Research Methods Exam 2

Victim-report

measure of crime obtained by asking people about their experiences as victims alternative measures of crime but are subject to error crimes such as prostitution, drug abuse, public order crimes and delinquency are excluded from this and are underestimated by police records of people arrested for those offenses

Mutually exclusive

only one attribute can apply to one situation for each observation only one box should apply when exhaustive and mutually exclusive together the attribute list has to provide one but one one observation every observation can be recorded in one and only one attribute category

Threats to external validity

represents a slightly different form of generalizability the question is whether results from experiments in one setting (time and place) will be obtained in other settings, or whether treatment found to be effective for one population will have similar effects of a different group greater threat when the experiment is conducted under carefully controlled conditions Field results have greater external validity

Quasi-Experiment Design

research design that includes most, but not all, elements of an experiment when random assignment is not possible, researchers can use different types of quasi-experimental designs used when randomization is not legally or ethically possible 1. non-equivalent group designs, which looks like classical experimental but without random/control group instead as a comparison group. which looses a good deal of controlling for spurious relationships Once we use non-equivalant group designs we can control for characteristics but can not have the ability to address non-spurious relationship Non equivalent groups and time series designs are two general types of quasi experiments

Conceptualization

sets up foundation for our examination of operationalization and measurement the process by which we specify precisely what we mean when we use particular terms the mental process of making fuzzy and imprecise notions (concepts) more specific and precise

Pre/Post Testing

simple experimental design: subjects are measured on a dependent variable (Pre-tested), exposed to a stimulus that represents an independent variable, and then remeasured on the defendant variable (post-tested) difference noted between the first and second measurements on the dependent variable are then attributed to the influence of the independent variable

Dimension

some specifiable aspect of a concept the end product of the conceptualization process is the specification of a set of indicators of what we have in mind, indicating the presence or absence of the concept we are studying. specifying dimensions and identifying the various indicators for each of those dimensions are parts of conceptualization

Control Group

subjects in an experiment who do not receive the experimental treatment the traditional way to offset the effects of the experiment itself observe a control group to which the experimental stimulus has not be administered using a control group allows the researcher to control of rate effects of the experiment itself designed to combat validity problems when we use a control group we get to see how much actually changes. it guards against measurement itself this is essential because without it people would assume the change was because of the independent variable

Time Series

takes dozens/hundreds measures of DV ex. Gun control gun control=more guns less crime or less guns less crime. wash d.c passed a law on gun control and a nearby city did not. used baltimore because the demographics was a good comparison sight. does gun control affect gun related deaths? expert other weapon deaths to be the same so looked at existing data to find that restrictions did not have a meaningful impact it is a load of observations time series and case studies are examples of variable-oriented research, in which a large number of variables are examined for one or few cases designs are common examples of longitudinal studies in criminal justice research interrupted time series is special type design that can be used in cause and effect studies. A series of observations compared before and after some intervention is introduced. Problem: they operationalize complex causal constructs in simple ways.

Reliability (Measurement Quality)

we need to have consistency it is not about accuracy. if you get same answer over and over again even if it is wrong it is still reliable because you are getting the same thing each time is a matter of whether a particular measurement technique applied repeatedly to the same thing will yield the same results roughly the same thing as consistency or stability does not ensure accuracy any more than precision does. concern every time a single observer is the source of data because we have no way to guard against that observers subjectivity also a problem when more than one observer makes measurement, because different interviewers get different answers because of there demeanors and attitudes also when we ask people for information about themselves

four types of validity threats

Face Validity Criterion-Related Validity Construct Validity Content Validity

why crime data is collected

monitoring agency accountability research

Experimental group

the group that a stimulus has been administered subjects in an experiment who receive the experimental treatment ex. if participation in the experiment leads the subjects to report less alcohol use, that should occur in both the experimental and control groups. if the overall level of drinking exhibited by the control group decreases between the pre test and the post test as much as for the experimental group then the apparent reduction in alcohol use must be a function of the experiment or of some external factor, not a function of watching the video specifically. ex. if drinking decreases only in the experimental group, then we can be more confident in saying that the reduction is a consequence of exposure to the video because that is the only difference between the two groups.

Reification

treating concepts as if they are real even if they are not

Reaching a conceptual definition

A conceptual definition is a working definition specifically assigned to a term. This definition does two important things. first being it seres as a specific working condition we present so that readers will understand exactly what we mean by a concept (word for conceptions). It also focuses our observation strategy. it does not produce observations but it channels our efforts to develop actual measures. The next step is operational definition.

