Research Methods Exam #1

Statistical Conclusion Validity

-Refers to our ability to determine whether change in the suspected cause is statistically associated with a change in the suspected effects. •Basing conclusions on a small sample size is a common threat to statistical conclusion validity.

Authority

new knowledge does appear everyday, whether we believe it or not depends on the status of the person discovering the new piece of information.

Units of Analysis

•CJ research involves a great deal of variation concerning who or what is studied - technically called units of analysis.

Retrospective vs. Prospective

•Combining Retrospective analysis with Prospective analysis is commonly done in Applied Research, specifically evaluation studies. •Goal is to analyze/measure what occurred prior to a program or policy change and then analyze/measure the same thing again after the program/policy implementation in order to show change.

Construct Validity

•Concerned with how well an observed relationship between variables represents the underlying causal process. •Construct validity refers to generalizing from what we observe and measure to the real world things in which we are interested. •i.e. - is what you are observing and measuring really what is critical to the process you are studying - example police supervision - is it just a matter of keeping officers in view or are there other ways of supervising that have not been observed or measured that will affect the outcome of the study.

Causation

•Criminal justice researchers are trying to find out things that involve the question of cause and effect. •The chief goal of social science research is to explain why things are the way they are - some things are caused by other things. •Causation is the foundation of explanatory research (why is something occurring - ex. Number of teenagers in gangs increasing)

Time Dimension Summary

•Cross-sectional - Snapshot: -produces an image at one point in time. •Trend Study - Slide Show: -series of snapshots in sequence - by viewing it you can tell how some indicators vary over time (tends to be aggregate data) •Panel Study - Motion Picture: -Captures moving images of the same individuals and gives us information about individual rates of offending over time

Ecological Fallacy

•Danger of making assertions about individuals as the unit of analysis based on the examination of groups or other aggregations. •Example: Can't study robberies by police precinct and then use the results to draw conclusions about individual residents or people who were in the precinct at the time of the robbery.

Considerations

•Defining your problem - conceptualization •Identifying Units of Analysis •Identifying the Time Dimensions •Choosing a research method: -Surveys -Interviews -Observation -Data Analysis

External Validity

•External validity is concerned with whether research findings from one study can be reproduced in another study, often under different conditions. •In federally funded research projects or evaluation projects this is a great concern - programs that can be replicated with equal success are considered "best practices" and have exceptional external validity.

Criteria for Causality

•First Requirement: There is a causal relationship between two variables when the cause precedes the effect in time. •Second Requirement: There is a causal relationship when the two variables are empirically correlated with each other - they occur together. •Third Requirement: There is a causal relationship when the observed empirical correlation between the two variables can not be explained by any other influences or third variable.

Internal Validity

•Internal validity threats challenge causal statements that are based on some observed relationship. •An observed association between two variables has internal validity if the relationship is in fact causal and not due to the effects of one or more other variables - related to the 3rd requirement for establishing a causal relationship (eliminate all other possible explanations). •Internal validity errors occur when some other variable is responsible for the observed effect.

Longitudinal Studies

•Longitudinal studies permit observations over an extended period of time. •Three specific types: Trend, Cohort, and Panel Studies •Trend: looks at changes within some general population over time (UCR data) •Cohort: examines more specific populations (cohorts) as they change over time (Age Group or groups of people that start or end something at the same time) •Panel Studies: similar to trend and cohort except the observations are made on the same set of people on two or more occasions. •Example: National Crime Victimization Survey - once chosen for the sample the household stays in the sample set for three years and is questioned every six months. •Problems: -Can be very costly -Take a long time to complete -Difficult to implement and maintain -Panel Attrition

Cross-Sectional Studies

•Many projects study a topic by examining a cross section of it at one time. •Exploratory and descriptive studies are often cross-sectional. •Examples: US Census or a Community Survey •Inherent Problem with cross sectional studies: aim is to understand causal processes that occur over time, but their conclusions are based on observations made at only one time - don't really know what happened before or after survey is completed.

