Test #2 Flashcards
Three threats to causal inference and how to control against each?
o Does cause precede effect?—Pre and post testing
o Relationship between DV & IV? –Comparing pre and post testing
o Third variable problem?—Post testing and random assignment
Classical experiment
Research design with 3 components- pre/post test, experimental/control groups and random assignment
quasi-experiment
- Research design that includes out, but not all, elements of an experimental design
- does not use random assignment
Experimental group
Subjects in experiment who receive the experimental treatment
double blind
Neither participants nor researcher know which group is experimental or control
Placebo
Drug administered that is not actually a drug to eliminate
Threats to internal validity (12)
- History
- Maturation
- Testing
- Instrumentation
- Statistical Regression
- Selection Bias
- Experimental Mortality
- Causal time order
- Diffusion of treatments
- Compensatory Treatment
- Compensatory Rivalry
- Demoralization
History
- External events can effect study results
- Ex. Terrorist attack, tsunami, death
Maturation
- People constantly growing
- Ex. young kids are deviant = mature and grow
Testing
- Process of testing & retesting
- ex. people change their answer to what they want to hear after seeing it already
Instrumentation
- Changes in measurement process
- ex. update survey and try and compare with old data using old surveys
Statistical regression
- Extreme scores regret to the mean
- ex. extreme scores will usually go back down to the mean
Selection bias
- Way in which subjects are chosen
- ex. people choose groups, experimenter selects place convenient for them
Experimental mortality
- when subjects drop out prior to experiment completion
Causal time order
- IV or DV come first, may only collect data at one time point
- Check which comes first
Diffusion of treatments
- Experimental group may pass on elements to control group, have contact w/ eachother
compensatory treatment
- control group deprived; researchers compensate
compensatory rivalry
- control group deprived; may compensate by working harder
Demoralization
- Feelings of deprivation among control group affect results at post-test
Non-equivalent groups design
- Select subjects in a way that makes the 2 groups as comparable as possible
time series deigns
- A type of quasi-experimental design where changes in a DV are monitored over some time period
Cohort design
- Group of subjects that enter or leave an institution at the same time
- ex. groups of people who graded in may
Case oriented research
- Many cases are examined to understand a small number of variables
variable-oriented research
- Large number of variables are studied for a small number of cases or subjects
Representative sampling
- All population members have known & = chance of being sampled
Probability Sampling
- Sampling in which the probability that en element will be included in a sample is known
Non-probability sampling
- Sampling which the probability that an element will be included in a sample is not known
Equal probability of selection method (EPSEM)
- representativeness achieved if sample charac. = population charac.
Standard error
- how far away we find scores device away from the mean
Confidence intervals
- how confident you will get results near the mean range
4 types of non-probability sampling
- Purposive/judgemental sampling
- Quota sampling
- Reliance on available subjects
- Snowball sampling
Purposive/judgemental sampling
- select a sample based on our own knowledge of the population, its elements, and the nature of our research aims
Quota sampling
- use matrix or table to describe the characteristics of the population we want to represent
Reliance on available subjects
stopping people on a street corner or other location
Snowball sampling
ask one subject, then that subject tells others, and etc.
