L3: inference & estimation Flashcards
(19 cards)
sample statistics
mean:x̄, var: s^2, sample prop: p
population parameters
mean:μ, variance: σ^2, pop prop: π
estimators
set of rules to calculate pop parameter estimate
estimators criteria
bias, precision
for higher precision
larger sample size (n increase = SE decrease)
point estimate for large sample
sample mean
what does 95% c.i tell you
95% confident that pop mean lies within 1.96 std errors of x bar (point estimate) (NOT 95% prob)
what does z val (e.g. 1.96) tell you at 95% c.i.
95% of data estimates is +/- 1.96 s.d. of mean (within 2.5 tail)
how does confidence level capture uncertainty
estimate drawn from sample (diff sample = higher / lower estimate)
smaller sample size on CL
wider - greater uncertainty
99% confidence interval on table
0.5% each tail = 0.005 on z table (=2.575)
sample proportion mean and variance
mean: 𝜋, variance: 𝜋(1-𝜋) / n
sample proportion distribution
p~N(𝜋, [𝜋(1-𝜋)]/n)
confidence level for sample proportion
same as mean; p +/- interval est √[𝜋(1-𝜋)]/n)
point estimate for sample proportion
p (=𝜋 bc 𝜋 unknown)
t distr used when
n is small, 𝝈 unknown
degrees of freedom
v=n-1
for d distr, 2 things you need to knwo
v=n-1, ⍺/2= area in each tail