Combining Units of Analysis

•Most CJ research will involve descriptions or explanations of more than one unit of analysis: •Community Policing Evaluation -How do citizens feel about their community police officers? (individual) -Did arrests increase in police beats with CP Officers? (group) -Do police departments that implement Community Policing have larger budgets than those that don't? (organization)

Necessary & Sufficient Causes

•Need to understand that all causal relationships in criminal justice are probabilistic in nature. •Given this you need to distinguish two types of causes: necessary and sufficient. •A Necessary Cause is a condition that must be present for the effect to follow. •Example: It is necessary to be charged with a crime if you are to be convicted. •A Sufficient Cause is a condition that more or less guarantees the effect. •Example: Pleading guilty to a crime is sufficient cause for being convicted. You could also be convicted by going to trial, but pleading guilty is also a "sufficient" cause for being convicted.

Individualistic Fallacy

•Opposite of Ecological - making assertions about the whole based on findings about an individual. •Example: Media reports tourist attacked in New York - does not mean that all tourists are in danger in New York. These media messages may distort how people initially approach research problems in Criminal Justice

Exploration

*A lot of research is done solely to explore a specific problem, crime, or policy issue. • Can involve simple methods - one report - or complex and multiple methods. *Exploratory studies can also be used if a policy change or new strategy is being considered. •Example - Mandatory Sentencing leads to prison overcrowding - leads to a search for alternatives to incarceration.

Why We Sample

- 1) we select samples to represent a larger population of people or things - 2)we want to generalize from a sample to an unobserved population the sample is intended to represent.

Selective Observation

- A danger of overgeneralization is that it can lead to selective observation. - Meaning, if we already conclude that a pattern exists we will tend to only pay attention to the events or incidents that fit the pattern and ignore any that don't. - To avoid this, use a large enough sample, follow through with the entire project and do not jump to conclusions based on a portion of the sample results.

Overgeneralization

- Can not assume a few similar events are evidence of a general pattern. - Tendency to overgeneralize is greatest when there is pressure to reach a general understanding. - Overgeneralization can misdirect or impede inquiry - can misrepresent or simplify a problem. - To avoid overgeneralization you must: • 1) observe or study a large enough sample • 2) ensure the sample is representative of the whole • 3) Replicate the inquiry (repeat the study) to see if you achieve the same results each time.

Ideology and Politics

- Crime is an important social problem therefore the policies surrounding the topic are often controversial - people feel very strongly about some of the issues (death penalty) - Ideological or political views can undermine the objectivity of the research process. - We all have our own beliefs, the challenge is to ensure that these beliefs do not distort how research problems are defined and how research results are interpreted.

Illogical Reasoning

- Do not fall into the trap of "the exception that proves the rule" - This is not logical - an exception can draw attention to a rule but it can not prove a rule it contradicts. - Avoid this by using systems of logic consciously and explicitly.

Description

- Key purpose of criminal justice research is to describe the scope of a crime problem or policy response to the problem. - A researcher can observe and then describe what was observed. - Descriptive studies are concerned with counting or documenting observations, while exploratory studies focus more on developing a preliminary understanding about a new issue. - UCR - Prison Census - Case studies through direct observation - implementation of community policing, development of neighborhood watch groups, etc.

Inaccurate Observation

- Keystone of inquiry is observation - Normally people are unconscious, sloppy observers of the world (witnesses to crime) - In contrast, scientific observation is a conscious, planned activity - Need to employ predetermined measurements, photos, checklists, etc.

