WEEK 2 - MEASURING PUBLIC OPINION

sampling

draw samples from overall pop, inferences from the sample about the nature of overall pop

sampling error

random sampling error, sampling bias

random sampling error

caused by random variation between samples, used to quantify exact uncertainty produced by rse

law of large numbers

so long as sample is random, as the sample size increases the sample avg will converge on the pop avg

margin of error

range of values where we expect the true population value to fall 19 times out of 20

sampling bias

anything that makes some people more or less likely to enter the sample, law of large numbers can’t save us

random sampling limitations

drawing a random sample solve sample bias, not perfect

convenience samples

don’t make claims to random sampling, cheaper but non-representative

level of measurement: categorical

places cases into categories based on whether characteristics are present or absent

level of measurement: continuous

place cases along a spectrum from more to less based on the degree to which characteristics are present, must be equal distance between categories

validity

the accuracy of an assessment, how accurate the measure represents the concept it’s measuring

measurement bias

production of scores that are systematically too low or too high

reliability

the consistency of an assessment, repeatable and consistent over time

common threats to measurement

social desirability bias v, leading questions v, double barreled questions vr

top of head considerations

ambivalence, avg across accessible considerations, accessibility = receny = top of their head answers

endogeneity

a correlation is driven by Y causing X