unit 1

population

set of subjects of interest on what we want to study

Can’t observe the whole target population so what would be the solution?

use subsets (sample) that is representative of the target population

parameter

population, number description of a population

• average income

• greek letters

• fixed, not observable

stats

sample, number of description of a sample

• average income in sample

• latin letters

• varies sample to sample

descriptive stats

for numerical summaries , sample average

inferential stats

inference make conclusions about population values (parameters)

why do researchers use descriptive + inferential stats

answers questions about social phenomena

unit of analysis “sample”

not the topic of the study it studies the individuals not attitudes

ecology fallacy “target populations”

“ecological” —> group / sets something bigger than individuals

EF assumes something learned about an eco unit says something about individual in unit

example of ecological fallacy

• study of neighborhoods shows a higher crime where there are many undocumented immigrants

• ef: cconclude that undoc imm more likely to commit crimes

• cant make conclusions about individuals from data describing cities

constant

character’s dont vary across observations

not to useful need to explain variation

outcomes

all variables have associated outcomes

different outcomes for different variables

• number of siblings —> #

• sex —> female/ male

• pot legal? —> agree/ disagree

• pol. party —> demo, rep, ind

categorical

category variable that provides less info

• non orderable (nominal)

• orderable (ordinal)

quantitative

interval variable, think of numbers

nominal

• variable

• gender and hair color

ordinal

• semesters at UCi

• height

• social classes

different interval variables outcomes

• crime rates —> # of crimes/ 1000

• # of siblings —> #

• income —> $

discrete (categorical)

outcomes separate number of categories

you are either one or the other

continuous (quantitative)

The outcome can take on any number on numberline

think of number line

sampling

subset of observations

purpose establish population value of the parameter

representative sampling

need to be representative of population

• random (probability)

• representative of population (randomization)

• non random

• doesnt representative of the population

simple random sampling

select the cases at random equal chances of being selected

sampling error

difference between estimates (statistics) value (parameter) want to estimate

sampling variability

stats vary from sample to sample, different random sample and yields

what are the different bias

sampling bias, reponse bias, non-response bias

sampling bias

• non random sample can make inferences

• results are invalid

response bias

• questions are poorly worded

• biased inference

non-response bias

missing data; unanswered questions

refuse to participate