Ch 4 Sampling How to Select a Few

Ch. 4, Sampling: How to Select a Few to Represent the Many (Pt. 1) Neumann, pp. 86 -105. 1

HOW AND WHY DO SAMPLES WORK? n A proper, representative sample lets you study features of the sample and produce highly accurate generalizations about the entire population n The most representative samples use random selection The random process allows us to build on mathematical theories about probability n Due to their use of random selection, probability samples are also called random samples n 2

Sample, population, random sample n sample: a small collection of units taken from a larger collection n population: a larger collection of units from which a sample is drawn n random sample: a sample drawn in which a random process is used to select units from a population 3

Sampling in qualitative vs quantitative research n Qual. & quant. researchers both use sampling, but qualitative researchers have different goals than to get a representative sample of a large population, so they rarely use random sampling n Instead, they actually want to learn how a small collection of cases, units, or activities, can illuminate key features of an area of social life n n Use sampling less to represent a population than to highlight informative cases, events, or actions Goal is to clarify and deepen understanding based on what's learned from highlighted cases 4

FOCUSING ON A SPECIFIC GROUP: 4 TYPES OF NONRANDOM SAMPLES n Random samples are best to get an accurate representation of a population, but they are difficult to conduct n Researchers who cannot draw random samples use nonprobability sampling techniques, e. g. , n n Convenience sampling Quota sampling Purposive or judgmental sampling Snowball sampling 5

Convenience Sampling n convenience sampling: a nonrandom sample in which you use a nonsytematic selection method that often produces samples very unlike the population n Also called accidental or haphazard sampling, it’s cheap and fast, but of limited use n With caution, can be used for the preliminary phase of an exploratory study 6

Quota sampling n n quota sampling: nonrandom sample in which you use any means to fill preset categories that are characteristics of the population Not as accurate as a random sample, much easier and faster 1) 2) 3) Identify several categories of people or units that reflect aspects of diversity in population you believe to be important (gender, age, etc. ) Decide how many units to get for each category, i. e. , what the quota will be After setting categories and # of units in each category, select units by any method 7

Purposive or Judgmental Sampling n purposive sampling: a nonrandom sample in which you use many diverse means to select units that fit very specific characteristics n It’s like convenience sampling for a highly targeted, narrowly defined population n Can be used in two types of situations: 1) 2) to select especially informative cases to select cases from a specific but hard-toreach population 8

Snowball Sampling n snowball sampling: a nonrandom sample in which selection is based on connections in a preexisiting network n Also called network, chain-referral or reputational sampling, it’s a special technique in which goal is to capture an already existing network It is a multistage technique n The crucial feature is that each person or case has a connection with the others n 9

Networks for which researchers used snowball sampling n Scientists around world investigating same issue n The elites of a medium-sized city who consult with one another n Drug dealers and suppliers in a distribution network n People on a college campus who have had sexual relations with one another 10

COMING TO CONCLUSIONS ABOUT LARGE POPULATIONS n sampling element: a case or unit of analysis of the population that can be selected for a sample n can be a person, a group, an organization, a written document or symbolic message, or a social action or event (e. g. , an arrest, a protest event, divorce, a kiss) 11

3 terms with similar meanings are often confused, but they’re related by degree of specificity (from less to more) n universe: the broad group to whom you wish to generalize your theoretical results n e. g. , all people in FL n population: a collection of elements from which you draw a sample n e. g. , all adults in the Miami metro area n target population: the specific population that you used n e. g. , all adults who had a permanent address in Dade country, FL in Sept 2007, and who spoke English, Spanish, or Haitian Creole 12

Once you have a target population… n you must create a list of all its sampling elements, your sampling frame n sampling frame: a specific list of sampling elements in the target population n population parameter: any characteristic of the entire population that you estimate from a sample n sampling ratio: the ratio of the sample size to the size of the target population 13

Why Use a Random Sample? n Random samples are most likely to produce a sample that truly represents the population n True random processes are: 1) purely mechanical or mathematical without human involvement 2) they allow us to calculate the probability of outcomes with great precision 14

All samples contain a margin of error n A random process makes it possible to estimate mathematically the degree of match between sample and population, or sampling error n sampling error: the degree to which a sample deviates from a population 15

Random samples all have 3 key features: 1) They are based on an accurate sampling frame or list of elements in the target population 2) They use a random selection process without subjective human decisions (e. g. , a computer program, random number table) 3) They rarely use substitutions for sampling elements 16

Types of Random Samples n Simple Random Samples n Systematic Sampling n Stratified Sampling n Cluster Sampling 17

Simple Random Samples n In simple random sampling, you: n n n First develop an accurate sampling frame Select elements from the frame based on a mathematically random selection procedure Locate the exact selected elements to be in your sample 18

Over many separate samples, the true population parameter is the most frequent result n sampling distribution: a plot of many random samples, with a sample characteristic across the bottom and the number of samples indicated along the side n The sampling distribution shows the same bell -shaped pattern whether your sample size is 1000 or 100 19

Systematic Sampling n If you lack the tools to select a pure random sample, systematic sampling is a quasirandom method n systematic sampling: an approximation to random sampling in which you select one in a certain number of sample elements; the number is from the sampling interval n sampling interval: the size of the sample frame over the sample size, used in systematic sampling to select units 20

Stratified Sampling n Sometimes researchers want to include specific kinds of diversity in their sample, e. g. , racial diversity n stratified sampling: a type of random sampling in which a random sample is drawn from multiple sampling frames, each for a part of the population n Because you control the relative size of each stratum rather then letting random processes control it, you can be sure your sample will be representative of strata. n In general stratified sampling results in a slightly m representative sample than simple random sampling 21

Selecting a Stratified Sample n Step 1: You divide population into subpopulations (strata) To use this method, you must have info about strata in population (i. e. , the population parameter). n Step 2: You create multiple sampling frames, one for each subpopulation n Step 3: Next draw random samples, one from each sampling frame 22

Cluster sampling n In some situations where there is no good sampling frame, you can use multiple-stage sampling with clusters n A cluster is grouping of the elements in the final sample that you are interested in n cluster sampling: a multistage sampling method in which clusters are randomly sampled, and then a random sample of elements is taken from sampled clusters 23

THREE SPECIALIZED SAMPLING SITUATIONS n Random-Digit Dialing (RDD) n Within-Household Sampling n Sampling Hidden Populations 24

Random-Digit Dialing n random-digit dialing: computer based random sampling of telephone numbers 25

Within-Household Sampling n A household can be thought of as a cluster in which there can be multiple sampling elements or individuals n To ensure random selection, create selection rules, and follow them consistently 26

Sampling Hidden Populations n hidden population: a group that is very difficult to locate and may not want to be found and is therefore difficult to sample 27