UCR validity and reliability problems

Because UCR does not count all crimes we can question is validity. If crime was defined more specifically such as crimes known to police and recorded in police files then it would be more valid Reliability- not all law enforcement agencies submit complete reports and the quality varies. Inconsistencies in reporting and pressures to reclassify incidents raise questions about reliability Validity problem because talking about crime but only actually looking at crime known to police construct validity problem because it does not capture the total range of meaning

Threats to internal validity

Hawthorne effect- the idea that measuring something changes it a type of contamination in experimental designs that occurs when members of the treatment group change in terms of the dependent variable because their participation in the study makes them feel special this refers to the possibility that conclusions drawn from the experimental results may not accurately reflect what went on in the experiment itself History-historical events may occur during the course of the experiment that confound the experimental results. can effect the dependent variable and make it seem like it was the IV that caused the change when it was not.

Experimental design and causal inferences

Experiments potentially control for many threats to the validity of causal inference, but researches must remain remain aware of these threats the iv/dv, pre/post testing, experimental and control groups created through random assignment are the building blocks of a research design to demonstrate a cause and effect relationship. Experimental design-ensures that the cause precedes the effect in time by taking posttest measurements of the dependent variable after introducing the experimental stimulus the second criteria for causation is deterred by comparing the pretest to the posttest for the experimental group after the stimulus is administered. A change from pretest to post test measures demonstrates correlation not due to third variable- the classical experiment makes it possible to satisfy this in two ways. the posttest measures for the experimental group(stimulus present) are compared with those for the control group (stimulus not present). if it is due to an outside factor then the two posttest scores will be similar Random assignment produces experimental and control groups that are equivalent and will not differ on some other variable that could account for the empirical correlation between cause and effect so it addresses covariation,, looks for time order by pre/post testing, and spurious relationship

Construct Validity

Logical relationships among variables Tests of construct validity can offer weight of evidence that our measure either does or doesn't evaluate the quality we want it to measure, without providing definitive proof. if we are successful in measuring some variable, then those measurements should relate to other measures in some logical fashion asks us to look at how multiple variables relate together IN THE SAME STUDY looking for a logical connection in this study take a quick check at variable in study and if blatant contridictions take place then it raises a red flag

NCVS compared to UCR

NCVS=victim surveys UCR=crimes known to police both have strengths and weaknesses UCR only provides a summary measure of aggregate units and the NCVS yields more disaggregated data on individual victims, offenders, and incidents so NCVS is better suited to studies of individual factors in the types of incidents covered by the survey , it cannot provide statically reliable estimates of crime for most cities, countries or states since it is just a survey and it limited to household residents age 12 and over. so it could be missing offenses.

Classical Experiment

Research design with four components: IV/DV, pre-post tests, experimental and control groups, and random assignment the term experiment refers to a specific way of structuring research, usually called the classical experiment the central features of the classical experiment are independent and dependent variables, pretesting and post testing, and experimental and control groups created through random assignment. in CJ we almost never use this because of legal and ethical issues tests the effect of an experimental stimulus on some dependent variable through the pretesting and post testing of experimental and control groups

Concept

Words for mental images in science we can not rely on concepts the words or symbols in language that we use to represent those mental images the words and phrases we use represent abstractions, because they are independent of the labels we assign them may vary based off of experiences and backgrounds

Self-report

ask people about crimes they may have committed skeptical of this technique because we wonder how truthful people actually would be many people also do not wish to disclose illegal behavior even if they are told it will be confidential others could deliberately lie to interviewers and exaggerate the number of offenses they have committed best method available for trying to measure certain crimes that are poorly represented by other techniques A third class of offenses that might be better counted by self report surveys are crimes that are rarely reported to the police, such as shoplifting and drunk driving all crimes require an offender and not all crimes have clearly identifiable victims who can be interviewed

"Blind" and "Double Blind"

blind=subjects do not know if they are in experiment or control experiment double blind=subjects and researchers do not know who got which. this involves needing two researchers, one who measures IV/DV and does not know and one who knows

Validity (measurement quality)

how close we are to measuring the thing we want to measure whether statements about cause or measures are correct (valid) or false (invalid) means that an empirical measure adequately reflects the meaning of the concept under consideration if you are really measuring what you think you are measuring valid measures are truly indicators of underlying concepts.

