CRJ 387 - exam 1

Experiental Reality

things we know from DIRECT experience - anecdotal

ex: touching hot stove

Agreement Reality

things we think are real because we’ve been told they are and everyone agrees

ex: Santa

Practitioner Lens

rely on “Professional Intuition”

research methods allow them to test their intuition against data to see if their “reality” matches results

UCR (Unified Crime Report)

A database that shows crime based on crime records

Personal Human Inquiry

curious species about the future and the world around us

the goal: to predict the future circumstances based on current patterns (fantasy football)

Causal reasoning

Cause and Effect

ex: Main Street Bridge will be late cause of traffic

Probalistic reasoning

likelihood, not certainty

ex: backroads usually quicker but school buses could slow it down

Inaccurate Observation

Error for reasoning

daily are casual; science make observation deliberate

Overgeneralization

error for reasoning

assuming few similar events are evidence of similar pattern

Selective observation

error for reasoning

focused on future events that fit pattern we already identified

Illogical Reasoning

error for reasoning

A gamble not support by facts

Hypothesis

starts all experiments

is the question you are trying to answer

Exploration Research (1 of the 4 purposes)

new or understudied phenomenon (widenet)

Description Research (1of the 4 purposes)

documenting scope of problems (UCR)

state-of-the-agency reports

Explanation Research (1 of the 4 purposes)

Why does this happen?

reporting relationships w/ variables

Application Research (1 of the 4 purposes)

uses policy analysis

did the evaluation research work.

predicting if a new policy will work and applying to people

Logic

rationality

do these things make sense

Attributes

characteristics or qualities that make up a variable

ex: sleep duration - the QUALITY of how many hours of sleep gotten

Variable

made up of attributes

Logical grouping (Dem or Rep)

ex: the NUMERICAL VALUE of representing sleep duration (5, 6, or 7?)

a measurable quantity

Quantitative Data Analysis

statistical, hard numbers, broad nomothetic research

heavy for CRJ research

Qualitative Data Analysis

Non-numerical, journalism, small sample size

“richness” idiographic research

Qual is less scientific

Nuremburg Code (1947)

concentration camps post WWII - seeing if voluntary consent is mandatory in medical experiments

Tuskegee Syphilis Study (1932-1972)

U.S Pub Health untreated syphilis in 60 black males and withheld the information that they had syphilis. Exploitation of disadvantaged groups

Belmont Report (1979)

response to Syphilis study, provides core ethics

no harm to participants, voluntary participation & informed consent, Anonymity vs. confidentiality, Deception and Historical Misbehavior

Standford Experiment (1971)

Lord of the Flies type situation

24 male college students were assigned as “guards” or “prisoners” in a mock prison at Stanford University. only lasted 6 days

Milgram Obedience Study

WWII based.

How soldier obedience happened with authority

teacher was instructed to administer increasingly severe electric shocks to the learner whenever an incorrect answer was given. The shocks were fake, but participants believed they were real. The shock generator ranged from 15 volts to 450 volts, with labels escalating from “mild” to “danger: severe shock”

this cause psychological harm to participants since information was withheld

No Harm

No physical, emotional, and psychological harm (stress, anxiety)

physical harm rare in Social Science

Social/Legal Harm

major risk in CJ

a “snitch” affects social status

Voluntary and Informed Consent

understand risk, benefit, and rights

not mandatory reporters unless dealing with Juveniles

prisoner dilemma: can inmate truly say not?

DO NOT offer incentives

Anonymity vs. Confidentiality

Anonymity: researcher does not know who the subjects are (GOLD STANDARD)

Confidentiality: know information of subject but promise not to tell anyone. However, can be subpoenaed.

Deception and Misbehavior

allowed if it is minimal and if the participant knew it would affect the results.

have to identify you are a researcher

Humphrey Tearoom Trade

Watchman who observed homosexual sex in bathroom then follows them when they go back to their lives w/ their wives. Noted everything then showed up to the house asking health questions

IRB (review boards)

committees review research proposals (rights, welfare, and privacy, etc..)

Levels of IRB

Full: More than minimal risk, involves physical or emotional discomfort

Expedited: NOT greater than minimal risk: research with exhumed docs

Exempt: less than minimal risk ex: anonymous survey most popular *

expedited most common for CJ research

The enforcer (OHRP)

can shut down ALL research funding if ethical rules are broken

Epistemological Challenge

how do we know? how do we measure?

use statistics we cannot measure

Translating theory to number

we cannot attach a meter to a person. must build systematic bridges & empirical data points

Three - Step Roadmap

conceptualization, operationalization, and observation/measurement

Conceptualization

“Defining the THING”

taking an abstract term and addressing/agreeing on what is means for specific study

EX: “bullying” has to be clear on the definition of bullying. creates a shared target.

Direct Observations (Maxfield and Babbie Classification)

things we directly witness (window breaking)

Indirect observations (Maxfield and Babbie Classification)

hidden records needing proxy (logs)

Constructs (Maxfield and Babbie Classification)

theoretical deviation built from multiple compound observations (antisocial personality)

Operationalization

Writing the recipe

spelling out exact steps, procedures, and questions you will use to measure - 5 senses

ex: survey question “how do you feel” (safe, unsafe, very unsafe, somewhat safe, etc)

moving from definition to questions

2 Golden Rules (Exhaustive and Mutually exclusive categories)

Exhaustive: every possible case or respondent must fit into AT LEAST 1 visible category

ex: employed vs. unemployed (traps students, disabled, veterans, etc.)

