Overview of Sampling
Why sample? We sample for two main reasons. 1.) we sample because it is usually impossible to sample the entire population; also, it is not practical as we have finite resources 2.) it is unnecessary to sample the entire population due to probability.
Representativeness Representativeness refers to how much your sample looks like the population from which it was drawn.
Generalizability If your sample is representative, you can generalize your findings to the population from which the sample was drawn; a desirable attribute of research
Types of Sampling; Strengths and Weaknesses
Sampling can be done either by probability or non-probability methods.
Probability methods include simple random sampling, systematic sampling, stratified random sampling, and cluster sampling.
1. simple random sampling - each element has an equal chance of being selected; can be done with or without replacement
2. systematic sampling - every nth element is selected
3. stratified random sampling - relies on the distribution of the characteristics of a population (often demographic characteristics such as age, gender, race) in order to select a sample that is representative of these characteristics
4. cluster sampling - sample is collected in 2 or more stages
Non-probability sampling includes purposive samples, convenience samples, quota samples, and snowball methods.
1. purposive - similar to convenience sampling; the sample is chosen based on the researcher's needs
2. convenience - used when samples are convenient; usually cheap and easy to obtain these subjects
3. quota - the non-probability analog of stratified probability sampling
4. snowball - may be used when subjects are difficult to obtain; once 1 subject is picked, subjects can then refer more subjects to the study
Why use probability sampling? Probability sampling reduces error, reduces bias, and allows us to calculate sampling error. Probability sampling allows for representative samples as opposed to non probability sampling designs which does not include equal probability of selection methods.
Overview of Research Design
Experimental Design
Classical design - 3 components: 1.) manipulation of the independent variable; 2.) post and pre-test; 3.) random assignment; subjects are divided into at least 2 groups, the experimental group and the control group; there is a minimum of two time frame
Experiments are strong in internal validity but may be weak in external validity because an experiment is a highly controlled environment.
Posttest only/ after only design - here, the classical design is cut in half by removing the initial time frame; since the initial time frame is removed, we can no longer be sure that the experimental group and the control group are indeed statistically significant prior to exposing the experimental group to the independent variable
Factorial design - when there is more than one treatment being applied; for example, different contact amounts with probation officers and its effect on recidivism
Quasi-Experiments
A quasi-experiment includes most, but not all, elements of an experiment. This can be used when random assignment is not achievable. A quasi-experiment meets the first 2 components for causality, empirical association and temporal order. A quasi-experimental design may be used when ethics, time, or money make an experimental design impossible.
3 types: nonequivalent group design, cohort design, time series design
Nonequivalent group design - here, the term "comparison group" is used over "control group"; the comparison group does not serve the same function as the control group in an experimental design; you might ask how we select important variables to match on and that answer is not an easy one to answer; as a general rule, the two groups should be comparable in terms of variables that are likely to be related to the dependent variable being studied
Cohort designs- for example, probationers that were sentenced in May with community service versus probationers sentenced in April without community service; the May group can serve as the experimental group and April can be treated as the comparison group
Time series design- longitudinal studies
Non-Experiment
Meets 1, possibly 2 criteria for causality, depending. If questions are asked to establish time order, we might meet 2 criteria for causality.
Criteria of causality
Criteria of causality - empirical association, temporal order, non-spuriousness
Threats to Internal Validity
history- historical events outside the study
instrumentation- concerned with changes in the measurement process itself
temporal order- when there is ambiguity about the time order of the experimental stimulus and the dependent variable
testing effect- if people are aware they're being tested, they may alter their behavior
selection bias- for example, if all subjects are volunteers, this may change results
statistical regression (regression to the mean)- occurs when statistical outliers move closer to the average
maturation- refers to change within the individual subject; may grow older and responses change
experimental mortality- aka attrition; people die, drop out, or becoming 'missing'
demoralization- people feel deprived and drop out
diffusion of treatment- subjects become aware of treatment and may experience spillover effects
compensatory rivalry- subjects in control group try to make up for not receiving treatment
compensatory treatment- people part of the study make up for treatment; similar to diffusion of treatment but intentional
Threats to External Validity
units- if the units are altered, will results change?
treatments- refers to the operationalization of the intervention; will the study's results change if the intervention is implemented differently?
outcome - refers to how the dependent variable is operationalized; will the results change if measures are taken differently
settings - the context in which the treatment is applied; will the studys results change when applied in a different jurisdiction by a different agency?
mechanism - the process by which the treatment produces the results within the study
Overview of Data Collection
Surveys/Questionnaires
- - surveys can be open or close ended
- - survey questions should be direct, umambiguous, and positively worded
- - for reliability, questions can be asked in different ways
- Interviews - Strengths & Weaknesses
- For one-on-one interviews, dress at the level or one above your audience.
- - strength: allows for probing
- Focus Groups - Strengths & Weaknesses
- - weakness: individuals may lead the conversation so the moderator must be cognizant of the situation
- Field Observations
- 1. Full participant
- - the researcher is a genuine participant in activities, or at least pretends to be
- - the researcher is covert
- - people see the researcher solely as a participant
- - researcher becomes part of the group without revealing his or her identity
- - strength: data is more reliable as people who do not know they are being watched are less likely to change their behavior and speech
- - weaknesses: collecting data accurately while fully participating may be a problem; difficult recording observations/ recalling information
- 2. Participant as observer
- - a researcher actively partcipates in activities
- - people know that research is being done
- - strength: being "known" as a researcher may be better than attempting to act as a colleague
- - weakness: reactivity; people know they are being researched and may change their behavior; the researcher faces "going native"
- 3. Observer as participant
- - the researcher does not actively engage in activity
- - an example of this is when a researcher sits in with a police officer on patrol
- - strength: in riding with police officers, citizens usually think the researcher is a plainclothes officer so the researchers presence does not contaminate the situation as much
- - weaknesses: "going native" is still a concern; for example, when researchers ride with police officers, they tend to become more sympathetic toward police officers
- 4. Complete observer
- - the researcher is detached
- - the researcher does not engage at all in social interaction
- - people do not necessarily know they are being observed
- - strengths: researcher less likely to go native and affect what is going on
- - weaknesses: less able to ask questions to better understand the situationå
No comments:
Post a Comment