Generalizability

construct validity and external validity How well a study's findings can be applied to other people and circumstances. Because experiments often take place under controlled conditions, results may not be generalizable to real world situations or findings from one sitting may not apply to other settings.

Crimes known to Police

crimes reported to police and recorded by police the most widely use measures of crime are based on police records and are commonly referred to as this important implications of understanding what police do and do not measure crimes not known to police obviously cannot be measured by consulting police records

Exhaustive

every observation can be recorded somewhere on the list-exceptions will need a box labeled other attribute list has to be exhaustive exhaustive and mutually exclusive are the bare bone minimum things required for a operational definition every observation must have a place to be recorded

Nominal

exhaustive and mutually exclusive no other characteristics, simplest level because it requires the only two minimum things when we organize at this level everyone has one thing in common and one thing marks them different than every other group examples: gender, race, city of residence, college major, Social Security number, martial status. merely offers names or labels for characteristics

Pulse Check

not quantitative anyone around drug market is invited, including researchers, professors, etc. gives idea of supply and demand throughout the country and get early knowledge about drug trends helps law enforcement stay in front because drug addicts come up with a lot of new things and can get creative

Criteria for measurement quality

the key standards for measurement quality are reliability and validity

Four Levels of Measurement

Nominal Ordinal Interval Ratio anything done at higher levels can be done at lower levels, but can't do lower level at higher level

Generalization

A conclusion, drawn from specific information, that is used to make a broad statement about a topic or person.

Ordinal

exhaustive, mutually exclusive, AND has an implied order of attributes if you disorganize this order then it does not work ex. months, ABCs, Fresh/Soph/Jr/Senior

Interval

exhaustive, mutually exclusive, has implied order AND uniform gaps between and within categories each category can only include same number of possibilities these are rare because it is rare to have uniform gaps

Experimental stimulus

represents the independant variable ex. showing a sad video to see alcohol consumption goes down, could be other factors contributing though

Independent variable

takes the form of an experiment stimulus that is either present or absent, that is a dichotomous variable have two attribute the given variable might serve as an independent variable in one experiment and as a defendant variable in another the cause

Dependent variable

the effect example: watching the video causes a change in alcohol use or that reduced alcohol use is an effect of watching the video both independent and dependent variable be operationally defined for the purposes of experimentation usually both are developed before the experiment begins

Ratio

exhaustive, mutually exclusive, has implied order, uniform gaps, AND there is a true zero that enables meaningful math and reflects the absence of the variable ex. Kelvin (temp) is actually ration because it has an absence of heat usually just a line because if it was checkboxes you would have to have one for every single possibility one by one

Progression of measurement steps

Conceptualization Conceptual definiton operational definition measurements in real world

Uniform Crime Reports (UCR)

police measures of crime form the basis for the FBI's UCR this data series has been collected since 1930 and widely used by cj researchers Part 1-criminal homocide, forcible rape, robbery, aggravated assault, burglary, larceny theft, motor vehicle theft, and arson. only what FBI counts as crime which is the more serious crimes Part 2- other offenses they are based off of crimes reported to police so they have measurement problems. crimes in this part only count if there is an arrest made Only 98% of departments answer the UCR reports, the 2% who don't are unknown because it changes about 30 of these 1st problem- the UCR does not even try to count all crimes reported to the police. Part 1 offenses are counted if they are reported and recorded by the police. Part 2 are counted only if a person has been arrested and charged with the crime. therefore crimes such as shoplifting, drug sale or use, fraud, stolen property, and all other non traffic offenses are not included unless someone is arrested. meaning a large number of crimes reported to police are not measured in the UCR 2nd Problem- part of the reason that part II offenses are counted only if arrest is made is that individual states have different definitions of crime. Inconsistency also exists in the quality of data submitted, meaning states, cities and counties vary in the quality and completeness of crime data sent to the FBI and reported in the annual UCR publication 3rd problem- produced by hierarchy rule, meaning if multiple crimes are committed in a single incident only the most serious crime is reported in the UCR UCR system produces a summary based measure. they can not be used in descriptive or explanatory studies that focus of individual crimes, offenders or victims. so therefore the data is restricted to the analysis of such units such as cities, counties, sates, or regions

Content Validity

the degree to which a measure covers the total range of meanings included within the concept asks have we captured the total range of meaning this is subjective because there is no real definition we can only compare our conceptual definition with there conception if you look at one study you compare and contrast compared to your own is the definition couple or only looking at one side, so is it broad enough to cover what we need ex. does playing video games make you more violent? some will say yes others will say opposite

