Exam Preparation Flashcards
var(X)
= E(X²) – E(X)²
E(Y|X)
= X × E(Y)
Variance using LIE
- w × var(A) + w × var(B)
- w × (E(A) – E(Total))² + w × (E(B) – E(Total))²
- Add (1) and (2)
Sample mean for more than two outcomes
Can do it but need two define variables - eg. as ‘1’ or ‘0’
Sampling Distribution
The probability distribution of a sample statistic over all possible samples from the population
Sample Mean is unbiased when
- Expected value of sampling distribution is equal to the population mean
- Can exist when n=1
- May not be true when the sample is not representative of the population
Larger Sample Size
- Reduces the variance of x̄
- Makes no difference to the bias of x̄
Standard Error of Difference
se(x̄1 – x̄2)
√(se(x̄1)² + se(x̄2)²)
Variance of Weighted Functions
On Formula Sheet
var(a + bZ) = b² × var(Z)
T Statistic
t = (x̄ - μ) / s.e
S.E for T Statistic
σ / √n
T Statistic for Difference of Two Means: Independent Samples
t = (x̄1 – x̄2) / s.e.(x̄1 – x̄2)
Consistent Sample
- Variance changes with the sample size
- Unbiased
Independent Samples
Each individual has equal chance to be chosen.
Therefore, independent surveys have a change of including the same person across them.
T Values for Regression
t = β1 / s.e(β1) or t = β0 / s.e(β0)
Su
=√(Sum of Squared Residuals/(n-2))
or =√(Sum of Squared Errors/(n-2))
R²
=var(Ŷ)/var(Y)
Ŷ
β0 + (β1 × X)
Type I and Type II Errors
+——–| True ——| False —-|
+——–+————–+————+
Accept | No error | Type II
+——–+————–+————+
Reject | Type I | No error |
Variance for Bernoulli Distribution
Var = p(1 – p)
Interpreting β1
1% increase is associated with a x% increase in…
Interpreting R²
The % that y is explained by x
Confidence Interval for T Statistic
[x̄±t(α/n,n-1) × s.e.(x̄)]
Confidence Interval for Independent Samples
[(x̄1– x̄2)±t(α/2,n1+n2 – 2) × s.e.(x̄1– x̄2)]
Confidence Interval for Regression
β̂1± t(α/2,df) × s.e.(β̂1)
Variance of T Distribution
v/v–2
Probability of a Type I Error
Significance level %
Power
Probability of not making a type II error
Standard Error for Proportions only
√(p̂(1 – p̂)/n)
T Statistic for Proportions
(p̂ – p)/se(p̂)
T Statistic for Difference of Two Means: Matched Pairs
tn-1 = (d̄)/(sd/√n)
d̄
x̄1– x̄2
Degrees of Freedom for Regression
Minus two for beta1 and beta0