Examples of Sampling Frames

- Membership Lists are a good source but you need to know how inclusive they are of the entire population you are studying. - Telephone Directory - problems: people who don't have phones, unlisted numbers, cell phone only households, etc. - City tax logs - homeowners - Resident Lists for apartment buildings

Errors in Personal Human Inquiry

Besides the inherent dangers of relying on traditions and authority, there are other common errors people make in conducting personal inquiries. - Inaccurate Observations - Overgeneralization - Selective Observation - Illogical Reasoning - Ideology and Politics

Application

Fourth purpose of CJ research is its application to public policy. • Conducting "applied research" comes from a need for specific facts and findings with policy implications. • Two types of applied research: evaluation and problem analysis. Evaluation - applied research used to evaluate the effects of specific criminal justice programs - to determine the effectiveness of program strategies. • Basically - comparing the program goals to the program results. • Example - Weed & Seed - Firearms - number of incidents, number of arrests, case dispositions, repeat offenders, etc.

Explanation

Third purpose of CJ research is to explain things. • Difference between description and explanation: - Reporting the number of 14 year olds involved in gang activity vs. discussing why the number of 14 yr olds involved in gang activity has increased.

Tradition

everyone is born into a culture with strongly accepted knowledge about how the world works. This knowledge is cumulative and constantly building as time goes on - allows us to not have to "start over" with every generation. Some things have already been figured out by others before us.

NCVS Problems

• 1) Over-reporting - victims can misinterpret events and report incidents that weren't crimes (lost wallet as stolen wallet) or report crimes outside of the time frame • 2) Under-reporting - victims will be too embarrassed to report some crimes, forget some crimes, or won't say so they won't get in trouble. • 3) Inability to gather data on personal criminal activity (drug use, gambling, murder, etc.) and commercial crimes since only households are in the sample - so this limits the crimes that this survey can report on

NCVS Problems

• 4) Sampling errors - may produce group of respondents that do not represent the nation. • 5) Inadequate question format that invalidates the response - can't ask leading questions, all must be asked in the same way, etc. • 6) Does not reveal anything about personality, attitudes, or behavior

Sampling Frames

• A sampling frame is the list or quasi-list of elements (unit about which information is collected) from which a probability sample is selected. • Or a sampling frame is a list or quasi-list of our target population. • Properly drawn samples provide information appropriate for describing the population of elements that compose the sampling frame - nothing more.

Purposive or Judgmental Sampling

• At times it is appropriate to select a sample based on our knowledge of the population, its elements, and the nature of the research project. • This sample would be based on our judgment and the purpose of the study - hence the name Purposive or Judgmental. • Gang Members are a perfect example - we may be able to identify some but it is clearly not feasible to identify all and then conduct a probability sample of the group. • Conducting a study on those that are known may be enough for the purpose of the study. • Other examples: - Choosing specific sites or programs to study because they exhibit specific attributes necessary for the project ("best practice" studies). - Choosing voting precincts for exit polls - can't poll everyone so specific precincts are chosen because they best represent the area as a whole.

Reliance on Available Subjects

• Best example of this type of sampling is the surveys that are asked of passers-by in shopping malls or on street corners. • The researcher is relying on the people available crossing a specific site to obtain data on a specific issue - crime, shopping habits, feelings of safety in that area, etc. • Example - interview people walking in a park before and after new security measures are put in place to ascertain their level of safety and satisfaction with the changes. • Studies of homelessness are another common use for reliance on available subjects. • Researchers will go to shelters and areas where the homeless are known to gather to obtain data.

NCVS - How Does it Work?

• Bureau of Justice Statistics - arm of DOJ - annually surveys households selected through a random sampling process. • Involves a lot of people - 2005 = 77,000 households, 134,000 individuals • The sample is interviewed twice a year about their experience over the prior 6 months. • Once chosen randomly, the household stays in the sample for three years.

Multistage Cluster Sampling

• Cluster sampling is used when it is impossible or impractical to compile an exhaustive list of elements that compose the target population (ex. All people that live in a city). • In these cases you would need to conduct an initial sampling of groups of elements (clusters) followed by the selection of elements within each of the selected clusters (cluster sampling). • Common Citizen Survey Example - Typically there is no single list of a city's population, but there are lists of census blocks or street addresses. - First step would be to draw a sample of census blocks from within the city. - Second step would be to draw a sample of people that live within the chosen census blocks. - If you wanted to get more complex you could add another layer and sample census blocks, then households, and then individuals that reside within each of the chosen households.