Precision

measurements can be made with varying degrees of precision which refers to the fineness of the distinctions made between the attributes the compose a variable example: saying that a women is 43 years old is more precise than saying that a women is in her forties in general precise measurements are superior to imprecise ones but not always necessary or desirable the operationalization of concepts must be guided partly by an understanding of the degree of precision required, if your needs are not clear be more precise rather than less refers to the exactness of the measure used in an observation or description of an attribute

NCVS revisions

most recently revised in 2006 by the BJS changes included how samples are drawn, how questions are phrased, and how interviews are conducted. 1993-focused on obtaining better measures of domestic violence and sexual assault, together with methods to help respondents recall a broader range of incidents presents more specific questions and cues, in effect encouraging respondents to think about specific types of incidents. anything we learn from surveys depends on what we ask and how we ask it. measures of crime are affected in part by the procedures we use to make those measures.

Face Validity

particular empirical measures may or may not fit with our common agreements and our individual mental images about a particular concept Does it make sense basically ex. peoples views on women being educated makes no sense to travel to country where women can not have an education just ask if you r definition is total nonsense or you can get where they are coming from

Summary (offense) based measure

UCR system produces this includes summary, or total, crime counts for reporting agencies

Operationalization Process

Clarifying abstract mental images is an essential first step in measurement. So after coming up with a conceptual definition we come up with a operational definition which is the further specifications of a concept, this definition spells out exactly what we will be measuring. It is a description of the "operations" undertaken in measuring a concept. Operationalization is the process of developing operational definitions. We begin by conceptualizing what we wish to study and then when we start to consider operationalizing we can revise our conceptual definition. Operationalization involves describing how actual measurements will be made and then the next step is actually making the measurements. So in operationalization we specify concrete empirical procedures that will result in measurements of variables. Operationalization begins in study design and continues throughout research project, including the analysis of data.

ADAM (Arrestee Drug abuse Monitoring)

Drug Surveillance System-suveillance systems have been developed to obtain alternative measures of drug use last collected data in 2003 and the program was terminated because lack of funding then in 2007 ADAM 2, a scaled down version was created and relocated to the Office of National Drug Control Policy it estimates drug use for a very specific population- which is persons arrested in 10 cities Over two 14 day periods each years, participating cities select samples of persons arrested for a variety of offenses. Anonymous interviews and urine specimens are obtained from those who agree to take part in the voluntary study. the main purpose=provide an ongoing assessment of the prevalence of drug use among persons arrested for criminal offenses Interviews with arrestees include self report items so responses can be compared with urine test results. Problem-any measure of crime is selective, so not all crimes or all people are included. it only operates in 10 cities so can't assume that participating cities are representative of other cities. simplicity assumes that at least some drug users will be involved in other offenses started out as DUFF system, ranges 23-35 american cities=juvenile/adult, male/female/, personal/property crimes. this is voluntary while they are in the holding tank-80% do participate. take arrestee to interview room and ask questions. not a national group of samples so does not give general pattern about half of juvenile arrestees, and 2 thirds of adults give positive drug results after urine test

NCVS

NCVS=National Crime Victimization Survey draws a new sample every six months the U.S census bureau has conducted national surveys of crime and victimization, which is known as NCVS. this was used to combat the large amount of unreported crime since UCR is only what is reported to the police it uses a uniform procedure to select and interview respondents, which enhances reliability of crime measures. It can not measure all crimes though because of the procedures used on select victims, like when businesses or commercial establishments are the victims. Bank robbers, liquor store holdups, shoplifting, embezzlement, political corruption, and securities fraud are examples of crime that cannot be counted by interviewing household members. along with victimless crimes such as drug sale and prostitution. Validity problem because it excludes many types of crimes. Reliability- since it is a survey it is subject to error. respondents may not remember some incidents, especially if it involves minor crimes. also the problem of telescoping which is when respondents just do not accurately recall when an incident occurred. also people who have been victimized multiple times. it also underestimates incidents in which the victim and offender know each other.

