research methods exam 1
refutable or falsifiable
(Applies more to theories than to science per se, but worth mentioning here.) Theories or other conclusions drawn as the result of scientific investigations must lend themselves to further testing.
Evaluation research
(programs, interventions, policies) •Effects of intervention •Does a new drug-rehabilitation program reduce probationers' drug use? •Does a cognitive-behavioral treatment program reduce prisoners' recidivism rates? •Do residency restrictions prohibiting sex offenders from living near schools reduce those offenders' recidivism rates? •Do tighter restrictions on the purchase and carry of handguns reduce gun-related crime?
Descriptive research
- Define and describe, count• How many youth become involved in gangs? •How many suspects shot by police Tasers get injured? •What percentage of female offenders have experienced abuse?
Common "Sense" / anecdotal
- Much of what we (think we) know about the world is observational and experiential - This knowledge helps us navigate our lives day to day •That guy is a jerk, so I stay away from him •That store has good prices, so I shop there
The Nuremberg Code's 10 Points
1. Informed consent. Participant must know that s/he is the subject of a research experiment and must be informed of all possible risks. S/he must provide the researcher affirmative consent of his/her willingness to participate. 2. The research must have potential to yield fruitful results and benefits to society, which would not be obtainable without use of human subjects. 3. The experiment should be based on previous research that provides some initial indication of expected impact on human subjects. 4. The experiment must minimize all possible physical and mental harm. 5. No experiment should occur where death or serious injury is a foreseeable outcome. 6. The risk to the participants must not exceed the anticipated benefit to society. 7. Preparations and facilities must be adequate to minimize risks. 8. Experiments should be conducted only by scientifically qualified personnel. Participants must be permitted to withdraw from the study at any time. The researcher must constantly monitor subjects for signs of adverse consequences, and remove subjects or terminate the study should a serious risk arise.
3 stages of rq formulation
1. identification What we've already talked about: within a researcher's field of study, unique contribution, builds upon existing knowledge 2. refinement Work a general question down to a manageable question or hypothesis Not too narrow, not too broad Don't paint yourself into a corner; give yourself an "out" in the event that your findings are unexpected Ex. Too board- is informal control effective? (hundreds of articles) Ex. too narrow- does informal social reduce rates of burglary in my town? (miss important literature) Ex. just right- is informal social control more effective in reducing crime rates than policing? 3. evaluation Feasibility "this is a good rq, but do I have the time and resources to gather the data?" Social importance "this is an interesting rq, but does it have relevance to policy? To theory?" For funded research, "what are the direct policy or other public benefits of this research?" Scientific relevance researcher should already know this, but if it is a new area of study for him/her, then additional research might be needed to ensure the question is meaningful
THERE ARE NO "RULES" FOR WHAT TO STUDY, BUT THERE ARE SOME BASIC REQUIREMENTS
1.THE RESEARCH QUESTION OR HYPOTHESIS MUST BE NEW. MUST OFFER A UNIQUE CONTRIBUTION TO EXISTING KNOWLEDGE•IF WE ALREADY KNOW THAT OFFICERS' RACE DOESN'T AFFECT THE AMOUNT OF FORCE THEY USE, THEN WE DON'T NEED MORE STUDIES EXAMINING OFFICER RACE AND FORCE•IF WE ALREADY KNOW THAT SERVING IN THE MILITARY GENERALLY IS NOT RELATED TO CRIME, THEN INSTEAD WE SHOULD EXPLORE IF CERTAIN SERVICE CHARACTERISTICS (E.G., BRANCH, COMBAT EXPERIENCE, LENGTH OF SERVICE) ARE RELATED TO CRIME.•THIS IS A NEW IDEA! 2. The rq or hypothesis must build upon existing knowledge in a logical manner It has to be new, but not totally out of left field No: "among college students, do cj majors differ from non-cj majors in their views about police?" Yes: "use-of-force Studies have considered officer race and suspect race, but none have analyzed the combination of the two" Yes: exploring specific service characteristics and crime.
Crime mapping and analysis
A form of secondary data analysis. Using location-specific crime incident data to create geographical pictures of crime distribution, and to determine whether certain crime events seem to affect other ones•e.g., create a map of robbery hot spots•e.g., find out whether a burglary committed at one location means that nearby houses are at risk
Sampling: A Valid Cross-Section
A good sample is a cross-section of the population from which it came Even well-intentioned sampling methods can break down and result in a non-representative sample The population I am studying is 30% Latino, but I send out my survey in English only I am interested in criminal victimization, yet I sample householders only, leaving out transient individuals and homeless persons, who have high rates of victimization I mail a survey to a sample of people in a city with a median age of 46, but the median age of my sample is 67
Target population
A set of elements larger than or different from the population that was sampled and to which the researcher would like to generalize any study findings
Dichotomous (Binary) Variables
A special subset of categorical variables Some dichotomous variables are nominal (male, female) Others do suggest some level of quantification (have used illegal drugs, have not used illegal drugs) Dichotomous variables will generally be treated the same as two-category nominal variables
Sample
A subset of elements from the population
Concepts
Abstractions/Ideas
Measurement Validity: Construct Validity
Another way of assessing one item's validity by comparing it to another item, but this time it's a different item e.g., I know that the use of one kind of serious illegal drug (e.g., heroin) often co-occurs with the use of other kinds of drugs (e.g., marijuana, alcohol). I can look for expected relationships between consumption of one illegal drug and consumption of others "Showing a measure is related to a variety of other measures as specified in a theory" (p. 106) Another example: Researchers comparing responses on the Addiction Severity Index (ASI) to other indicators of substance abuse, such as employment problems, family problems, and medical problems (McLellan et al., 1985). Did not have access to laboratory tests or observer reports, but extensive research has shown that these problems were related to substance abuse
Interitem Reliability
Applicable to multiple-item scales only (i.e., not single items) If multiple items tap into the same construct, then the items should be highly correlated with each other Everybody who scores high on one item should score high on the others Everybody who scores low on one item should score low on the others
Protecting Research Participants: Avoiding Harm
As we have seen, this is a grey area Some harm is clearly unethical Other harm is more subjective and raises questions about whether the value of the study outweighs the possible negative effects to participants e.g., Milgram's experiment
Policy motivations
Assess programs and policies to determine their success and develop ways to improve outcomes and better address problems
Nonprobability Sampling Methods: Availability Sampling
Availability sampling (aka convenience sampling): Selecting elements on the basis of their ease of access. Lots of research uses availability sampling Surveys of high-school students Surveys of college students Surveys of jail inmates Studies of inmates or probationers in treatment programs Characteristics of this type of sampling (see p. 131) Often, no population information is available Can't compare the sample to the population to determine representativeness Might not even know what the population actually is Generally, there is no sampling frame No systematic technique was used to select respondents Benefits Researcher needs a low-cost method for conducting initial studies so that s/he can apply for a grant to fund a large-scale project Researcher can target a specific group Particularly useful for low-prevalence behaviors (e.g., gang membership, serious drug use) Drawbacks No way to assess representativeness Limited generalizability Potential for serious systematic error to bias the sample In particular, self-selection bias
peer reviewed
Before a study is published in an academic journal, other scientists evaluate its methods to ensure proper scientific technique.