Survey
- A data collection method that applies a standard instrument in a systematic way to take measures from a large number of units
Matrix Questions
- ask respondents to select 1 answer against a range of related topics
- Ex. strongly disagree— strongly agree
Contingency question
- An answer to one questions prompts whether you answer the next or answer another question
Ex. If yes, answer ques 21
3 ways to administer surveys
- self administered questionnaires
- in person interviews
- Telephone surveys
Self administered questionnaires
- mail, computer online surveys,
cheap and easy, can allow for large number of responses - Not everyone will reply honestly
In person interviews
- ask questions orally
- More likely to get honest answers
- Costly, lots of time, not many people want to sit for it so small SS, people may feel uncomfortable talking about certain topics
Telephone surveys
- Researchers appearance will have no effect on answers
- low time and cost
- computer assisted telephone interviews , computers dial up phone numbers
3 rules of interviewing
- Appearance and demeanour: dress in fashion to those similar to who they are interviewing
Familiarity with the questionnaire: researcher should not stumble
Probing for responses: make sure probes are neutral
Qualitative interview
- A verbal interaction between an interviewer and a participant that follows a general plan of inquiry, but not necessarily a specific set of questions
3 main features of qualitative interview
- Richness, meaning and shared cultural views
- Critical realist approach
- Platform for creating questions
Critical realist approach
- A philosophical view that reality exists, but knowledge is constructed through multiple meanings
3 types of qualitative interviews
- structured/semi-structured interview
- Unstructured interviews
- Focus group interviews
Natural focus group
- individuals in group may already be acquainted or have an existing connection
Artificial focus group
- People selected according to criteria
unstructured interview
- most open style of interviewing, no guidelines
focus group
interview groups of people at same time
Tree-and-branch
main topic with branching questions
river-and-channel
main streams of questioning that lead into the main channel
4 main steps of gaining access to formal organizations
- Sponsor, find sponsor in related field
- Letter, email, tell what you will do for study
- phone call and follow up
- Meeting, further explain intentions and how this will benefit them
ethnography
- detailed and accurate description (what and how)
4 roles field observer could take on
- complete participant
- participant-as-observer
- observer-as-participant
- complete observer
complete participant
- Researcher told unknown to subjects; participate fully as member of studied group
- ex. study hells angels, become hells angels = break law?
- “Going native”= acting like group you study
Participant-as-observer
- Researcher role known to groups subjects; still participate as member of studied group
- Participants may act differently
Observer-as-participant
- researcher role known to subject; do not participate as member of group studied
- People may act different
complete observer
- researcher role (& often identity) unknown; observes w/o participating w/ subjects
- May not fully understand process
4 ways to record observations during field research
- note taking/tape recording: interviewing or dilation of observations
- Photos/videos
- field notes: field journal
- Structured observations: observers mark closed ended forms
Strength of field research
o High validity
o Depth of understanding
o Flexible
o Better for measuring behavior
Weaknesses of field research
o Not everyone can be studied this way
o Reliability- not asking everyone same ques
o Not generalizable
o May not represent entire population
Obtrusive measurement
- People are aware data are being collected
Unobtrusive measurement
- People are unaware that data are being collected
Agency records (3 subtypes)
- Collect vast amount of crime and CJ data
- Published ex. stats
- Nonpublic agency records ex. study internally (police officers)
- New data collection by agency staff
Content Analysis
- Examines class of social artifacts (any recorded document)
- Not created for research but can be examined
- Ex. blink video
Secondary Analysis
- Analyzing data collected by other researchers
“hybrid source”
- Combines collection of new data by agency staff w/ day to day agency activities
Manifest content
- visible, surface content vs. ex. blinks
Latent content
- Underlying meaning, ex. lying
Descriptive statistics
- Computations that describe either the characteristics of a variable or that relationship among variables
Inferential statistics
- Computations for making inferences from findings based on sample observations to some larger population
Univariate
examines distribution of cases only one variable at a time
Bivariate
two variables involved
Multivariate
examines relationships among several variables
Central tendency
- Clustering of observations to a midpoint in a distribution
- Mean (average), Median (middle attribute) and Mode (most frequent)
Dispersion
- Distribution of values around some central value, such as average
- Range, Standard deviation, CI
Rate
- Standardize some measure for comparative purposes
- Allow us to make more meaningful comparisons
Proportional reduction of error
- Used to summarize the association between two variables
Lambda
- Represents the reduction in error as a proportion of the errors that would have been made based on the overall distribution
- used for two nominal
Gamma
- Computed from 2 quantities: # of pairs that have the same ranking on the 2 variables and # of pairs that have opposite ranking on same variable
Pearson product moment correlation
- Used for interval and ratio
- guessing the value of 1 variable by knowing the other
Levels of significance
- probability of the measured associations being due only to sampling error
Chi square test
- Predict the joint distribution that would be expected if there were no relationship between the 2 variables
Statistically discernible difference
- considered statistically significant if found in a random sample
Hawthorne effect
- Individuals modify an aspect of their behavior in response to theis awareness of being observed
