Research Sampling Flashcards
In general ,what are the two types of sampling? Which is preferred?
nonprobability and probability
probability is preferred
What are the negatives with nonprobability sampling?
there are problems with representativeness, so it’s less generalizable
lower confidence in findings
What are the four common methods of nonprobablility sampling?
purposive
convenience
snowball
quota
What does probability sampling avoid?
Sampling bias
What is random selection?
all members have an equal chance of being selected
Equal Probability of Selection Method
What theory allows us to estimate the accuracy and representativeness of a sample?
probability theory
In sampling, what is an element?
an individual member of the population
what do you call the list of all the elements in a population?
the sampling frame
What is a parameter?
a summary of a given variable in a population - can be the population mean for example
What is a statistic and how is it different from the parameter?
It is the summary of agiven variable in a SAMPLE, not the whole population
the goal is to get the statistic as close as possible to the true parameter
What is the sampling idstribution?
the set of all the possible random samples that oculd be selected
What are the 4 commonly used types of random sampling design?
simple random
systematic
stratified
multistage cluster
Describe simple random sample
just straight random
get a sampling frame, assign numbers to eah, select by random number
Describe systematic sampling
you figure out the n you want
divide the population total by the n, this is the sampling interval k
list and number into the sampling frame
then start with a random number and select every k-th element within the population
What does it mean to avoid periodicity in systematic sampling?
you have to make sure the list isn’t oragnized in a specific way because you may miss out on a whole set if your random start number is far down
Describe stratified samplin
you random sample from sub populations to ensure better representativeeness and decrease sampling error
So stratify the list, figure out what percentage there are in each group, then randomly sample that percentage from each group.
What is oversampling?
It’s when you use stratified sampling and take more out of a group than their general percentage.
do this is the group you want to look at is really small percentage wise
What is cluster sampling?
you cluster the gruop of elements first , randomly sample
then randomly sample them again
so randomly sample hospitals first, then randomly sample PAs from the selected hospitals
What is the pro and con of cluster sampling?
it helps with cost and dispersed populations
it increases sampling error potential because you take two samples, so you double the error opportunity
FOr experiments, what is a greater concern than representativeness?
comparability of the control of experimental groups
they need to be randomized in sch a way that the groups don’t differ in a systematic way
what is sampling error?
variation in values ofyour sample mean compared to the population mean due to sampling
there will ALWAYS be some sampling error, the goal is to minimize
What are the 2 ways you can reduce smapling error?
increase the n
increase homogeneity (decrease variance)
In the normal curve, what percentage falls within 1 SD? 2SD? 3 SD?
68% between -1 and 1
95% between -2 and 2
99.7% between -3 and 3
What normal distribution is called the z-distribution
when the mean is 0 and the SD is 1
What is the critical z-score for a 90% co95%? 99%?nfidence interval?
- 65 = 90% CL
- 96 = 95% CL
- 58 = 99% CL
What is the confidence level? What is anothe rname for it? What do we usually set it as?
it’s the probability our sample statistic will fall within a confidence interval
it’ salso called alpha
usually set at -05, meaning we are 95% confident it will fal within a confidence interval
What is a confidence interval?
It’s the range within which the true parameter should lie - it’s a range around the estimate with an upper and lowe rlimit
If we want a confidence level of 95%, what is the equation for the confidence interval?
CI = mean +/- (1.96)(standard errors)
If they give standard deviations, what do you need to calculate before finding the CI?
the standard errors
SE = SD/square root of the n
What wideth of the confidence interval depends on which 3 things?
the alpha (confidence level)
the n - higher n, narrower interval
variation - higher variation more error wider interval