Constructing Questions: Closed-Ended
CE questions should be: Mutually exclusive: A respondent falls into one and only one category. Exhaustive: Every respondent has a category to choose. Mutually exclusive and exhaustive: "What is your annual household income?" $9,999 or less $10,000 - $34,999 $35,000 - $74,999 $75,000 or more Not mutually exclusive: "How old are you?" 18 or younger 18 - 25 26 - 30 31+ Not exhaustive: "What is your race?" Black White Each item should contain one and only one concept No: "Do you think that police officers who use excessive force should be criminally and civilly prosecuted?" No: "Have you ever been sentenced to jail or probation?" Confuses the respondent, s/he doesn't know what answer to provide, and reduces the quality of the research because the researcher doesn't know how to interpret respondents' answers
Closed-ended questions
Closed-ended questions: Questions that allow respondents (participants) only one answer option "Have you ever been treated unfairly by a police officer?" Yes No "How many alcoholic beverages do you consume each week?" _____ "The best way to punish criminals is to put them in prison." Agree Disagree Most often associated with quantitative research
secondary data analysis
Collection of existing administrative or survey data•e.g., unemployment data from the census bureau and crime data from the Uniform Crime Reports
Research Ethics
Concerns with ethics date back to 5 BC Nuremberg War Crime Trials Experiments conducted in the name of "science" Formal procedures Tuskegee Syphilis Experiment (1932-1972) Public-health officials followed 399 black men in Alabama who had syphilis Didn't tell them they had the disease, and didn't offer them a cure (penicillin was proven effective in 1940s) Many participants died, many suffered other effects of the disease. Their wives and children contracted it Experiment only ended when a leak to the press caused public outcry Today, we have laws and regulations governing research and preventing people from being exploited or abused But much research continues to have ethical implications As biosocial researchers discover the genetic differences between criminal offenders and non-offenders, what will be the implications for sentencing? We know that poor, single mothers are at high risk of producing children with behavior problems. Do we force these mothers to accept outside assistance raising their children? Do we put them on mandatory birth control to prevent more children?
Sample Planning
Define the population, sampling frame, elements, and (if applicable) units Sampling frame is the data source used to pull information about all people in the population and to construct the sample Elements are the people, places, or objects to be sampled Units are proxies that researchers might use if it is impossible to get information directly from the elements (see p. 117 for an example). Example: Binge drinking among criminology majors at UF Population: All criminology majors at UF Sampling Frame: Obtain a list of criminology majors from the registrar's office Elements: Individual members of the sample Evaluate generalizability In reality, sampling usually has certain limitations that reduce or constrict generalizability Hard to get a sample of youth that represent high-school students nationwide, or a sample of prisoners that is representative of all prisoners in the country Researchers need to be clear and honest about how their sampling methods might limit generalizability Sample generalizability (generalizability): Ability to make a statement about the population from which the sample was derived Sample of high-school students from a single school all high-school students in that district Cross-population generalizability (external validity): Ability to make a statement about other populations Sample of high-school students in a single school high-school students in the neighboring district We will primarily focus on how the findings from a sample can be generalized to the population from which it was drawn. Evaluate generalizability, continued All samples will have some level of sampling error either due to the inherent nature of sampling or to choices (or errors) made by the researcher Random error is white noise; background variation that is inevitable in sampling Your population is 51% female, but your sample is 55% female; your population experiences an average of .90 victimization incidents per year, but your sample experiences .55 Not much a researcher can do about this. A researcher using proper methods will still probably not get a perfect representation of the population. Systematic error is caused by an error or limitation in the sampling strategy that either includes or excludes disproportionate numbers of certain groups Your population is residents of a particular city, but you use drivers-license records as the sampling frame from which to pull addresses and phone numbers You send out surveys in English only to a population with a large portion of non-native English speakers You decide to use online surveys instead of mail or phone surveys Assess population diversity Know your population whenever possible The U.S. Census Bureau is a good source of information on demographics and social facts about cities, counties, and census tracts Compare your sample to the population to determine representativeness Consider a census A census includes all members of the population Usually not feasible, but worth considering sometimes
Survey research
Designs involving the administration of questionnaires to a randomly selected sample•e.g., mail surveys to a random sample of city residents asking them about their opinions of local police•Your research project in lab will use survey research
Quasi-experimental designs
Designs that mimic experiments but are missing an element•e.g., matching each sex offender who did the community program to a similar sex offender who served prison time instead
Measurement Validity
Did you measure what you thought you measured? Is your empirical variable truly a good indication of the concept the variable is supposed to represent? Validity: Accuracy. The extent to which an argument, formula, variable, etc. is logically sound and correct. Sometimes, validity isn't an issue (age, race, sex) But there are lots of times when a researcher must select a way to operationalize a concept. This is where measurement validity becomes important I want to study the relationship between unemployment and violent crime. I gather a sample of cities. For each city, I get unemployment data from the census and crime data from the Uniform Crime Reports I measure crime as the number of Part I violent index offenses (homicide, robbery, rape, aggravated assault) per city How accurate is this as an estimate of the true amount of violent crime happening in that city? The UCR Captures only crimes that come to police attention, either through officer's intervention or victim/witness reports Hierarchy rule means that in a multiple-crime incident, only the most serious offense gets reported How valid is the UCR, then?
Measurement Validity: Content Validity
Does the measure "cover the full range of the concept's meaning?" (p. 105) Not an issue with simple things like age, race, sex But very important when concepts are complicated and multifaceted A researcher wants to know about people's attitudes toward the police No: Single item "Do you like the police?" Yes No (no face or content validity) Yes: Multiple-item battery of questions Recall previous discussion regarding scales; this is the same thing Much CCJ research focuses on people's assessments of the social environment of their neighborhoods. How well neighbors get along with each other, and whether they would help each other "Do you get along with your neighbors?" Yes No is inadequate Researchers use multiple-item indexes to gather the data. Once they have the data, they sum these items to form single scales Criminal Thinking Scale (from last lecture) Attitudes-toward-police scale Social-cohesion scale These scales thus represent multi-faceted composites of complex concepts. They have content validity Another example: The Michigan Alcoholism Screening Test includes 24 questions which experts have agreed capture the full range of possibilities of direct consequences of substance abuse
replicable
Each study can be conducted again by another researcher to determine if the results are the same.
Justice
Equitable distribution of research risks and benefits; Treat people fairly
Content analysis
Examination of media for themes•e.g., analyze all crime-related local TV news stories reported in one year to find out if particular types of crimes are being exaggerated
Elements of a Research Article
First, the overarching purpose or point of the study. Two parts to this What the researchers are studying Why they are studying it Have to identify the contribution that the article is making to the study of criminal justice or criminology Answer the "who cares?" question Identify the research question Review the current state of the research third, identify the data the researchers used and the methods they employed to collect that data Surveys? Official crime reports? Etc. Where did they get their respondents? Where did they pull the crime reports from? Etc. fourth, findings from the analysis You won't understand the stats and that's fine IN the Results section, the authors will explain the results in words, so you don't have to be able to read the numbers to know what they found What you need to pull from this is the answer to the research question That is, the main finding Finally, the discussion/conclusion We won't spend a lot of time on this; our main concern stops at the answer to the research question But the discussion section will help you understand the results (especially if you didn't understand what the authors wrote in the Results section)
So putting it all together....