Conception

Our unique mental image for things Conception is subjective clarifying abstract mental images is an essential first step in measurement the technical name for those mental images a subjective thought about things that we encounter in a daily basis, but they can not be communicated directly because there is no way we can clearly reveal whats written in our mental images therefore we use terms

Operationalization

Our goal is to create a definition of the concept that allows for uniform measurement, has consistency, is viable (it has to be there with a way to measure), has to serve purpose of research study concrete steps, or operations, used to measure specific concepts describing how to obtain actual empirical measures begins with operationalization the process of developing operation definition although we begin by conceptualizing what we wish to study, once we start to consider operationalization we may revise out conceptual definition involves describing how actual measurements will be made

Benefits of randomizations/random assignment

Random assignment is a central feature of the classical experiment, randomization produces experimental and control groups that are statistically equivalent. meaning random assignment reduces sources of systematic bias in assigning subjects to groups. if subjects are assigned to experimental and control groups through a random process such a flipping a coin, the assignment process is said to be unbiased and the resultant groups are equivalent. "All other things being equal", random assignment makes this possible brings outside third variable to lowest possible controls for every variable possibly related to the change if you do not randomly assign there will always be a threat to the 3rd criteria guards against every other threat every other variable out there besides IV is statistically equivalent, we don't have to know anything about you but we can guard against any threat to IV to the greatest extent if there is no pattern at all to who is placed in experimental and control group then no statistical difference in each group other than the independent variable nothing can effect DV because of gender, race, etc. only way we can be confident that we do not have a spurious relationship the best way to achieve comparability in the experimental and control groups and it guards against most of the treats to internal validity

Issues with self-report measures

ask people about what crimes they may have committed How truthful are people when asked? Many people do not want to disclose the illegal information that they have done even if they are promised confidentiality they are the best method available for trying to measure certain crimes that are poorly represented by other techniques such as prostitution and drug sale and use. Researchers and policy makers are best advised to be critical uses or measures obtained from self report surveys although they can and should be used to measure offending researchers, public officials, and others should be aware of there strengths and weaknesses

Incident v. Summary or offense based measures

Summary based measure- means that data includes summary, or total, crime counts for reporting agencies: in most cases cities or countries represents groups as units of analysis. can not represent individual crimes, offenders or victims as units. Incident-based police records- the first one of these was Supplementary Homicide Reports (SHR), which was the spin off of UCR. Local law enforcement would submit detailed information about individual homocide incidents under this program which includes information about victims and if known offenders (age, gender, and race), the relationship between victim and offender: the weapon used: the location of the incident: the circumstances surrounding the killing Incidents are the basic unit and can include one or more victims and offenders can use this to conduct descriptive and explanatory studies of individual events Crime measures based on incidents as units of analysis have several advantages over summary. but still flawed. NIBRS (National Incident Based Reporting System) would be ideal over UCR and we started to make the shift in 1980s but it is a slow process it would include much more detailed information about individual incidents, together with the offenses, offenders, and victims within each incident. also has a broader array of offenses. it would drop the hierarchy rule.

Incident-based measure

The U.S department of Justice sponsors two series of crime measures that are based on incidents as units of analysis, incident based measure is one of them Police reports based on individual crime incidents incidents are the basic unit and can include one or more victims and offenders have several advantages over summary measures

Operational Definition

this prepares us to measure the best definition is the one that lets us answer our research question has to be identical every single time when we measure with this you take the idea of what you are and are not able to measure and create a way to measure, most of the time it is a survey must specify exactly what we will observe, how we will do it, and what interpretations we will place on various possible observations a definition that spells out precisely how the concept will be measured an operational definition is a description of a the "operations" undertaken in measuring a concept moves us closer to measurement

Leading up to a conceptual definition

We start out with a conception which is a subjective thought about things that we encounter in daily life, and this conception can not be directly communicated. People have different types of conceptions for everything we observe or hear about they also vary based on our backgrounds and experiences. In order to clearly communicate these conceptions we use concepts which are the words or symbols in language that we use to represent these mental images. So we use concepts to communicate about our conceptions. These concepts are abstract because they are independent of the label we assign them. These concepts are a "family of conceptions". Since concepts are just words to go with our mental images we use conceptualization to specify precisely what we mean when we use particular terms. When go through the process of conceptualization we use a set of indicators of what we have in mind, which indicates the presence or absence of the concept we are studying. We use dimensions which is a specifiable aspect of a concept. for example the "victim harm dimension" of crime seriousness could include indicators including physical injury, economic loss, or psychological factors. Specifying dimensions and identifying the various indicators for each of those dimensions are both parts of conceptualization, which paves the way for a more sophisticated understanding of what we are studying. A conceptual definition is a working definition of a concept or term.