Mutually Exclusive Categories: no respondent or case can fit in 2 categories

ex: overlapping age brackets (18-24 + 24-30) causes inaccurate data

NOIR Hierarchy - levels of measurement

Nominal - (names/labels) categories with zero math ranking. (political affiliation)

Ordinal - (rank/order) can be put in logical order, but distance is not math equal (strongly agree to strongly disagree) cannot know mathematically how much higher one is)

Interval - (Equals, Variables) equal distance between numbers but 0 does not mean “absolute” (temperature) RARE in basic CJ surveys

Ratio - (true 0 point) equal intervals and an absolute zero point where zero mean none of the variable exists. (# of prior arrests) THE GOLD STANDARD

Reliability vs. Validity

Consistency = reliability

Truth = validity

Bullseye Analogy

Reliability = the same every time

Validity = Accuracy

science in collaborative

perfect experiment has both reliability and validity

Classical Experiment

3 components

independent variable and dependent variable, pre-testing + post-testing, experimental and control group

Independent variable

the cause (policy/program)

have at least 1 in your study

ex: study time - when testing its effect on test scores

Dependent Variable

the effect (outcome)

all for explanatory research

ex: exam score - depending on study method

Pre-testing and post-testing

measuring the dependent variable before and after intervention

Experimental + control groups

groups that change vs. group that doesn’t

selected at random to ensure statistic equivalences at the start

example: one group get to study the other does not

Ethical Dilemmas

cannot random assign people to high-risk conditions (prison vs. freedom)

Retrospective Research

much of it look at crime after it happened. Makes clean “before and after” difficult

5 threats to internal validity

History - external event that influences outcomes (police brutality or current events)

Maturation - changes through passage of time and growth (aging out of crime or learning right from wrong)

Testing - pre-test effects responses to post-test (given the same test and scores better because they remember the answer)

Instrumentation - “ruler” could be affected by changes (changing from paper to digital system during an experiment)

Statistical Regression - High + Low extreme score will naturally move closer to average (extremely deviant naturally move to average)

Quasi-Experimental designs

used in CJ since we cannot randomize

Does not use all 3 parts of classical experiment

2 Major Types (Non-Equivalent-Groups Designs & Time-Series Design)

Non-Equivalent-Groups Designs - using a “comparison group” that is similar to experiment groups but not randomly assigned (inmates at 1 prison with education and one without)

Time-Series Design - examining a series of observations over time before and after interventions (a new DUI law)

both of these scientifically acceptable

What are surveys?

collecting data by asking questions through questionnaire can be multiple-choice, open-ended, or a combo of both

MOST FREQUENT TYPE of survey in CJ

Why surveys are used in CJ

Limit of official records: National Incident-Based Reporting System (NIBRS) data that tells us what has been reported or caught

Measure the Unobservable: allows to study; public attitudes to the police + courts, fear of crime and victim experience, self-reported offending + illegal drug use among juveniles + inmates.

National Crime Victimization Survey (NCVS)

is anonymous and tracks crime not reported

2/3 of crime not seen

Constructing: Operationalization + variables

turning abstract concepts into concrete survey items

ex: scale of how they treat the community fairly (concept: legitimacy)

rule of thumb: every questions must connect back to the core concept

Close-ended questions

mutually exclusive and exhaustive

multiple choice questions (1 answer)

benefit: quantitative coding + statistical analysis

Open-ended Questions

qualitative depth + answer in own words

conflict: different to code and quantify

Double - Barreled Question

two questions in one

fix: split the questions up

Biased or Leading Words

flaw: nudging the respondent (Don’t you agree)

fix: use neutral phrasing

Social Desirability Bias

respondents underreport illegal behaviors or unpopular attitudes to be socially acceptable.

Modes of Data Collection

1: Self-administered (web/mail) - Primary way pro: low costs, great for sensitive topics, convenient. cons: lower rates, risk of non-response bias

2: telephone service - pros: faster, moderate to high response rates. cons: mobile-only, caller ID screening

3: Face to Face interviews - pros: highest response, complex questions + observations. cons: expensive, high risk of interviewer bias

Problems of Non-response bias

non-response bias- are the people who responded only responding because they are extremely passionate about something where others did not care

mitigation strategies: follow ups, incentives, and demographic comparisons of your sample to the population

Logic of sampling

cannot study everyone, sampling selects a subset of population

Probability sampling

every unit has a known, non-0 chance of being selected GOLD STANDARD for reducing bias

(random, simple, stratified)

ex: picking numbers out of a hat

Non-Probability Sampling

used when sampling frame is not available

ex: snowball sampling for active drug dealers or hidden population

used in CJ research because we are finding the hard to find

Sampling Bias

systematic exclusion of certain members of the population. selection process favors some characteristics over others

Generalizability

ability to apply findings from sample to larger population

Case Study - Literary Digest

predicted president based on surveys of millions - 1936

reality: their prediction was wrong

why? the great depression, wealthy individuals favored rep. parts, poor people did not have the information they were using to give out this survey

Simple Random Sampling

pulling numbers out of a hat

Systematic Sampling

select every “nth” element from a list

ex: “every tenth”

Stratified Sampling

population divided in strata’s (groups) based on characteristic (race, gender) then random samples taken from each group

Cluster Sampling

used when full list of population not available, researchers sample groups first (police precincts) and sample individuals in those groups

Purposive Sampling

selects a sample based on own knowledge of the population and research aims (“gang leaders”)

Snowball sampling

method where existing people recruit future people from friends; essential for studying active use members.

GMM - things in beer

shows hypothesizing and predicting what will happen

GMM - lie-detector tests

if it violates the ethical guidelines (no harm, social/legal harm, voluntary and informed consent, anonymity vs confidentiality, and deception and misbehavior).

GMM - identifying brands

use of conceptualization, operationalization, and measurement

GMM - stuff in Hot Car

looks at independent variable, dependent variable, threats of internal validity, and control groups