For my dissertation, I explored the association between military service and criminal behavior among incarcerated military veterans Concept being examined: Criminal behavior Conceptualization: Lifetime arrests, and type of offense committed Operationalization: Lifetime arrests: Self-reported number of times ever been arrested; Offense type: Self-reported offense type
Reliability and Validity
Ideal is to have both Statistical tests can help determine whether a given measure is valid, but if the measure appears invalid after the data has been collected, than little can be done to correct the situation. A valid measure should also display reliability If the measure accurately captures what it is supposed to, then it should yield the same results across different times and raters But reliability can exist without validity A bad measure can yield consistent results, even if those results are consistently wrong You weigh 160 pounds You step on a scale and it says 160—validity Today and it says 160. You measure yourself again tomorrow and it says 160—validity and reliability Each time you step on it, it says 175—reliability, but no validity Over five days, you get 140, 190, 165, 178, 156—no validity, no reliability To enhance both, researchers generally rely upon measures that have been used in the past Never advisable to blaze your own trail—if a measure already exists and has been proven to display validity and reliability, stick with it A researcher who develops his/her own measure must demonstrate its quality. Can never simply say "Here it is" without offering justification
Units of Analysis
Important to remember that study results are limited to the unit of analysis used in that study Sample of prisons cannot be generalized to jails Sample of police departments cannot be generalized to prosecutors' offices or parole agencies Sample of cities cannot be generalized to countries Unit of observation: Unit from which data were collected. Often, but not always, the same as the unit of analysis I want to know about the institutional environment of prisons, so I survey prisoners in 10 prisons and average their scores together to get an idea about what life is like in each facility. Unit of observation: individual; unit of analysis: prison
personal motivations
Improve society, solve a problem of personal interest
Respect for persons
Individual autonomy; Protection of individuals with reduced autonomy
Elements
Individual members of the sample
Protecting Research Participants: Informed Consent
Informed consent must be obtained by each participant before that person can be entered into the study Researcher must disclose the nature of the study, the participant's expected role/contribution, the duration of the study, the participant's right to quit, etc. (see pp. 72-74) If the study involves deception, the researcher will have to justify to the IRB that the informed consent document not disclose full details Part of consent is in informing people that they are part of a study Researchers generally cannot go in "undercover" 1960s: Researcher seeks more information about men who engage in homosexual acts in public restrooms. Policy at the time was to arrest them, assumption was that they were public nuisances. Researcher wanted more information about them, in hopes of helping to shape a more productive policy response. He went undercover as a "lookout." Recorded license plates of some and showed up at their houses posing as a health-services interviewer and asked them questions about their marital status, sexual orientation, etc. Found that vast majority actually lived conventional, upstanding lives. Many were heterosexual. Most had children. Contradicted stereotypes and, arguably, led to police deciding not to waste resources on this victimless offense. But "participants" never knew they were being studied. Only adults (18+) who are of sound mind (not mentally retarded, severely mentally ill, etc.) can provide informed consent For others, parents or legal guardians give consent, and the participants, if possible, are asked for assent Special populations (prisoners, pregnant women, individuals with cognitive impairments, etc.) are given additional protections, though are usually permitted to provide consent
Continuous variables
Interval Ratio
Organization of a Research Article
Introduction Literature review Methods Results Discussion/conclusion
Probability Sampling Methods
Known, non-zero probability of selection Chance alone dictates who is and is not selected; conceptually no different than flipping a coin for each element of the population to see who gets and doesn't get selected All else being equal, larger probability samples are better than smaller ones The minimum size of the sample depends upon the size of the population A good sample from a population of 1,500 might be 200; with a population of 100,000 you might want a sample of 1,000 However, a large sample will not compensate for bad sampling methods
academic motivations
Learn more about complex social phenomena to better understand crime and society's response to it
Beneficence
Maximize benefits and minimize harms
The role of theory
Most research is grounded in some theory, though some more elaborate than others Some research tests theories (e.g., "are people with low self-control more likely to commit crime?") Other research is grounded in theory even though it doesn't test it directly we know that crime is more common in socially disorganized neighborhoods, so is police misconduct also more likely to occur in these areas?") We know that inmate subculture is reflective of the offender subculture on the streets, so does military experience shape the prison experience among veteran inmates?
Sample Planning: Methods and Generalizability
Most samples have some limitations Respondents to a general-population survey are mostly older and white Only people age 18 and older were eligible to participate Researchers studied prisoners who were convicted of identity theft Some level of nonrepresentativeness is not inherently fatal to a study. A key is that the researcher be clear about which groups were—and were not—represented (i.e., what the population is) If the sample contains only 18+, then the population is adults Some types of nonrepresentativeness (or reasons for it) are worse than others
Justifying Research
Much research carries some level of risk. The key is whether the risk is as minimal as it can possibly be, and whether the risk is outweighed by potential benefits Those benefits can only be arrived at when the research is valid and objective "It is the pursuit of objective knowledge about human behavior... that motivates and justifies our investigations and gives us some claim to the right to influence others to participate in research" (p. 66) Some risk-benefit balances are obvious No to Nazi experiments No to Tuskegee experiment But others are trickier Some fascinating insight into human thought and behavior has been achieved by research that wouldn't pass IRB review today Milgram's obedience to authority experiment Zimbardo's Stanford prison experiment "I wanted to know who wins-good people or an evil situation-when they were brought into direct confrontation" (p. 66) These were carefully controlled studies, no glaring errors and no reason to think their results aren't valid (or are completely invalid). Is the risk and damage to participants justified?
research methods
NVOLVE THE PROCEDURES USED TO COLLECT DATA•THE STUDY DESIGNS APPROPRIATE FOR CERTAIN SITUATIONS•THE SCIENTIFIC RULES THAT RESEARCHERS MUST FOLLOW TO GATHER VALID, RELIABLE DATA•STATISTICS IS THE MATHEMATICAL FIELD DEVOTED TO THE ANALYSIS OF THE DATA ONCE IT IS COLLECTED
Categorical variables
Nominal Ordinal
Limited or No Generalizability
Nonprobability methods create samples that are not representative of populations Or, at least, representativeness cannot be assessed Therefore, nonprobability samples cannot be generalized to populations A researcher talking about her/his findings must keep all conclusions confined to the sample No: "Returning prisoners have problems staying away from drugs and alcohol during their transition back into society." Yes: "Returning prisoners in this sample had trouble staying away from drugs and alcohol during their transition back into society."