Self-Report Survey Accuracy

• Critics say it is unreasonable to expect people to honestly talk about their criminal activity. - Some may exaggerate to be "cool", others may under-report for fear of getting in trouble. • However, research done on these surveys show that people do answer them honestly - Peer review, test twice to see if answers changed, designed questions to see if people were lying, polygraphs to ensure honesty • Even with this verification - self-report survey methods are questionable • What about the students that don't participate - are they the really bad 10% that would provide a different picture • Research shows the worst kids will not report or easily forget criminal activity • Also, worst kids are ones most likely to not cooperate with a survey form • Institutionalized kids are not included • So, hard core chronic delinquents may actually be excluded from the survey results

Idiographic and Nomothetic Explanations

• Every day we explain things - why did something happen. • Idiographic explanation means we fully understand the many causes of what happened in a particular instance - you are just trying to explain that one incident very carefully. (exam grade) • Nomothetic explanation explains a class of situations or events rather than a single one using only one or a few explanatory factors not every possibility. (teams win at home - use "usually") • Powerful CJ research tools - A case study of one gang - ideographic - What factors lead to juvenile delinquency - nomothetic • Both are useful and good science - sometimes need a broad, less detailed explanation (nomothetic), and other times you need specific details about a specific event/incident (ideographic).

NCVS Redesign

• Focused on obtaining better information on domestic violence and sexual assaults as well as helping respondents recall incidents more completely. • Summary of Changes: - Revised screening questions and added cues throughout the survey to help respondents recall minor incidents - More direct questions about rape and other sexual assaults - Greater effort to obtain information about victimization by someone the respondent knows (DV) - Increase in use of telephone interviews to replace inperson method

Simple Random Sampling

• Foundation of unbiased sampling. • Once a sampling frame is established, a random sample can be produced by assigning a single number to every element in the frame. • Computer programs can be used to choose random numbers associated with the elements that are then included in your sample.

Methodological Problems

• How each crime is counted • Many different offenses can occur in one incident • The UCR only counts the most serious of the offenses • So many crimes go uncounted because of the way the data is scored by the UCR

UCR Data

• If you are in need of historical national crime data 2019 or earlier the UCR would be the "Official Data" source to reference. • Uniform Crime Reports were issued by the FBI on an annual basis. • Data comes from over 17,000 police departments across the country. • PDs report the stats to the FBI on a monthly basis - get them from police reports or computer systems. • Records number of offenses reported to the police, does not consider solved/prosecuted/convicted. etc. • Also reports monthly how many cases are cleared. • Meaning: - At least one person is arrested and turned over for prosecution, or - The police know who did the crime but can't arrest because of exceptional circumstances like the person is dead or left the country.

Quantitative and Qualitative Data

• In other words - numerical and non-numerical data. • Most observations are qualitative to begin with. • Useful to convert qualitative observations into a numerical form - helps to aggregate and summarize the data and allows for statistical analysis. • Non-numerical or qualitative data conveys more "richness" of meaning. (example - heroin user interviews)

Inductive and Deductive Reasoning

• Inductive reasoning moves from the specific to the general - from a set of particular observations to the discovery of a pattern that represents some degree of order among varied events under examination. • Doesn't tell you why this is so just that it is. • Example (studying for exams) - figure out by looking at each exam specifically that you do better on exams in general when you study in a group.

