Econometrics - up to week 6

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
full-widthPodcast
1
Card Sorting

1/147

encourage image

There's no tags or description

Looks like no tags are added yet.

148 Terms

1
New cards
What is econometrics?
Using statistical methods and data to estimate economic relationships; test hypotheses; and forecast outcomes.
2
New cards
What is a population?
The complete group we want to learn about.
3
New cards
What is a sample?
The observations collected from that population and used in the analysis.
4
New cards
What is an observation?
One recorded unit in the dataset such as a person; firm; country; or time period.
5
New cards
What is sample size n?
The number of observations used in the analysis.
6
New cards
What is a variable?
A characteristic that can take different values across observations.
7
New cards
What is the dependent variable y?
The outcome we want to explain or predict.
8
New cards
What is an explanatory variable x?
A variable used to explain differences in the outcome. Also called a regressor or independent variable.
9
New cards
What is a regression?
A statistical model describing how an outcome relates to one or more explanatory variables.
10
New cards
What is simple regression?
A regression with one explanatory variable.
11
New cards
What is multiple regression?
A regression with more than one explanatory variable.
12
New cards
What is a parameter?
An unknown numerical feature of the population model such as a true regression coefficient.
13
New cards
What is an estimator?
A rule or formula that uses sample data to estimate an unknown population parameter.
14
New cards
What is an estimate?
The numerical result obtained by applying an estimator to a particular sample.
15
New cards
What does β represent?
An unknown population regression coefficient.
16
New cards
What does β̂ represent?
A regression coefficient estimated from sample data.
17
New cards
What is a coefficient?
A number multiplying an explanatory variable in an equation. Its interpretation depends on the model's form and measurement units.
18
New cards
What is the intercept β₀?
The constant term. In a fitted level model it gives the prediction when all explanatory variables equal zero.
19
New cards
What does _cons mean in Stata?
The estimated intercept or constant term.
20
New cards
What is a fitted value ŷ?
The outcome predicted by the estimated regression for given explanatory-variable values.
21
New cards
What does a hat over a symbol mean?
It usually indicates an estimated or fitted quantity.
22
New cards
What does a bar over a symbol mean?
It usually indicates a sample mean. For example x̄ is the average observed x.
23
New cards
What does the subscript i mean?
It identifies a particular observation in the dataset.
24
New cards
What does Δ mean?
Change in a quantity. It is the new value minus the old value.
25
New cards
What does Σ mean?
Add the specified quantities together across observations.
26
New cards
What is the error term u?
The part of the population outcome not explained by the included regressors. It captures unobserved influences.
27
New cards
What is a residual û?
Actual outcome minus fitted outcome: û = y − ŷ.
28
New cards
What does a positive residual mean?
The actual outcome is higher than the fitted value.
29
New cards
What does a negative residual mean?
The actual outcome is lower than the fitted value.
30
New cards
What distinguishes an error from a residual?
An error is relative to the unknown population equation. A residual is relative to the estimated sample equation.
31
New cards
What is OLS?
Ordinary Least Squares. A method that chooses coefficients to minimise the sum of squared residuals.
32
New cards
What is a squared residual?
A residual multiplied by itself. Squaring removes its sign and gives larger deviations more weight.
33
New cards
What is SST in your lectures?
Total Sum of Squares. The total squared variation of observed y values around their sample mean.
34
New cards
What is SSE in your lectures?
Explained Sum of Squares. The sum of squared differences between fitted values and the sample mean of y.
35
New cards
What is SSR in your lectures?
Residual Sum of Squares. The sum of squared differences between actual and fitted values.
36
New cards
What is RSS in your sample tests?
Residual Sum of Squares. It has the same meaning as SSR in your lecture notation.
37
New cards
Why must you check sum-of-squares abbreviations?
Different sources use different conventions. Identify whether the quantity means total; explained; or residual variation.
38
New cards
What is the relationship between SST and SSE and SSR?
For OLS with an intercept: total variation = explained variation + residual variation. Therefore SST = SSE + SSR.
39
New cards
What is goodness of fit?
How closely the fitted model matches the observed outcomes.
40
New cards
What is R²?
The fraction of sample variation in the dependent variable explained by the fitted regression.
41
New cards
What does R² = 0.60 mean?
The model explains 60% of the sample variation in the dependent variable around its mean.
42
New cards
What does R² not tell you?
It does not establish causation and it is not the percentage of predictions that are correct.
43
New cards
What is adjusted R²?
A measure of fit that adjusts for sample size and the number of regressors. Unlike ordinary R² it can fall when a regressor is added.
44
New cards
What is a mean?
An average calculated by adding values and dividing by how many values there are.
45
New cards
What is variance?
A measure of spread based on squared deviations from a mean.
46
New cards
What is a standard deviation?
The square root of a variance. It measures spread in the original units of the quantity.
47
New cards
What is covariance?
A measure of how two quantities vary together. Positive covariance indicates that above-average values tend to occur together.
48
New cards
What is correlation?
A standardised measure of linear association ranging from −1 to +1.
49
New cards
Does zero correlation mean independence?
Not generally. Variables can have a nonlinear relationship despite zero correlation.
50
New cards
What is a sampling distribution?
The distribution an estimator would have across repeated samples drawn using the same sampling process.
51
New cards
What is sampling variability?
The variation in estimates that would occur because different samples contain different observations.
52
New cards
What is a coefficient's standard error?
An estimate of the standard deviation of that coefficient estimator's sampling distribution.
53
New cards
What does a small standard error indicate?