Nonprobability Sampling Methods
Nonprobability samples: Elements of the population have unequal and/or unknown probabilities of selection There are times when resource limitation or the specific type of research being conducted require nonprobability methods The downside of nonprobability sampling is that "because the chance of any element being selected is unknown, we cannot be certain the selected sample actually represents our population" (p. 130) This means that generalizability suffers The benefit of nonprobability sampling is that it is more flexible than probability sampling is Resource limitations prevent full use of probability sampling Research question is exploratory; researcher wants to conduct initial study Researcher is interested in a specific subset of the population, not in the population itself Qualitative research
Major Laws in Research Ethics
Nuremberg Code and Belmont Report have shaped federal and state laws Universities also have requirements of their own and requirements that federal and state laws be observed Each university has an Institutional Review Board (IRB) that oversees all research Everybody wishing to conduct a research study must receive IRB approval first
Why Operationalization Matters
Often better to avoid using the names of the concepts themselves No: "Do you binge drink?" Yes: "Have you consumed five or more drinks in one sitting?" No: "Have you participated in prison misconduct?" Yes: "Since your admission, have you been written up or found guilty of breaking any of the prison rules? Question: What percentage of Americans support the death penalty? Concept being examined: Death-penalty support Measurement strategy: Survey self-report Operationalization: How to word the question? Two options for operationalizing this concept "Do you support the death penalty for people convicted of murder?" Yes No "Which penalty do you think is best for people convicted of murder?" Death penalty Life in prison without the possibility of parole Life in prison with the possibility of parole So, what percentage of Americans support the death penalty??? The "answer" I get depends on the way I ask the question Operationalization matters because it can (dramatically) impact the results of the analysis and the conclusions a researcher draws from the study Example: National Crime Victimization Survey (NCVS), Supplemental Victimization Survey(SVS) Some responses regarding "repetitive and unsolicited communications" were incorrectly classified as stalking or harassment Were actually spam email, internet scams, and contacts from telemarketers or bill collectors When these cases are excluded, the estimate of persons who experienced behaviors consistent with either stalking or harassment dropped to 5.3 million from 5.9 million
Operationalizing Concepts
Once a concept is clarified, we must decide the way to measure it Operationalization: The process of selecting an empirical variable to represent a concept Good conceptualization and operationalization can prevent further confusion later Basically, translating an idea into a variable Unemployment: % unemployed persons in a county Crime: Number of property crimes (burglary, larceny, MV theft) per 10,000 population Crime: Self-reported number of times ever been arrested Gang: Group of 3+ youth who have designated signs, colors, and turf
Test-Retest Reliability
One measurement administered a two time points Applicable only if the concept being measured has not changed OK to administer the same math test twice for someone who is not taking a math class Not OK if that person is enrolled in a math class
Ordinal Variables
One step up from nominal in complexity - categories that can be ranked "What is your year in college?" Freshman Sophomore Junior Senior "How old are you?" 17 or younger 18 - 35 36 - 55 56+ Does allow some "more than" or "less than" judgment, but not much Knowing that one person is a senior and another is a junior doesn't allow me to calculate how many more credit hours the senior has completed A person in the "36-55" category is older than someone in the "17 or younger" category, but I can't do any math to figure out how much older Another example: Level of agreement with a statement from strongly agree to strongly disagree
Constructing Questions: Open-Ended
Open-ended questions: Questions that do not have response options and that instead allow respondents to elaborate on their thoughts either verbally or in writing. Associated with qualitative research "Tell me about your most recent contact with a police officer." "How do you feel about the death penalty?"
Ethical
Pertaining to moral principles of right and wrong Researchers must always consider ethical questions, both of the research project itself (e.g., effect on participants) and for society (e.g., implications for policy) Ethical dilemmas arise in all fields of science
Nuremberg Code (1947)
Product of the Nuremberg Trials, which exposed the vicious torture Nazi doctors inflicted upon Jewish prisoners in the name of medical science Code has 10 points All points refer to human-subjects research Other codes and laws apply to research with microorganisms, animals, etc.
Honesty and Openness
Publication of research findings is critical to honesty and openness- nothing that a scientist does should need to be kept secret or covered up This is part of the importance of the "Methods" section Many funding agencies require the researcher to turn his/her data over to the funder at the end of the study period. This data is usually then made publicly available for other researchers to inspect and use
Nonprobability Sampling Methods: Purposive Sampling
Purposive sampling: Targeting people known to be knowledgeable about certain issues and/or able to speak for others Researchers may want to know about an attitude or event in a community, but they cannot or don't want to do a large-scale survey Generalizability might not even be important to a researcher—might be a case study or in-depth exploration of a specific phenomenon in a specific area Why do people living in a particular community often refuse to report crimes to police? What kind of interactions occur between gangs and community members? Are these interactions all bad, or do gangs sometimes help others out? How well do ex-prisoners with felony convictions return to society? What problems do they have adjusting? Researchers may target community organizers, religious leaders, social-service workers, etc. Anybody who has extensive first-hand knowledge about the phenomenon being studied and can reasonably be trusted to speak for others on the subject The key to the usefulness of this research (what separates this from polling and anecdotes) is the quality and quantity of the people who are selected for inclusion Good: 30 people with at least 5 years' experience as outreach workers Not good: 10 people who live in the area
Interval Variables
Quantitative/numerical. Rank orderable. A more precise way of measuring quantity (i.e., "more than" and "less than") Attitude scales ("How do you feel, ranging from 1 - 10?") Interval variables can be added and subtracted, though generally not multiplied or divided "Jane scored 2 points higher on the scale than Bob did" is fine, but "Jane's score was 3 times higher than Bob's" is not How does someone have "3 times" the attitude as someone else? Represent fixed measurement units but have no absolute zero point Has values of equal intervals that mean something More examples: Temperature in degrees Fahrenheit IQ (intelligence scale). SAT scores.
Brief Review: Quantitative vs. Qualitative Research
Quantitative: Records the variation in social life in terms of categories that vary in amount (numerical data) Explanatory, descriptive, evaluative, and sometimes exploratory Qualitative: Tests explanations of social phenomena using textual or observation data Generally used in exploratory research only
Nonprobability Sampling Methods: Quota Sampling
Quota sampling: Deliberately targeting people according to predetermined information about the population This is a small step up from availability sampling, in terms of rigor. Slight increase in the sample's representativeness of the population Not much! Method: use a basic fact about the population and seek to get a sample that reflects that fact Researcher studying gangs knows that 15% of gang members are female, so he seeks out 8 - 10 female gang members to interview in a sample of 50 gang members Researcher knows that approximately 50% of female offenders have been sexually assaulted, so she seeks a sample of 20 victims and 20 non-victims Benefits Resource limitations and preliminary research Greater diversity than pure availability sampling Ensure enough respondents of each relevant group to facilitate understandings and comparisons Drawbacks Limited representativeness, limited or no generalizability Continued potential for systematic error/bias
We also have to consider collateral consequences
Research showing that maternal drug abuse could cause permanent damage to babies led to laws requiring hospitals and clinics to drug test all pregnant women. Women who tested positive could be criminally prosecuted. Intention: Deter pregnant drug abusers from using Collateral consequences?
transparent
Researchers are open about their methods so that others can see and evaluate the correctness of the procedures.