Self-Report Survey Purpose

• Measuring attitudes towards crime • Computing the association between crime and important social variables (education, family relations, income) • Enabling crime experts to evaluate the distribution of criminal behavior across racial, class, and gender lines • Test theories of crime and behavior

Action Research Purposes

• Most commonly used Probability Sampling: - Stratified Sampling - Systematic Sampling - when sampling frames are available for larger community surveys • Most commonly used Non-Probability Sampling: - Purposive or Judgmental Sampling - Snowball Sampling - Reliance on Available Subjects

Snowball Sampling

• Most commonly used in field observation studies or specialized interviewing, snowball sampling begins by identifying a single subject or small number and then asks the subject(s) to identify others who would be willing to participate in the study. • CJ research often uses snowball techniques when trying to identify a population that is unknown, i.e. drug dealers, gang members, etc. • The method has disadvantages in that you are trusting an untrustworthy population and you may need to sift through many false leads to actually find useful participants. • But, similar to self-report surveys, sometimes it is the only method you have available to get at an unknown population you want to study.

NIBRS: Pluses & Minuses

• Much more comprehensive police report needed to complete NIBRS reporting requirements. • Meant to collect much more detailed information about individual incidents, offenders, offenses, and victims. • Eliminates the hierarchy rule or only counting the reigning charge. • Allows data to be analyzed by victim, offender, and incident not just by incident. • Also expanded the crime categories to include: - Blackmail - Embezzlement - Drug Offenses - Bribery - Hate Crimes • This is a strange expansion list since blackmail and embezzlement or not high on a crime priority list for the majority of police departments across the country • In theory - it is an excellent idea. • In practice - Initially it was a disaster for the early "pilot" police departments - No training involved - new reports just distributed, so police officers have no idea how to complete them, so they just continue to fill in the same information as before. - Almost too much data to make sense of (if it was all submitted) - Claims of technical assistance and funding - but most departments have not received anything like this

UCR Validity Concerns Summarized

• No federal crimes reported • Reports are voluntary and vary in accurateness and completeness • Not all police departments submit reports • FBI uses estimates in its total crime projections • Only the most serious crime is recorded in incidents involving multiple crimes • Each act is listed as a single offense for some crimes but not for others - ex. Robbing 6 people is one robbery, killing 3 is three homicides. • Incomplete acts are added to completed acts • Important differences exist between FBI crime definitions and state and local crime definitions • Victimless crimes go undetected • The majority of Child Abuse cases and Domestic Violence are unreported.

Reporting Practice Problems

• Not all crime gets reported to the police - actually less than 40% • Victims don't report for a number of reasons: - Don't trust the police - Fear of reprisal - Shame - DV situations • So, UCR may significantly under-represent the total number of crimes - especially sexual assaults, Agg Assault, etc

Probability Sampling

• Probability Sampling enables you to make relatively few observations and then generalize from those observations to a much wider population. • It helps researchers generalize from observed cases to unobserved ones. • It is a method of sampling in which each member of a population has a known chance or probability of being selected.

Quota Sampling

• Quota sampling is similar to probability sampling in that it is concerned with representativeness. • Quota sampling begins with a matrix or table describing the characteristics of the target population - it is a complicated process that attempts to place all elements into boxes that are assigned characteristics (i.e. male, under 25, working, living in urban area, etc.) • Once the table is complete, each cell/characteristic is weighted based on what portion of the entire population that cell represents.

NCVS Redesign Consequences

• Results show an increase in sexual assault victimization and domestic violence offenses. • Important Points: - 1) What we learn from survey results depends on what we ask and how we ask it. - 2) If you redesign a survey tool you can not compare trends and changes in crime overtime because the survey design itself has changed.

Self-Report Surveys

• Self-Report Surveys were developed because neither the UCR or NCVS can get at the country's most important social problem - drug abuse among teenagers • These are anonymous questionnaires administered through a mass distribution to high school students. • They are supposed to measure attitudes, values, personal characteristics, and behaviors.

