7.1 Sample Surveys Flashcards
Statistical inference
For a conclusion on TOI
Population (P)
Only those that HAVE THE ability to qualify in the sample
Look at the context of the “random sample”, this is the same qualifications as pop BUT just sample version !! EXCEOT without TOI applied
!!! WORD form
UPPERCASE P
KW!!!- when definition doesnt have “part” and instead indicates ALL of a “group” even when it mentions words like “group”
Unit
ONE qualified thing from the POP, OR sample (ofc diff terms for both- pop size and sample size)
Population SIZE (N)
The actual NUMBER manifestation of POP (qualified but participating or not)
it is N !!!!
Parameter (p) - POPulation
Successes
Mean
LOWERCASE p
PERCENT (this will be the percent that is in question) That represents TOI of POP (NOT sample of it tho)
MUST come from census only (actual votes through elections not just predicted action/ approval)
KW!!-defitnion references it as a N value that characterizes SOME part of pop (TOI)
Census
The place / thing that actually accounts and gathers the DATA (TOI- and CV((voted at election/ SAMPLE)) of POP (not pop size)
Sample
PART of the population you collect data (successes/ not) from.
Meets requirements, not TOI, BUT DONT NEED to meet qualification of being polled at election (parameter)(Samping error)
WORD FORM
Sample SIZE (n)
Number manifestation of sample
# of units in sample (check requirements for both )
SUP IMP!!! Estimate / probability (p^)- SAMple
Estimate/ statistic(percent form) = success guess (shouldn’t be a %) / SAMPLE size
This is all estimation- because we dont “sample” the WHOLE pop (what makes it a statistic)
BUT if it comes from this one sample (yes, considering the above), its a PROBABILITY/ Statistic (p^)
!!!
USED to estimate the parameter (guessing if their thoughts will match up with ACTUAL outcome of votes (p))
Sampling error
Guessing from small portion mistake
Missing some of the MOST imp factors to what will lead to the parameter
!!!
Probability - parameter = p^ - p
Will give negative # -> off by this much BUT depends on how big the sample was (if big then distance of tick marks (SE) will be smaller, making this negative value even riskier)
POP parameter (outcome) v.s SAMPLE statistic (guess of outcome)
POP(parameter) sample( statistic)
Mean = MU x_
SD= sigma s
PROPORTION of successes-> p (parameter) p^ - estimate
Number of units (pop)= N n - sample size
KW!!- LOOK OUT FOR “ALL” - ALL is the parameter/ actual outcome of EVERYONE, not just outcome of a SAMPLE
The only percentages in the context (parameter VS estimate/ statistic)
Parameter
Population (all) proportion (successes/ all) = OUTCOME
Estimate
SAMPLE (part) proportion (guesses/ SAMPLE)= GUESS of outcome (because proportion percentage was out of sample, not all)
Need clarification- (8, A)
It asked for sample, not sample size but whatever- either way that number represents Americans of all ages (randomly- not ALL)
Sampling w/ and w/o replacement (10)
In stats it’s WITH, but MOST are W/O
Doesn’t matter tho if pop is significantly bigger than sample size (which most in our world aren’t- unless its within M and Ms bag)
It’s For probability/ RANDOM sampling/ deciding what UR sample is from pop:
WITH replacement:
When you pick a POP unit, you re PUT it in the pop so it MAY get picked again (makes it LESS random) !!!! Makes it BIASED
Simple random sampling (or use random sampling)
Each group of units (pop) has a chance at being picked to be in sample (!!!! MAY NOT be picked twice)
Done WITHOUT replacement (unbiased)