The Uses of Research
Researchers, as scientists, (should) strive to be objective However, everyone has social issues that mean a lot to her/him personally Basic science is science done purely for discovery Is Pluto a planet? Applied science is science with a purpose—it's supposed to accomplish something useful and practical Vaccines Sometimes the goal of helping humanity can seem incompatible with scientific detachment and objectivity Zimbardo wanting to improve prison for both inmates and guards Milgram wanting to help us avoid falling in the trap of blind obedience to authority Weigl wanting to help Nazi victims and weaken Nazi forces Another question revolves around the amount of evidence needed to support a policy intervention before that intervention is put into practice How many white swans before the probability of a black swan is low enough? Part depends on the urgency of the problem, and part on the amount and type of harm required to effect a positive outcome After-school program that results in a 10% drop in youth gun violence Zero-tolerance policing that results in a 30% increase in youth arrests for curfew violations, and a 10% drop in car thefts by youth
Belmont Report (1979)
Response to the Tuskegee Syphilis Experiment Three major ethical principles (p. 64) Respect for persons Beneficence Justice
Choosing a Level of Measurement
Rule: Always use the highest level of measurement possible Criminal history "Have you ever been arrested?" Yes No "How many times have you been arrested?" 0 1-2 3-4 5+ "How many times have you been arrested?" ____ You can always make a continuous variable categorical later; you can never turn a categorical variable into a continuous one Go with the one that gives you the maximum possible amount of description/information/usefulness Example: Age questions from the introduction survey
systematic
Rules for the proper way to collect evidence
Unobtrusive Measures
SSO belongs to the class of unobtrusive measures. Data that can be gleaned about people's behaviors without direct interaction with participants. Archival research is also unobtrusive. Examining historical documents, published letters, media reports, etc.
Ratio Variables
Same as interval except that there is a true zero, which means that multiplication and division are possible Ratios can be formed between the numbers Number of times a person has been arrested "Jack has been arrested twice as many times as Jill has" Number of homicides in a city in one year "This city's homicide rate is double that city's" or "The homicide rate is down 50% compared to last year"
Sample Planning: Sampling Frames
Sampling frames can be diverse; key is that they contain contact information on the entire population Water department's list of all residential addresses in the city US Postal Service's list of all business addresses in the county Some sampling frames are not good Driver's license records Voter registration records A good sampling frame must not systematically miss identifiable segments of the population; a good sampling frame captures the entire population
Scales
Scales are useful because they have better construct validity (that is, they represent the entire spectrum of ideas within a concept) Procedural justice is about people's perceptions of fair treatment, respect, and unbiased decision making Therefore, a single item ("How well do you think the officer treated you?") is not good enough. The full range of the concept is not represented Another example Legal cynicism is the theory that certain groups in society hold skeptical or jaded views about the law and the CJ system on the basis of both their own negative experiences and the negative experiences of those within their social group This is a complicated concept. Contains elements of skepticism about the law itself (legislative motives and processes) and about the ways in which the law is enforced (police, courts) Legal cynicism operationalization "Criminal laws protect everyone equally, regardless of race or class." Agree Disagree "People in power try to use criminal laws to control people who don't have any power." Agree Disagree "Police officers treat everyone equally, regardless of race or class." Agree Disagree "Courts treat everyone equally, regardless of race or class." Agree Disagree Multiple-item indexes are also necessary to ensure you actually capture your intended conceptual target Drug use "How many times have you smoked marijuana?" Not an accurate picture of someone's drug use Need to ask about all sorts of drugs
Probability Sampling Methods: Simple Random Samples
Simple random sampling: The most basic probability sampling method. Random selection of cases from sampling frame. Each element in the population has an equal probability of selection. One option is to number each element of the frame and use a random-numbers table to select Random-digit dialing: A simple random sampling method where a sample of phone numbers is generated through automated generation of local telephone numbers (407) 321-???? Very useful, particularly when no sampling frame is available. Incorrect numbers can be replaced with new ones Characteristics of simple random sampling Known probability of selection Equal probability of selection All selections are independent; that is, nobody's selection depends upon the actions of anybody else in the sample Each element can be selected only once Random sampling is a popular and well-established method of obtaining representative samples But simple random sampling can be expensive and inefficient, particularly when a population is very diverse and/or no sampling frame exists or can be located. The researcher might find it more effective and efficient to manipulate the sampling procedure. Systematic Stratified Multistage cluster
Measurement Validity: Face Validity
Simplest validity assessment: Does it look OK? "Have you ever served on a jury?" seems fine Sensitive questions might have less face validity. "Have you ever been sexually assaulted by a partner or spouse?" Some measures might not seem to be closely related to the concept they are intended to operationalize. "Do you smoke cigarettes?" as a measure of self-control Another example: The question, "how many days per week do you smoke marijuana", has face validity if used to measure the variable, "marijuana use"
Probability Sampling Methods
Smaller populations require a larger fraction of the population to be sampled May need 20% of a 700-person population but only .1% of a 1 million-person population Small samples are untrustworthy To ensure large enough sample, increase the percentage of the population sampled as the population shrinks Also depends upon the homogeneity of the population A city population that is mostly white, older, and retired vs. one that is very diverse racially and economically Homogeneity means a smaller sample will likely suffice Heterogeneity means smaller samples carry high risk of missing certain demographic segments
Nonprobability Sampling Methods: Snowball Sampling
Snowball sampling: Getting recommendations from respondents about who else to interview; contacting new possible respondents on the basis of referrals from other respondents. Highly useful for hard-to-reach populations Gang members Active criminal offenders Prostitutes "Underground" populations are often interconnected, so a researcher can gain access to one person and then ask that person to vouch for her/him to gain access to others Benefits May be the only way for a researcher to gain access to multiple respondents from a secretive group May increase respondents' trust of the researcher, encourage thorough and honest responses Drawbacks Limited or no generalizability Might be hard to get much diversity in the sample—people tend to comingle with others who are like themselves
Operationalization: Scales
Sometimes effective operationalization requires use of multiple items (variables) that are later summed to form a scale Procedural justice "Did the officer treat you fairly?" "Did the officer take time to listen to your side of the story?" "Did the officer make a decision based on facts and law rather than on her/his personal opinion?"