Stratified Sampling

• Stratified sampling is a method used for obtaining a greater degree of representativeness. • Rather than select a sample from the total population, elements are selected from subsets found within that population. • You can stratify by any number of variables depending on how complex a sample is necessary. • Ultimate function of stratification is to organize the population into homogeneous subsets and to select the appropriate number of elements from each. • Examples: - Housing Survey - Crimes

Victim Surveys

• Surveys that ask crime victims about their experience with criminals and crime. • What is the most important and widely used victim survey? • NCVS - National Crime Victimization Survey

NIBRS Transition

• The reliability of the UCR had long been questioned and a new generation of reporting was developed over a 15-year time frame - NIBRS. • NIBRS took over as the ONLY Official Data crime data national source as if 1/1/21 when the FBI discontinued the UCR. • Problem is not everyone has transitioned to NIBRS for a number of reasons so there is a gap in data availability right now. • Gov't is attempting to tie the transition to federal grant funding, so PDs are forced to transition or loose federal funds.

NCVS

• The survey asks questions about: - Frequency of criminal victimization - Characteristics of criminal victimization - Consequences of criminal victimization • Considered very valid because of the complex sampling methods employed and the high completion rate (90%) • Survey is important because it provides insight into all the unreported or underreported crimes (ex. UCR 92,000 rapes in 2006 vs. NCVS 270,000 rapes in 2006)

Sampling Definition

• The value of your research depends on what you choose to observe and how data is collected. • Sampling is the process of selecting observations because: - 1) you can't always collect information from all persons or units you want to study - 2) it is often not necessary to collect data from all persons or units

Law Enforcement Problems

• The way police departments record and report criminal activity seriously affects the validity of the UCR report. • Problem 1 - Interpretation of the definition of Index Crimes - Example Aggravated Assault can be many different crimes in different towns, who is counting what?? • Problem 2 - Increasing police efficiency and technology - Crime rates will actually rise as new computer systems are implemented or new enforcement strategies are created - but this is not a rise in crime, but a by-product of improved policing and better record keeping. - The UCR can not account for these administrative or procedural changes only the strict number of crime • Problem 3 - Validity of numbers reported - If a police department does not have the technology (records management system) or personnel (crime analyst) to complete these reports the numbers are based on best estimates or guesses.

Differing Avenues for Inquiry

• There is no one way to do CJ research - many valid methods available. • Method Choices: - Idiographic and Nomothetic Explanation - Inductive and Deductive Reasoning - Quantitative and Qualitative Data

Disproportionate Stratified Sampling

• This method deliberately produces a sample that IS NOT representative of the population. • Why? If only a small number of people exhibit some attribute of interest, then a large sample must be drawn to produce adequate numbers of elements that exhibit the uncommon characteristic. • Disproportionate stratification is a way of obtaining sufficient numbers of these "rare" cases by selecting a number disproportionate to their representation in the population. • Example: - Crime victimization - specifically armed robbery is relatively rare on a national scale so in order to capture information about these "rare" victims, you would want to include more people who live in large urban areas where the serious crime is more likely to happen in your sample - so "disproportionately sample" people from large urban areas.

UCR Offenses

• UCR only reports on a certain group of crime types they call Part I Offenses or Index Crimes • Part I crimes include: - Homicide - Forcible Rape - Robbery - Aggravated. Assault - Burglary - Larceny - MV Theft - Arson

Systematic Sampling

• Very close to random sampling • If you have 10,000 elements and you need a sample of 1000 every 10th element should be selected for the sample. • The first element chosen to start the selection should be randomly chosen from a number between one and ten. • Technically - this is a systematic sample with a random start. • If the sampling frame is randomized to start with this is very much a simple random sample. • Problems arise if the list is in some particular order that would coincide with the sampling pattern then your sample will be biased. • Example: an apartment list in numerical order - the sampling interval could coincide with similar apartments just on different floors - all near a stairwell or an exit and depending on what you are researching this could bias your results. • Solution - make sure you know how the list was made and determine any bias before you choose the sampling interval.