Conceptual definition

Working definition of concepts or terms has a list of indicators might have dimensions (groupings) We have these because there is no such thing as a real definition and we have to be uniform in what we are counting There is no real definition so conceptual definition is the one we pick from a list of possible ones the point is not to create a right definition, but count what and what is not being measured it takes a really long time to make a conceptual definition because it goes off of what you are trying to measure is a working definition specifically assigned to a term serves as a specific working definition we present so that readers will understand exactly what we mean by a concept and it focuses our observation strategy

DAWN (The Drug Abuse Warning Network)

assumes drug use can produce acute health problems that cause users to seek treatment in hospital emergency rooms it collects emergency medical treatment reports for drug episodes from samples of emergency rooms and medical examiners nationwide drug episodes=visits to a hospital emergency room that are produced by or directly related to the illegal use or non-medical use of drugs. which includes direct effects of drug ingestion (overdoses or physical/psychic reactions) as well as injuries or deaths in which drug intoxication was a contributing factor designed to be a data system for health surveillance long-term series that monitors the most serious medical consequences of drug use and best suited to measure trends. it includes demographic and other information about the individual . unusually though one individual can account for multiple drug episodes most common drugs seen= alcohol, coke, optied category

Hierarchy rule

in a multiple offense incident, only the most serious crime is counted The Hierarchy Rule, which requires counting only the highest offense and ignoring all others, applies only to the crime reporting process and does not affect the number of charges for which the defendant may be prosecuted in the courts. we have better counts for higher offenses the lower offenses are less reliable

Criterion Related Validity

involves comparing a measure with some external criterion A measure can be validated when it shows that it predicts scores on another measure that is generally accepted as valid approach to combat this: show that our measure of a concept is different from measure of similar but distinct concepts compare operational definition in this study to other widely accepted definitions good researchers will explain how the widely accepted definition and their definition stack up with each other compare unusual definition in the particular study to widely accept one in other studies

NHSDUaH (National Survey on Drug Use and Health)

is based on a national sample of households that has been being conducted since 1971 with the sampling and questions being revised several times. it is sponsored by the Substance Abuse and Mental Health Services Administration in the U.S Department of Health and Human Services. designed to to monitor nationwide patterns it is used to obtain self reports of drug use since it has been being conducted for over three decades it is able to provide information on trends and changes in drug use among respondents the 2010 sample survey was designed to obtain statistically reliable samples from the eight largest states in addition to the overall national sample problems: do people tell the truth? there are procedures taken like interviewers conduct the rest of the interviews in a private area away from other household members and do not directly ask questions about drug use instead the questions are administered through computer-assited self interviewing. Drug use is still underreported but it varies by age and type of drug. Problem 2- military housing or work release community corrections centers along with homeless peoples are not included in household samples. they do try to make special efforts to include these people

MTF (Monitering the future)

it is different because it targets a specific population and it asks sampled respondents a broader variety of questions Since 1975- the National Institute on Drug Abuse ponders the annual survey of high school seniors which is MTF. it is intended to monitor the behaviors, attitudes, and values of young people. it is used to measure such behaviors such as drug, alcohol, and tobacco use among youths it actually includes several samples of high school students and other groups, totaling about 46,000 respondents Each spring- 400 secondary schools are sampled within particular geographic areas. Students complete questionnaires contacting questions that include self reported use of alcohol, tobacco, and illegal drugs. The 8th and 10th graders are anonymous but the seniors are confidential so that researchers can contact them years later it provides a cross section for measuring annual drug use and other illegal acts problem- students absent that day are not included and are not eligible for follow up survey. trend and panel study feature

Tradeoff between internal and external validity

threats to internal validity are reduced by conducting experiments under carefully controlled conditions but these conditions do not reflect real world situations, which restricts our ability to generalize results. then external validity is better if field research but there is a threat to internal validity because such studies are harder to monitor. "diabolical dilemma" Explanatory studies that test cause and effect theories should place greater emphasis on internal validity while applied studies should place a greater emphasis on external validity. not perfect hard fast rule because internal validity must be established before external validity becomes an issue meaning applied researchers must have confidence in there cause and effect relationship before they worry about if similar reactions would be found in other settings

Comparison Groups

we match individuals or groups looks for things associated with something and then list of characteristics that go with that outcome and match entire group/individuals with these characteristics Look at past research/literature to figure out these main characteristics that threat spuriousness the two groups(comparison and experimental) have matching characteristics. and all characteristics are related to the DV-at least the obvious factors that put us at risk for spurious relationship still can't be confident though because not completely random numbers have to be close in each group if matching groups and individuals would be one to one, not enough would be statistical conclusion error to validity