Measurement Validity: Criterion Validity
Sometimes, a subjective score can be compared to an objective measure Think—a validation process e.g., "How much did you have to drink this evening?" and blood alcohol content - or - A probation officer checks self-reports of drug use with urine tests. This is usually not feasible in CCJ research: for concepts like attitudes, there is no criterion, and for other concepts (e.g., annual income) researchers cannot access criterion documents (e.g., tax returns)
Alternate-Forms Reliability
Sometimes, the wording of a question or the presentation of an answer-option index can influence the responses participants provide To test for this, a researcher alters the wording and/or presentation slightly to see if the results change Consistent results suggest reliability (i.e., nothing about the question is influencing the results) Inconsistent results suggest that the question is interfering with the results
Probability Sampling Methods: Stratified Random Samples
Stratified random sampling: Using information about the population to select a sample that is over- or under-representative of certain groups. Useful when the population has minority groups If my population is 70% white, 10% black, 15% Latino, and 5% other racial groups, then even a perfectly representative sample of 250 people would only have 25 black and 37.5 Latino respondents, and probably none from the "other" racial category. These numbers are too small to be useful in statistical analyses. In my population, 5% of adults have been arrested in their lives, and I am interested in their experiences. But in a sample of 200, only 10 would have been arrested. In these situations, researchers need ways of weighting their sampling methods to get larger portions of the small groups Proportionate stratified sampling: Adjust sample size of each group to get a representative sample. Decide in advance that you need to have a sample of 500 to contain 350 white (70%), 50 black (10%), 75 Latino (15%), and 25 other (5%) Use random sampling within each group to pull the predetermined numbers of people Disproportionate stratified sampling: Intentionally increase the proportions of the sample that contain minority groups (and reduce the proportion that is from the majority) Decide you want 25% each of black, Latino, white, and other-race respondents (100 of each group; 500 total) This is useful when a researcher is interested in something that might vary across the groups, and therefore needs large samples of each minority group Attitudes about the effectiveness of the CJ system Voting preferences and behavior Researchers using stratified sampling apply a weight during the statistics phase. Thus, the sample is appropriate for statistical analyses but still generalizable
Probability Sampling Methods: Systematic Random Samples
Systematic random samples: Randomly selecting a starting number, then selecting every nth case after that Useful when there is no sampling frame, or when it's impossible to number the cases for random-number generation Collecting a sample of criminal defendants processed by a large urban court in one year, where case files are stored electronically or in file cabinets Three steps (p. 127) Calculate sampling interval (#population/#sample) (1,000/50 =20) Randomly select a number from the interval as the starting point (1-20) Select every nth case after that, varying if interval is a decimal
CRIMINOLOGY
THE STUDY OF THE CAUSES OF CRIME•INDIVIDUAL FACTORS (E.G., LOW SELF-CONTROL, BIOSOCIAL CAUSES)•COMMUNITY FACTORS (E.G., HIGH UNEMPLOYMENT, GANGS)•CRIMINOLOGICAL RESEARCHERS ASK QUESTIONS LIKE, "WHAT CAUSES CRIME?" (EXPLORATORY) OR "DOES THIS PHENOMENA CAUSE CRIME?" (EXPLANATORY)
CRIMINAL JUSTICE
THE STUDY OF THE SYSTEM DEVISED TO DEAL WITH CRIME AND THE PEOPLE WHO COMMIT IT•POLICE, COURTS, CORRECTIONS •CJ QUESTIONS MIGHT BE "WHY DOES IT TAKE ABUSED WOMEN SO LONG TO LEAVE THEIR ABUSERS?" OR "WHAT ARE THE ELEMENTS OF THE POLICE CULTURE?" OR "DO JUDGES' RACE OR GENDER AFFECT THEIR SENTENCING BEHAVIOR?"•CJ RESEARCHERS CONSIDER CRIME, TOO, AND CRIM RESEARCHERS CONSIDER CJ QUESTIONS. THE DISCIPLINES ARE DISTINCT, BUT THE RESEARCHERS MOVE BACK AND FORTH3
Real World Example- CTS
Texas Christian University Criminal Thinking Scales (TCU-CTS) Survey The 6 CTS scales include entitlement, justification, personal irresponsibility, power orientation, cold heartedness, and criminal rationalization, which represent concepts with special significance in treatment settings for correctional populations.
Level of measurement
The amount of precision in the way a variable is measured or coded. Variables can be more or less descriptive, depending on 1) Their innate properties 2) How a researcher chooses to measure or code them
Experimental research
The gold standard/ideal. •Random assignment to experimental and control groups •Double-blind administration •Assessment of outcome and comparison between exp. and cont. groups •Rarely feasible in CJ and crim research•Can't have judges randomly assign people convicted of sex offenses to prison or to a new community-based treatment program•Can't have police officers randomly select which domestic-violence suspects to arrest and which not to
Operationalization
The identification of specific variables and indicators that reflect our definition of a term
Reliability
The other major important characteristic of a good variable Reliability: "A measurement procedure yields consistent scores as long as the phenomenon measured is not changing" (p. 107) That is, a repeatedly used measurement/operationalization strategy should produce the same results over time
Conceptualization
The process by which we define main terms (measurable)
Why Sampling Matters: Generalizability
The purpose of quantitative research methods is to make a claim or draw a conclusion about a population All adults in the United States All prison inmates in the United States All prosecuting attorneys in a particular state All juveniles involved in gangs But populations are too large to study directly See example on p. 116 The solution: sampling. Draw a subset of the population (a sample) and use that subset to make inferences about the population Sample of adults across the U.S. Sample of inmates in a prison All youths in class on a certain day Merely polling people is not good enough Talking to people on the street Sending out an email or posting an ad online asking people to take a survey There are rules (methods) to sampling-sampling methodology is what separates scientific research (high-quality information) from polling and other anecdotal approaches (low-quality information) The key to good sampling—and to generalizability—is the representativeness of the sample If the population is 51% female, 20% black, 30% Latino, with a median household income of $45,000 and an average age of 47 years, the sample should have very similar demographics A representative sample contains roughly the same proportions of certain characteristics as seen in the population as a whole
Role of IRB
The role of the IRB is to protect the rights and welfare of individual research subjects. This is accomplished by having the IRB assure that the following requirements are satisfied: 1. Risk to subjects are minimized 2. Risk to subjects are reasonable in relation to anticipated benefits 3. Selection of subjects is equitable, (i.e., fair) 4. Informed consent is sought form each subject or his/her legally authorized representative 5. informed consent is appropriately documented 6. when appropriate, the research plan makes provisions for monitoring data collection 7. privacy and confidentiality of research subjects are appropriately protected, and 8. when some or all of the subjects are likely to be vulnerable to coercion or undue influence, additional safeguards have been included IRB must approve these requirements BEFORE the start of the study and review the documents on an annual basis
Population
The set of individuals or other entities to which we wish to be able to generalize our findings
Nominal Variables
The simplest, least descriptive level of measurement No quantifying information (i.e., no "more than" or "less than") - purely descriptive Values have no mathematical interpretation Putting people, objects, or places into categories Race, sex, political party, country of origin Basically, categories with no rank ordering Can only be listed out; cannot be arranged hierarchically Nominal variables must be mutually exclusive and exhaustive
Conceptualization: Why Measurement Matters
The way we define a concept can have (serious) implications for the results of our study For example, "binge drinking": Consumption of several alcoholic beverages over a short period for purposes of intoxication How do we measure (define) this? The Substance Abuse and Mental Health Services Administration (SAMHSA) defines binge drinking as drinking 5 or more alcoholic drinks on the same occasion on at least 1 day in the past 30 days "39% of university students engage in binge drinking versus 34% of non-university students." Yikes But by that definition, anyone who attends parties or watches sports is a binge drinker But if we are more conservative in the measurement, we get a much different picture An episode that lasted more than one day Drinking to the point of losing consciousness The results may be (drastically) different So let's go back to our concept of violence How would you define the concept of violence? How would you conceptualize (measure) violence? In your definition/conceptualization of violence did you include Certain crime types? Specific behaviors or actions? Harm done to another person? Concept of violence cont. In my dissertation, I chose to define violence as being incarcerated for the following crimes: Murder, unspecified homicide, voluntary/non-negligent manslaughter, manslaughter-vehicular, manslaughter-non-vehicular, kidnapping, armed robbery, unarmed robbery, aggravated assault, simple assault, assault public officer, blackmail/extortion/intimidation, hit and run driving, child abuse, and violent offenses-other I did not include sex offenses in the "violent offense" category
Conceptualization: What is it?