UCR Accuracy

• What are the three areas of concern regarding UCR data accuracy? • Reporting Practices • Methodological Problems • Law Enforcement Practices

Official Crime Data

• What is the definition of Official Crime Data? - Crimes known to and recorded by police departments across the country. • What is the best known and most widely cited source of official crime statistics? - UCR - Uniform Crime Reports (up to 2019) - NIBRS - National Incident Based Reporting System (1/2021 on)

Sampling Bias

• When creating samples you must avoid conscious and unconscious sampling bias. • Bias means those selected for the sample are not "typical" or "representative" of the larger population they have been chosen from. • This bias is inevitable if a sample is chosen casually.

Multistage Cluster Sampling with Stratification

• You can also use the same stratification techniques discussed with random sampling when you are using multistage cluster sampling. • Law Enforcement Personnel Example: - Stratify the enforcement agencies at the first stage by type or size, etc - So all county agencies with more than 50 officers would be in the first cluster - then you could use a random or systematic sampling method to choose specific elements (officers) for the sample at the second stage once you collected personnel lists.

Representativeness

• Your sample MUST be representative of the larger population from which it was selected. • Meaning: The sample must have the same basic characteristics of the larger population. • The sample does not have to have ALL the same characteristics, just the ones that are pertinent to whatever you are studying.

Individuals

•Any variety of individuals may be units of analysis. •Important because social science suggests that scientific findings are most valuable when they apply to all kinds of people. •This is not what CJ research does however - it usually focuses on specific people because of characteristics or memberships that they have (gang members, victims, police officers) •Descriptive studies that have individuals as their units of analysis typically aim to describe the population that comprises those individuals.

Reductionism

•Overly strict limitation on the kinds of concepts and variables to be considered as causes in explaining the broad range of human behavior represented by crime and CJ policy. •Example: Economists will consider only economic variables, Sociologists will consider only sociological variables, etc. •Social scientists depending on their specialty will look at some explanations for crime problems and ignore others.

Retrospective Studies

•Retrospective research asks about the past - it is another way of approximating observations over time. •Common retrospective CJ research involves analyzing offenders' criminal histories. •Problems: -If using surveys - people forget or lie -If using data - records may be unavailable, incomplete, or inaccurate.

Social Artifacts

•Social artifact is the product of social beings and their behavior - critical to the CJ Action Research field. •Social artifacts can be newspaper articles, police reports, criminal histories, meeting minutes, etc. •How is a police report a social artifact? -Police reports are records of social interaction, police record and interpret victim and witness statements and determine who is telling the truth. The report becomes the social artifact that represents one among a population of all assaults or all B&E's, etc. •Social artifacts require information about individuals but the recorded social interaction between people is the unit of analysis.

Groups

•Social groups can also be units of analysis (juvenile gangs, police beats, cities, households). •Example: if we want to know why teenagers join gangs, the teenager (individual) is the unit of analysis - if we want to compare gangs who deal drugs to gangs that steal cars, the gang (group) is the unit of analysis. •Example: can study police beats to determine if those containing schools report more drug incidents.

Organizations

•Social or political organizations can also be units of analysis. (correctional facilities, police departments, probation departments, victim service agencies). •Example: Studying federal prisons vs state prisons to determine types of crime each house. •Example: Comparing victim service agencies to determine which agency has better service provision to sexual assault victims.

Concerns About Units of Analysis

•Studying the cause of increase in juvenile crime: -Sociologists might think it is the changing family structure - so the unit of analysis would be families (groups) -Economists might use states as the unit and compare state economies to juvenile crime rates (organizations)

Prospective Studies

•This research requires an examination or approximation of the future. •Retrospective vs. Prospective -What percentage of abuse victims later abuse their children? (Prospective) -What percentage of abuse victims have parents who were abused? (Retrospective) •The way you frame your question dictates which time dimension you should use.

Time Dimension

•Time order is a requirement for causal inferences so the time dimension of a research design requires careful planning.

Validity Threats

•Validity threats - reasons we might be incorrect in stating some cause produced some effect. Researchers assess the truth of statements about cause by considering the threats to validity