Ch 6: Sampling Flashcards
(23 cards)
Sample
A subset of the target population (who/what you want to report about).
Representative Sample
when the characteristics of the individuals in the sample are consistent with those of the larger population being studied
sample of a classroom of 30 students could be 6 students, with 3 males and 3 females.
Response rate
The percentage of those people selected to participate in a study who actually complete the study
Incentive
researchers offer participants something to encourage them to agree to the study (improve the quality)
Sample size
number of individuals who take part in the study
larger the more powerful
Statistical power
indicates our ability to detect relationships in our population
failure to do so is type II error, which occurs when we mistakenly conclude that no relationships exists
Sampling frame
a list of all the individuals in a population
ex: membership list for a non profit, list of all students in campus directory
Internet panels
when samples are collected and recruited through the internet
Population
Complete set of persons, objects, or events that share some characteristic
Census
differs from a sample, all individuals within a population are meant to participate
time consuming, people leery to share info w government, difficult
SAMPLING METHODS: Non probability
-only a subset of the pop. is accessible
-target pop. is unknown, rare, or hidden
-when the aim is to develop a holistic (encompassing the whole) understanding of a group
Convenience Sampling
individuals are selected based on purely their availability (typically used for experiments)
ex: mall-intercept
Deliberate sampling or purposive sampling
attempts to target specific types of individuals to include within the study, often targeting a specific demographic group, behavior, or characteristic (such as college students, new parents, or people who own a particular brand of phone).
get a list of recent buyers from several local car dealerships and distribute surveys to those individuals by mail
Quota sampling
recruits specified numbers (or percentages) of individuals with specific traits
focus on basic demographics (gender, age, and race)
once reach 500 men, you wouldn’t allow any more men to participate
Snowball sampling
or network sampling, refers to when potential participants are identified by other participants in study
ex: suppose you are studying online social networks. You might start by recruiting a small group of individuals through some other sampling technique, then would ask those participants to give five other people in their social network, etc, etc
Volunteer sampling
researchers open the study to people who willingly participate with no attempt to recruit
an online poll dealing with opinions regarding gun control would likely draw people who are either strong supporters or strong opponents of control. (can be the most flawed)
SAMPLING METHODS: Probability sampling
or random sampling, all members of pop. have equal and known chance of being selected
Random sampling
refers to a process of choosing individuals that meets two criteria: each individual has an equal chance of being selected, and the selection of each individual is independent from that of all other individuals
ex: Suppose you have a roster of every student in a large introductory class with 1,000 students enrolled. A random number generator could produce 200 unique numbers, representing your sample.
simple random sampling
each individual should have an equal chance of being selected, and the likelihood of an individual being selected must be independent of that of other individuals being selected.
cluster sampling
type of sample in which entire groups of individuals are first identified and randomly sampled before individuals within those groups are sampled (reduces cost but error occurs at each stage)
ex: you want to conduct detailed, face-to-face interviews with individuals to learn their reasons for supporting a specific political candidate. You might, therefore, get a list of zip codes and randomly select 30 of those zip codes. Then, within the zip codes, if you could obtain a proper sampling frame for each, you could randomly sample individuals to recruit for your study. Interviewing 20 people within each zip code would give you a total study size of 600 while reducing your travel to just 30 locations.
stratified sampling
variation of a simple random sample in which the population is first divided into groups, and then a predetermined number of individuals in each group are selected randomly.
ex: if you are studying how men and women respond to advertisements, you might want to be sure you have equal numbers of each gender. If you have a sampling frame that identifies individuals by gender, and you want a total sample of 500, you could randomly select 250 men and 250 women. If the strata chosen match the distribution of individuals in the population, the overall sample will be representative. or
Divide a habitat into zones based on vegetation cover, then take samples from each zone
Systematic sampling
This approach to probability sampling uses a sampling interval to select units of analysis from the sampling frame, Select every {n^th} element from a list of the population, with a random start.
one potential drawback, the phenomenon known as periodicity-want to select random months of television content to analyze over a period of 20 years. In theory, each year would give you a randomly selected month, so you would have a good chance of having each of the 12 months represented in some way. However, suppose that the sampling interval ended up being 12. Because there are also 12 months in a year, you would end up selecting the same month for every year
Probability Sampling (Representative Sampling)
there will always be “random error”: statistically known as the confidence interval or margin of error (N= ~ 1,000 +- 3%)