Theoretical constructs and concepts are abstract. They are general notions Strain Violence But work needs to be done to turn abstract ideas into something concrete that can be collected and analyzed After you identify a concept, you must conceptualize it Conceptualization defines concepts into a way that can be measured
The social-research process
Theories are like chains and hypotheses are the links Multiple links, all of which must be true for the theory to be correct Individual research studies only test one or a small number of hypotheses, never the entire theory itself Hypothesis supported = support for the theory Hypothesis not supported = possible indication of problem with the theory Hypotheses involve two or more variables. Variables are characteristics, properties, or events that are measured in a study. The dependent variable is the outcome of interest. The independent variable is the hypothesized cause of the dependent variable. Does arrest deter future domestic violence? Independent variable: arrest Dependent variable: domestic violence recidivism Causal model: a method of depicting the proposed relationship between two or more variables Hypothesis: arrest will deter (that is, reduce) future dv offending See description on negative vs. positive relationships on p. 46 Causal models can also be more descriptive Can include the hypothesized causal mechanism Hypothesis: arrest will deter future dv abuse by making offenders think their likelihood of arrest is high Induction: using empirical findings to inform larger ideas/theories Deterrence: dv offenders who were arrested did not think it was likely they would be arrested again; or recidivism dropped among arrested offenders who were employed and married but rose among those who were unemployed and unmarried Induction sometimes causes us to reexamine theories/ideas. modifications must then be empirically tested to see if the newly renovated theory is a better match to reality. Inductive research: exploratory research that uses empirical findings to create a new theory If a police precinct noticed (from its arrest data) that arrests of at least one person from the scene of a domestic violence case resulted in fewer subsequent calls to that same scene, the police officers might develop (induce) a theory that mandatory arrests decrease domestic violence recidivism Useful, but potentially dangerous. All theories premised upon data from a single study must be empirically tested in new studies.
The role of theory in public policy
Theories can help us formulate ideas for public policy domestic-violence offenders Deterrence theory: arrest is the best solution; this will prevent future acts because the offender doesn't want to be punished again. Labeling theory: non-arrest solutions are best; arrest will stigmatize an offender and cause his abusive behavior to continue. Procedural justice theory: doesn't matter if you arrest or not; what matters is that the offender felt the police treated him fairly and respectfully. But those ideas must be tested before they are put into practice on a large scale Do dv offenders who are arrested have lower (or higher) rates of future abuse than those who do not experience an arrest? Do dv offenders who were arrested and who felt police treated them fairly have lower rates of recidivism compared to arrestees who felt unfairly treated? What about compared to dv offenders who were not arrested?
Falsification: black swans
Theories must be testable or falsifiable ("Show to be false") In search of the black swan: all the swans I have ever seen are white; therefore, I theorize that swans can only be white If you find a black swan, you have disproven my theory Supporting evidence (more white swans) isn't enough. The true test is to look for contradictory evidence. Neighborhoods with high social disorganization and low crime rates Police officers who arrest high-status suspects and let low-status ones go It is easy to fall into the trap of unfalsifiable claims get-tough proponents: putting more people in prison will reduce crime Critics: incarceration rates have increased, but crime rates are still high Proponents: not tough enough, more prison Critics: it's not working Proponents: tougher! Anyone advancing a claim has a responsibility to be specific about their predictions. If a prediction doesn't hold true, time to reexamine the claim.
Protecting Research Participants
There are a variety of principles that must be adhered to and steps that must be taken to protect human subjects
Errors in Causal Reasoning
There are two ways to draw erroneous conclusions by mixing up units of analysis 1. Ecological fallacy: Using group-level statistics to make statements about individuals within the group. Concluding that a correlation between unemployment and crime at the city level means unemployed people commit crime or that someone who loses his job will start committing crime Concluding that a group-level correlation between early childhood abuse victimization and adult criminal behavior means that people abused as children become criminals 2. Reductionist fallacy: Using individual-level statistics to make statements about the group. Concluding that because there is a relationship between low education and crime that all people with low education must be crime-prone or that all crime is committed by people with low education Concluding that neighborhoods with high concentrations of lower-educated residents will have crime problems
The social-research process
There should be a cyclical relationship between theory and empirical research That is, the collection and analysis of data (see Exhibit 2.4) Constant process of deductive an inductive reasoning Deduction: Starting from a general proposition (theory) and testing a specific subcomponent of that proposition (hypothesis) Deterrence: dv offenders who are arrested will perceive a greater likelihood of future arrest for re-offense, relative to those who were not arrested
Sampling Methods
Two basic types: probability and nonprobability Probability sampling methods: All people, places, or objects in the population have a known probability of being selected into the sample, and nobody's probability is zero Nonprobability sampling methods: No known probability of selection or non-selection; some elements of the population have zero chance of selection Probability sampling methods are key in quantitative research (again, qualitative is different). This is the one and only way to get a representative, generalizable sample
Qualitative research
Use of ethnography, participant observation, face-to-face interviews to collect in-depth data from a relatively small sample. Generally used in exploratory research only.•e.g., a researcher spends one year observing the behavior of youth in a high-crime, low-income neighborhood to find out how these youth adapt to the dangers and pressures of life in this type of area
Options for operationalizing concepts:
Using available data Constructing questions Making observations Collecting unobtrusive measures
Mixed methods
Using both qualitative and quantitative methods to study one research question
Quantitative research
Using numbers to describe social phenomena •Count events •Analyze information with statistical techniques
Probability Sampling Methods: Multistage Cluster Samples
Very common in CJ and crim research. Useful when No sampling frame is available (population is too large) Researcher wants information on both larger elements (e.g., neighborhoods) and smaller ones (e.g., the people living in those neighborhoods) Researcher wants to ensure diversity and variety in the sample Recall unit of analysis: Whatever is being studied (people, places, objects, events) Multistage cluster sampling: The process of selecting primary, aggregate sampling elements and then randomly selecting secondary elements from within those aggregates Cluster: Naturally occurring group (e.g., schools, neighborhoods, prisons) This sampling strategy has two (or more) units of analysis: the cluster and the elements within the cluster Step 1: Select the clusters or level-2 elements (aka, primary sampling units) No rule about the number, but for statistical purposes, probably want at least 10, depending on the size of the population and the sample Can select clusters entirely randomly or with some deliberativeness (or do a census, if possible) Step 2: Select the individual or level-1 elements from within each cluster Random selection of smaller elements (e.g., people) from within each cluster The level-1 elements are said to be nested within the level-2 clusters Students within classrooms Residents within neighborhoods Inmates within prison Cities within countries Generally, need at least 15 - 20 level-1 units per cluster
heuristic devices
We use our experiences to build up a set of heuristic devices. These are mental shortcuts, convenient ways of organizing the world •Our brains can only process so much incoming information at a time. We need shortcuts to conserve energy for the new or unfamiliar inputs.
