Vakken
Engels
Frans
Duits
Spaans
Nederlands
Grieks
Portugees
Italiaans
Latijn
Japans
Biologie
Aardrijkskunde
Natuur- en scheikunde
Wiskunde, rekenen
Economie
Geschiedenis
Eigen methodes
Alle vakken
Home
›
Alle vakken
›
Eigen methodes
›
Private
› 1 Ectrie
Helaas is de overhoormodule niet beschikbaar. Wel kun je deze lijst overhoren via StudyGo. Klik op 'Overhoren'
Private
, deel 1
1 Ectrie
Jaar 6 (Gymnasium)
Link voor email / website
Link naar overhoring, zonder bewerk/reactiemogelijkheid (ELO)
Open met deze code de oefening in miniTeach
Twitter
Facebook
Google+
LinkedIn
what are the tests for heteroskedasticity? = goldfeld-quandt, white test, likelihood ratio, breusch-pagan summary goldfeld-quandt = you split data in 3/2 groups. you apply ols on each group and compare the estimates. test statistic + distribution goldfeld-quandt = s2^2/s1^2, f(n2-k, n1-k). summary white tests = apply ols on model and store the residuals. then regress the residuals on a constant, all original regressors and squares. test statistic + distribution white tests = nR^2, chisquared(p-1), p is number of parameters. summary breusch-pagan = regress squared ols residuals on variables that possibly affect the variances test statistic + distribution bp = nR^2, chisquared(p-1) summary likelihood ratio = estimate model under h0 and h1 and compare log likelihood values test statistic + distribution lr = -2(log Lr - logLu), chisquared(k), k is number of restrictions con of goldfeld-quandt = the observations have to allow ordering in groups with increasing variances, rejecting h0 gives no information on what kind of model can be used con white test = doesn't require ordering, but test does not provide information on correct variance model pro bp = breusch pagan does provide information, since homoskedasticity is tested against a specific model con lr = the function h() must be specified which assumption concerns heteroskedasticity? = a3 which assumption concerns autocorrelation? = a4 consequence of heteroskedasticity = ols estimater b is unbiased and consistent, but inefficient. usual ols se no longer correct. consequence of autocorrelation = ols estimator b is unbiased and consistent, but inefficient give name of variance estimator when there's heteroskedasticity = white estimator give name of variance estimator when there's heteroskedasticity and autocorrelation = newey-west estimator how can serial correlation be included in the model? = include lagged variables, using autoregression model called cochrane-orcutt procedure what is the idea of generalized least squares? = to transform data in such a way that the conditions hold under which ols is efficient give the names of the tests for autocorrelation = breusch-godfrey, durbin watson, ljung-box, box-pierce distribution durbin watson = 2(1-r1) test statistic, distribution breusch-godfrey and name of test sort = nR^2, lm-test, chisquared(p) distribution box-pierce = chisquared(p) distriubution ljung-box = chisquared(p) pro breusch-godfrey = most generally applicable. summary breusch-godfrey = regress squared ols residuals on regressors and lags of ei. calculate nR^2 procedure for detecting outliers = inspect histograms of residuals, run jarque bera test, consider leverage and studentized residuals what is the connection between outliers and studentized residual or leverage? = a large ej* can be caused by a high leverage hjj for j. this in turn could mean that observation j is an outlier. what does the test for normality do? = jarque bera, it compares the third and 4th moments of residuals with normal distribution what's kurtosis and skewness under normality = k is 3, s = 0 distribution of joint test for normality = chisquared(2) which assumption does normality concern? = a7 what could be an alternative option to outliers? = alternative weighing schemes may be more robust to outliers which assumption concerns endogeneity? = a1 with what is a1 replaced if the regressors are stochastic? = plim(1/nX'X) is Q when is there endogeneity? = there is a correlation between x and errors what are causes of endogeneity? = ommited variables, measurement error, simultaneity what are consequences of endogeneity = ols estimator is inconsistent, standard tests not applicable what is a solution to endogeneity? = instrumental variables what is the idea of instrumental variables? = find m variables that do not affect y, but are correlated with x what are requirements of 2sls? = z and e uncorrelated, z correlated with x, z stable and not multicollinear, sufficient instruments what are the properties of iv/2sls estimator? = it is consistent and asymptotically normal give name of test for exogeneity = compare ols and iv, hausman test give name of test for validity instruments = sargan test, examine correlation exogenous vars with z what is a problem with comparing ols and iv when checking for exogeneity? = covariances negative in finite samples procedure exogeneity test = hausman test. regress y on X and store residuals. regress all endogenous variables on instruments and store end_residuals. regress ols residuals on all x and end_resisduals test statistic & distribution of hausman test = nR^2, chisquared(k0), k0 number of endogenous variables what is a crucial assumption when applying iv? = exogeneity of instruments. if it doesn't hold, then hausman test not valid procedure of sargan test = apply iv on model with instruments, store iv_residuals, regress iv_residuals on all instruments. test statistic & distribution of sargan test = nR^2, chisquared(m-k), k = number of restrictions with ols model which assumption does a linear model concern? = a6 what if there are too few variables in model? = bias what is there are too many variables in model? = efficiency loss what do we want with trade-off bias and efficiency? = small/no absolute bias and small variance when is the restricted model better? = beta2 is small, variance of b2 is large what are commonly used methods to select explanatory variables? = aic, sic, out of sample predictions, test results give name of test for non-linearity = reset test what does the test for non-linearity do = checks for nonlinearity by adding powers of the fitted values con of reset = general misspecification test. does not tell how the model should be adjusted which assumption does constant parameters hold? = a5 give names of tests for parameter variation = chow break test, chow forecast test, cusum, cusumq distribution of chow break test = f(k, n1+n2-2k) distribution of chow forecast test = f(n2, n1-k) distribution of cusum test = N(0,r-k) distribution cusumq test = chisquared(r-k)
Ingezonden op 17-04-2017 - 1049x bekeken.
Nog niet genoeg stemmen voor waardering: geef je mening!
voting system
1
2
3
4
5
Maak gratis account aan
Toon volledig menu
Door deze site te gebruiken, ga je akkoord met het gebruik van cookies voor analytische doeleinden, gepersonaliseerde inhoud en advertenties.
Meer informatie.
Overhoor en verbeter je talenkennis op woordjesleren.nl. De grootste verzameling van Franse, Engelse, Duitse en anderstalige oefeningen. Naast talen zijn ook andere vakken beschikbaar, zoals biologie, geschiedenis en aardrijkskunde!