to test for the _____ _____ (both variables have _____ _____ impacts on the dependent variable) | Same impact, statistically similar |
| In a first stage regression equation, the _____ variable is used as the dependent variable, and the _____ variable is used as another independent variable, with other independent variables also present | Endogeneous, instrumental |
| This condition is used to measure if an instrumental variable is valid or not | Valid if F > 10 |
| List 4 situations where we might log-transform a variable | 1. Dollar denominated 2. Variables span a large range (population, income) 3. Data is right-skewed with outliers 4. Interpreting based on percentage changes makes more sense |
| A term included in polynomial regressions that indicates a convex/concave U shape in the relationship | Squared term |
| The purpose of the ramsey reset test | Check whether the model suffers from ommitted polynomial terms |
| The number of _____ variables can't be larger than the number of _____ in instrumental regression | Endogeneous, instrumental |
| A variable is endogeneous if there is _____ _____ between that variable and the dependent variable | Reverse causality |
| _____ _____ is the stata command to test for endogeneity _____ _____, _____ is the stata command to test for instrument validity | estat endogeneous estat firststage, all |
| The classic difference in differences regression includes an _____ _____ between the treatment group and the time after policy implementation | Interaction term |
| Difference in differences measures _____ _____ in the treatment group relative to the control group. By subtracting D-B from C-A, we remove the _____ _____ and isolate the _____ _____ | Additional changes, common trend, treatment effect |
| Regression discontinuity designs exploit a _____ that assigns treatment based on whether a _____ variable crosses a _____. They key idea is that individuals near the cutoff are nearly _____ except for treatment assignment | Policy, forcing, threshold, identical |
| The variable that determines treatment assignment in regression discontinuity | Forcing variable |
| The threshold that determines treatment in regression discontinuity | Cutoff |
| Observations close to the cutoff in regression discontinuity | Bandwidth |
| Regression discontinuity relies on the assumption that individuals cannot precisely _____ the running variable at the cutoff | Manipulate |
| In regression discontinuity, we aren't actually interested in the running variable. We are interested in how the _____ changes the observations right around the _____ inside the _____ | Treatment, cutoff, bandwidth |
| _____ regression discontinuity: Assignment to treatment only depends on X _____ regression discontinuity: Having X doesn't guarantee assignment to treatment, but it does increase the probability | Sharp Fuzzy |
| The main difference between regression discontinuity and difference in differences is that in RD, assignment to treatment is _____ | Random |
| Simple regression equation in regression discontinuity: Y = Beta0 + Beta1(_____ - _____) + Beta2(_____ dummy variable) + Beta3((_____ - _____) * _____ dummy variable) | Running - cutoff, treatment, running - cutoff, treatment |
| Regression discontinuity credibility depends on: 1. _____ of potential outcomes around the cutoff 2. No _____ or _____ of the running variable 3. _____ should be smooth at the cutoff 4. Correct local _____ form | Continuity Sorting, manipulating X Functional |
| Choosing a bandwidth is essentially a tradeoff between _____ and _____, since we need to get enough data points but we don't want to manipulate the outcome | Bias, precision |
| To perform a regression discontinuity in stata: _____ _____ _____, _____ | rdrobust outcome running_variable, c(cutoff) |
| Multinomial logit and probit are used for _____ _____ dependent variables | Unordered categorical |
| The two necessary characteristics of count data | 1. Discrete values 2. Non-negative integers |
| Effect of X on Y with a quadratic term: _____ = _____ + _____ * _____(_____) | Dy/dx = Beta1 + 2 * Beta2(X) |
| Quadratic term coefficients tell us if the effect on the X variable is _____ (U-shaped) or _____ (inverted U-shaped) | Convex, concave |
| We must omit one dummy variable when analyzing categorical outcomes to protect the model against _____ _____ | Perfect multicollinearity |
| We interpret time fixed and entity fixed effects in relation to the _____ | Baseline |
| Two-stage least square corrects for endogeneity by _____ the endogeneous variable with a clean version driven only by _____ factors | Replaces, exogeneous |
Created by:
Liechtensteiner