Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
32 Cards in this Set
- Front
- Back
|
Generalizability (external validity) |
Extend to which a study's findings hold true outside of (external to) the particular context of the research |
|
|
Population of interest |
Some studies aim to represent a broad national, or even international population
Population the study aims to investigate
The broader the pop of interest, the more generalizable the results -Much more can be learned from studying unique or specialized groups (narrower pop) |
|
|
Random (probability) samples |
more generalizable
RDD
CBS poll |
|
|
Voluntary or convenience samples |
Less generalizable
voluntary: participants volunteer to be part of study
Convenience: researchers rely on most readily available participants-less representative |
|
|
Experiments (often generalizable) |
Even when they involve small, voluntary samples
Bc they focus on causal (what if) relationships |
|
|
Replication |
enhances generalizability
repeating a study with different sample, in different place, time perps, or policy context |
|
|
Meta-analysis |
A method for pooling together multiple smaller studies to get a more generalizable synthesis of findings |
|
|
population--universe |
Not always the general, human population of a country, region...
ex: air quality (air shed in city) |
|
|
Sample |
subset of people or elements selected from a population |
|
|
inference |
making conclusions about the population, based on data from a sample
sample--> inference--> population
To make proper inference, must be clear about definition of population |
|
|
Sometimes a study does not involve sampling |
Rather, data is gathered on the entire population
CENSUS |
|
|
Sampling frames |
List or enumeration that represents the population
ex: directory of listed phone numbers for households in city |
|
|
Coverage Bias |
When the sampling frame is systematically different from the target population
Ex: household phone directory might be all older ppl |
|
|
target population
Sampling frame
RDD |
everyone with a household land-line
yellow pages phone book--> biased
Sampling frame based on creating a list of phone numbers from random digits |
|
|
Nonresponse bias |
response rate= contact rate x cooperation rate |
|
|
Contact rate |
how successful researchers are in contacting people or units |
|
|
cooperation rate |
how willing the contracted people are to participate |
|
|
Nonresponse causes bias when |
propensity to response ℗ is related to what the survey is trying to measure |
|
|
Steps in assessing coverage bias |
1. define carefully the target population 2. identify the sampling frame --who is sampling frame but now target pop? 3. assess any systematic differences between frame and target pop 4. determine whether likelihood to be covered is related in any way to what study is measuring 5. describe analogous steps for non-response bias
|
|
|
Voluntary sampling |
putting out an ad or call for volunteers to participate in study |
|
|
Convenience sampling |
including people or elements in a study because they are conveniently available |
|
|
Sampling online--open web polls |
open web polls: polls posted on a web site that are open for any visitor to take
often biased due to interest in topic driving the likelihood to respond |
|
|
Internet access panels |
Lists of people who have agreed to participate in online research
less biased bc motivation to join panel is not as related to likelihood to respond
still a non probability sample |
|
|
purpose sampling |
choosing people or cases with a unique perspective, important role, or theoretical relevance
often used in small n (sample size) qualitative research
ex: sampling schools |
|
|
random (probability) sampling |
uses chance to select people
like drawing numbers from a hat |
|
|
Contribution of random sampling |
foundation for many gov't statistics
basis of major social surveys, public opinion research
methods for assessing precision of quantitative results |
|
|
Random sampling |
selecting elements from a population in order to make inferences describing the population |
|
|
Randomized experiments |
assigning people or elements to conditions to test causal relationships
rely on volunteers or convenience samples |
|
|
simple random sampling SRS |
selecting people from population in such a way that each individual has an equal chance or probability of selection
assumed most basic stat formula in stat software |
|
|
Sampling distribution |
distribution of estimates from many samples (of same size, drawn from same pop)
normal distribution (with big enough sample) |
|
|
Standard error-- standard deviation |
standard deviation of sampling distribution
measures precision of the estimate
formulas for proportion (p)
P is population proportion and n is sample size
all stats have SEs |
|
|
Empirical rule |
68% will be within -/+ 1 SE of mean
95% will be within -/+ 2 of mean
99.7% will fall within -/+ 3 of mean |