Protecting Research Participants: Privacy and Confidentiality
When possible, research should be anonymous Anonymity often not feasible, though, so confidentiality is second best Never release identifying information Researchers must also take steps to protect Ps' privacy e.g., storing data in locked cabinets and secure computers
Protecting Research Participants: Avoid Deception
When possible, researchers should avoid lying to or deceiving participants Deception can cause psychological distress But deception is sometimes necessary: "[R]esearchers deceive their subjects when they believe that knowledge of the experimental premise may actually change the subjects' behavior" Minor deceptions are common Experiment: Researcher is interested in the social psychology behind people's failure to assist others who are clearly in need of help. Participant (P) is put into group of confederates (Cs) and the group walks across campus. Along the way, another C collapses on the sidewalk in front of the group. The Cs calmly step around or over the crumpled C. Question: What will the P do, surrounded by others who do nothing? Experiment: Researcher wants to know what effect groups have on the quality of individuals' decision making. One P is in a group of Cs. Two cards are presented: left one has a single line drawn on it, right one has three lines of varying lengths. Researcher asks everyone to say which line on the right is the same length as the one on the left. Cs provide clearly erroneous answer. Question: Will P insist on the right answer, or will he go along with the group and select the obviously wrong line? Clearly, can't disclose nature of these types of experiments to Ps beforehand. Study ruined. Small amounts of harmless deception probably OK, as long as the question the researcher is asking has demonstrable benefits Risk-vs.-benefit balancing analysis
Observational Research
When qualitative, part of ethnographic research. Researcher spends time with subjects, gets to know them, watches them interact with each other and with their environment. When quantitative, called systematic social observation. Way of quantifying observations so they can be used in statistical analyses.
Interobserver Reliability
[Note that intraobserver reliability is the same thing as test-retest] Interobserver reliability is the use of more than one observer to rate a single event If both people provide similar ratings, reliability is good If the observers' ratings differ, reliability is bad
theory
a "logically interrelated set of propositions about empirical reality" (p. 41) Theories are grounded in initial observation
Theoretical constructs
core elements of theories "These [...] describe what is important to look at to understand, explain, predict, and 'Do something about' crime" (p. 41) So, essentially items that we can measure (turn into variables) "theories usually link one or more theoretical constructs to others..." (p. 41) Ex. Differential association ( links definitions favorable or unfavorable to the violation of the law to crime) Social disorganization (concentrated economic and social disadvantage) Police culture (officers' views of citizens and management) Self-control (ability to control impulses and to think about long-term consequences)
common sense problem
occurs when we extrapolate our anecdotal knowledge; that is, when we confuse heuristics with facts •Our knowledge is nearly always incomplete •That guy is mean to me, but he volunteers at the children's hospital on weekends •That store has good prices, but there is a store closer to my house with equally good prices •Our anecdotal knowledge paints an extremely limited and often inaccurate, biased, even prejudiced way of viewing the world
pseudoscience
sometimes, people claim to be scientists (or claim that a certain belief is scientific) but the elements of science are missing. •It can be hard to tell the difference when a source of pseudoscience is persuasive. People can be talked into believing that information is trustworthy when, in fact, it is not phrenology: the detailed study of the shape and size of the cranium as a supposed indication of character and mental abilities.
Explanatory research
•Cause and effect •Are people with low self-control more likely to commit crime? •Does the easy availability of firearms influence the gun-homicide rate? •Comparing two competing theories or proposed causes• Fact: Large portions of people living in high-crime areas don't call the police to report crimes or information about suspects. Is this because... •Subcultural tolerance of deviance theory: They don't call the police because they don't care about crime and disorder in their neighborhoods, or they even support it. •Legal cynicism theory: They dislike crime and disorder as much as anyone else, but they don't think the police will do anything about it so they don't bother to call.1/10/2019
The Scientific Approach
•Criminological and criminal justice research is an area of social science research •"Science relies on logical systematic methods to answer questions, and it does so in a way that allows others to inspect and evaluate its methods... This transparency allows other researchers to see if the same results can be reproduced... Transparency also relies on peer review, the process by which other independent researchers evaluate the scientific merit of the study" (p. 8). •SOCIAL SCIENCE= investigate individuals, societies, and social processes; the knowledge produced by these investigations
Strengths and Limitations of Social Research
•Despite adherence to scientific methods, no single study ever "proves" something to be true •Example: One gang-prevention program in one school district had a small impact on youths' involvement in gang crime•What about other school districts, maybe ones serving students in high-crime communities with more gang presence?•What about the same program administered by different people (teachers, coaches, police officers, etc.)?•Is the program equally effective for kids who are at high vs. low risk of joining gangs?•Lots of questions remain, even when initial results are promising. •"After an exhaustive review of all existing studies, the committee concludes that more research is needed"•Social research requires a balance between not leaping recklessly to conclusions, while still recognizing the merit in preliminary findings•No proof that the gang-prevention program works, but initial evidence suggests that the program should be tried in other school districts
Illogical reasoning
•Draw conclusions from invalid assumptions •People who commit school shootings have mental illnesses; therefore, people with mental illness are dangerous •Most adult offenders engaged in delinquency as juveniles; therefore, juvenile troublemakers will become adult criminals •Poverty causes crime
Overgeneralization
•From one person to a group •That guy is mean, and he works at a bank, so all people who work at banks are mean •From one person's single action to a behavioral trait •That guy cut me off in traffic, so he's a bad driver and probably kicks his dog, too. •Can lead to stereotypes
Exploratory research
•Gather more information on newly identified areas/concerns •What led a mass shooter to his terrible crime?•Why do so many sexual-assault victims refuse to go to the police?
Inaccurate observation
•I thought the price tag said $5.75, but it was actually $8.75 •Perception involves interpretation •Example: Eyewitness misidentification
Selective observation
•Look at phenomena according to personal beliefs, regardless of fact •I focus on the fact that he takes my parking place every morning, and I ignore his volunteer work at the children's hospital •The "Yeah, but..." syndrome
HOW DO CCJ RESEARCHERS DECIDE WHAT TO STUDY?•
•MOST CCJ RESEARCHERS SPECIALIZE IN A PARTICULAR AREA OF STUDY•BIOSOCIAL CRIMINOLOGY•POLICE USE OF FORCE•SELF-CONTROL THEORY•SOCIAL DISORGANIZATION THEORY•CORRECTIONAL REHABILITATION•SEX OFFENDER POLICY•PROBLEM SOLVING COURTS •GENERALLY, RESEARCHERS STICK TO QUESTIONS WITHIN THEIR AREA OF EXPERTISE. THEY ASK QUESTIONS THAT HELP THEM GAIN A BETTER UNDERSTANDING.•BIOCRIM: "ARE THERE DIFFERENCES IN BRAIN CHEMISTRY BETWEEN OFFENDERS AND NON-OFFENDERS?"•USE OF FORCE: "DO CHARACTERISTICS OF THE OFFICER AND/OR SUSPECT AFFECT THE AMOUNT OF FORCE USED?"•SEX OFFENDER POLICY: "DO REGISTRATION AND COMMUNITY NOTIFICATION LAWS IMPROVE PUBLIC SAFETY?"•MILITARY SERVICE & CRIME: IS MILITARY SERVICE RELATED TO DEVIANT BEHAVIOR?
Resistance to change
•New information fails to change attitudes •Ego: Once we declare our stance on something, we feel a need to stick to our guns, even when evidence of our error is overwhelming •Tradition: Assuming that since previous generations did it, it must work and must be the right way to do things. •Deference to authority: Failing to examine the evidence yourself and instead relying on others to tell you what to think about an issue.
Social Research Philosophies
•Positivist and Postpositivist Reality & perceptions don't match •Interpretivist and Constructivist Reality is socially constructed •Feminist Research Focus on women's lives
Triangulation
•The combination of methods to answer a question •Clearer picture of the social reality being studied