Relatively little sampling uncertainty in the coefficient estimate compared with a larger standard error measured on the same scale.
54
New cards
What is precision?
How tightly an estimator's sampling distribution is concentrated. Lower sampling variance means greater precision.
55
New cards
What is σ² in the homoskedastic regression model?
The common conditional variance of the population errors.
56
New cards
What is σ̂²?
An estimate of the error variance. In the usual OLS model it equals SSR divided by residual degrees of freedom.
57
New cards
What is Root MSE in Stata?
The square root of the estimated error variance. It estimates the error standard deviation.
58
New cards
Is Root MSE the same as a coefficient's standard error?
No. Root MSE concerns unexplained outcome variation. A coefficient's standard error concerns uncertainty in an estimated coefficient.
59
New cards
What are degrees of freedom?
The number of independent pieces of information remaining after accounting for estimated parameters or constraints.
60
New cards
What are residual degrees of freedom in your regression models?
n − k − 1 where n is sample size and k counts slope coefficients excluding the intercept.
61
New cards
What does k represent?
The number of slope coefficients or explanatory variables in the regression. It excludes the intercept.
62
New cards
What does ceteris paribus mean?
All other relevant factors held constant.
63
New cards
What is a partial effect?
The effect of changing one explanatory variable while holding the other relevant factors fixed.
64
New cards
What is causation?
A relationship in which changing one factor produces a change in an outcome.
65
New cards
What is a confounding factor?
A factor that affects the outcome and is also related to the explanatory variable being studied.
66
New cards
What is zero conditional mean?
The assumption that the population error averages to zero at every combination of included explanatory-variable values.
67
New cards
What does E(u|X) = 0 mean in words?
The average error is zero conditional on the included regressors. X represents the explanatory-variable information.
68
New cards
What is endogeneity?
A relationship between regressors and the population error that violates the relevant exogeneity condition. Correlation with the error is a standard example.
69
New cards
What is exogeneity in the assumptions used for this course?
The included regressors satisfy the zero conditional mean condition required by the model.
70
New cards
What is bias?
The difference between an estimator's expected value and the true population parameter.
71
New cards
What is an unbiased estimator?
An estimator whose expected value across repeated samples equals the true parameter.
72
New cards
Does unbiased mean correct in every sample?
No. Individual estimates can be above or below the truth even when the estimator is unbiased.
73
New cards
What is homoskedasticity?
The conditional error variance is constant across explanatory-variable values.
74
New cards
What is heteroskedasticity?
The conditional error variance changes with explanatory-variable values.
75
New cards
What is perfect collinearity?
An exact linear relationship among regressors including the intercept when present. It prevents unique estimation of the separate coefficients.
76
New cards
What is multicollinearity?
Strong linear relationships among explanatory variables. Imperfect multicollinearity can make coefficients imprecisely estimated.
77
New cards
What is an auxiliary regression?
An additional regression run to obtain information needed for analysing the main model.
78
New cards
What is Rⱼ² in the coefficient-variance formula?
The R² from regressing explanatory variable xⱼ on the other explanatory variables and an intercept.
79
New cards
What is SSTⱼ in the coefficient-variance formula?
The sum of squared deviations of xⱼ from its sample mean. It measures sample variation in that explanatory variable.
80
New cards
What is VIF?
Variance Inflation Factor. It measures variance inflation associated with linear relationships between a regressor and the other regressors.
81
New cards
What is statistical inference?
Using sample evidence to draw conclusions about population parameters while accounting for uncertainty.
82
New cards
What is a hypothesis?
A claim about a population parameter or a set of population parameters.
83
New cards
What is the null hypothesis H₀?
The specific population claim being tested. Example: a regression coefficient equals zero.
84
New cards
What is the alternative hypothesis H₁?
The competing claim specifying which departures from the null count as evidence against it.
85
New cards
Does a null hypothesis always say a coefficient equals zero?
No. It can specify another value or a relationship between several coefficients.
86
New cards
What does H₀: β₁ = 0 mean?
The population coefficient on x₁ is zero in the specified model.
87
New cards
What does H₀: β₁ = β₂ mean?
The two population coefficients are equal.
88
New cards
What is a test statistic?
A number calculated from sample data that measures evidence against a null hypothesis.
89
New cards
What is a t-test?
A test commonly used to assess a single claim about a regression coefficient by comparing its estimate with the null value relative to its standard error.
90
New cards
What is a t-statistic?
The estimated coefficient minus its hypothesised value divided by its standard error.
91
New cards
What does t = 2 mean?
The estimated coefficient is two standard errors above the hypothesised value.
92
New cards
What does t = −2 mean?
The estimated coefficient is two standard errors below the hypothesised value.
93
New cards
What does Stata's reported t-statistic usually test?
Whether the corresponding population coefficient equals zero.
94
New cards
What is a one-sided test?
A test whose alternative specifies a direction such as β > 0 or β < 0.
95
New cards
What is a two-sided test?
A test whose alternative allows departures in either direction such as β ≠ 0.
96
New cards
What is an upper-tail test?
A test that rejects for sufficiently large positive values of the test statistic.
97
New cards
What is a lower-tail test?
A test that rejects for sufficiently negative values of the test statistic.
98
New cards
What is a critical value?
A cutoff from the reference distribution used to decide whether a test statistic lies in the rejection region.
99
New cards
What is a rejection region?
The set of test-statistic values that lead to rejection of the null hypothesis.
100
New cards
What is the significance level α?
The chosen probability of rejecting a true null under the test's assumptions.