Research Design

How do we know things

Experiential knowledge (self-experience) and we are told by experts and other people we trust

Two criteria for agreement knowledge

Logical (must make sense) + emperical (observaitions)= agreement knowledge

Personal human inquiry

Observe the world (we process things to create predictability)

What does this inquiry do?

It helps us make decisions and reduce anxiety

What does scientific research do

Uses the scientific process/methodologies to look at the world though others’ experiences

Probabalism

Every relationship is probablistic (when x happens y will usually follow, but not always)

Scientific errors

Inaccurate observations

Overgeneralization

Selective observation

Illogical reasoning

Ideology/politics

Inaccurate obeservation

Human observation is casual, not critical; science specify what we are looking for and how to record it

Overgeneralization

See a few times and assume something is true; science selects a representative sample and replicates the study

Selective observation

Look for stuff that supports what we believe; specify how many observations

Illogical reasoning

Jump to conclusions based on invalid assumptions and lack of evidence; make sure research question is logical

Ideology/polotics

Experiences shape how we see the world; science should try not to allow beliefs to influence research or findings (follow scientific methods and peer review)

Class Summary

• Experiences and being told is how we know things

• Ideology/prior experience change how we view things

• Every relationship is probabalistic

Understanding research

Critical because CJ system is based on research and makes us more informed

Purpose

Exploration (specific problems that can provide info for the future, description (scope), explanation (why something happens), and application (study implications)

Variant

Is there a relationship between two things

Steps of research

1.) Interest

2.) Read

3.) Question

4.) Conceptualize (must be precise)

5.) Method (how to collect data)

6.) Operationalization (how to measure concepts)

7.) Specify (define population)

8.) Collect

9.) Analize

Summary 2

Exploration and description are similar (base line) and explanation is why

Social science

Strategically observing what we see

Corrilation

Does not equal causation (can’t be fact)

Spurious corrilation

Things that are corrilated that make no sense

Logic/theory

They go hand in hand (research built around theory)

Theory

Use research to test (use existing to justify research question) and observation to generate theory (alway exceptions)

Hypothesis

Expectation about relationship between variables (basically research question, but yes/no)

Variables

Concepts in research question (usually more than 2); most are about people and their experiences

Empirical

To say that two things are related we need to have observations that

confirm these two things are relate

Research to test a theory

Find an existing theory that is interesting to you (use new method and stuff)

Hypothesis testing

Variables are made up of attributes (need to be clear otherwise can’t get everything we want to)

Attributes

Characteristics or qualities of something

Expected causality

X variable (independent) causes a change in Y variable (dependent)

Idiographic

Look at few case in detail (open ended questions)

Nomathetic

Examine phenomenon as a whole (less detail)

Deductive

Testing a theory (general to specific)

Inductive

Creating a theory (observing the world to develop theory)

Qualitative

Words

Quantitative

Numbers

Summary 3:

Attributes make up variables and qualitative is words (quantitative is numbers)

Establish causality

Use random assignment to remove other factors

Criteria for causality

Correlation, temporal order (x before y), and. no other explanations

Valididty and causation

Sometimes relationships are not causal because we can’t remove and account for a variable

Validity threats

Reasons relationship isn’t actually causal

Unit of analysis

Who/what you are studying

Individuals

People

Groups

Organizations, households, etc.

Ecological fallacies

Not every person/group is the same (overgeneralizations)

Research time

Cross-sectional (one time) or longitudinal (over time)

Cross-sectional

Harder to establish order, but is cheap/quick

Longitudinal

Multiple times; better with causation, but is expensive, takes time, and people drop out

Summary 4

Logic/empirical, variables, assume causality, etc.

Sampling

Process of selecting observations (ex:people)

Sample

Subject/ group we want to study (our unit of analysis)

Why we sample

Can collect data that could be used to generalize about the population

Probability sample (random sample)

Gives everyone a known/equal chance to be included (don’t need if have access to whole population)

Sample bias

If we don’t use probability sampling it might not be representitive of the population

Sampling element

Who/what makes up our sample

Population

Full grouping of sample elements

Population parameter

Given variable in the full population

Sample statistics

Summary description of a given variable

Summary 5

Key to probability sampling is random selection

Sample distribution

The distribution of all of the sample (ex: pick two people and get average combinations)

More samples

What will get us closer to the mean (larger is better)

Probability theory

Gives rules about sampling distributions (large will be more normal)

Sampling error

Will not be representative of the population (also known as standard error)

Sample size

As the sample size increases the standard error will decrease

Probability sampling rules

1.) Define population

2.) Sampling frame (list of people)

Accuracy

Depends on simple random sampling (very common)

Summary 6

As the sample size increases the standard error decreases

Systematic smpling

Time consuming, start with sample frame and then choose a random place to start (no if inharant ranking)

Sample frame

Numbered list

Stratified sample

Split sample into groups that are related to your study (can do random pick from strata, but can’t be independent or dependent)

Disproportionate sampling

Can get rare elements (unequal, but known chance of being selected); can fix by giving less contribution to those with greater chance of being selected (1/likelyhood of selection)

Multi-stage cluster sampling

When we don’t have a managble list (start with group and select from the group until you get mannageable list)

Non-probability sampling

Purposive, quota, convenience, and snowball

Summary 7

Most probability sample need a sampling frame (not multi-stage)

Purposive sampling

Purposefully choose people in your study (good for baselines and rare elements)

Quota sampling

Pick people that can meet pre-established criteria (will not be representative of population)

Convenience

People who are readily available (easy, but not representative)

Snowball sampling

Variation of purposive sampling (one person leads to more people; good for rare, but doesn’t go far)

Conceptualization

Define what our variable means (results in indicators and dimentions)

Dimentions

Broad category of concept (main catagories)

Indicators

Specific catagories of dimentions

Operationalization

Spells out exactly how you are going to measure your variable (usually survey)

Exhaustive

All possibilities are included for the questions

Mutually exclusive

Respondents should only fall into one category (unless select all that apply)

Nominal measure

No inherant ranking (yes/no)

Ordinal

Measure that has inherant ranking (ordered in specific way)

Interval measures

Ordered attribues with equal distance (no meaningful zero)

Ratio measures

Similar to interval, but has a true/meaningful zero (absence of something)

Summary 8

Operationalization and there are different levels of variables

Is our measure good

Does it capture what you intend (validity) and is it consistent (reliable)

Relaiability

Will be low when measuring a complex or subjective subject

Testing reliablility

Test-retest- stays the same in different samples

Internal consistancy- items measuring the same concept are highly corrilated

improving reliability

Conceptualize clearly, have a good measure, several questions about the same concept, and look at what others have done

Validity types

Face- does it appear right

Content- do you have all the dimentions/indicators

Construct- Are your measure related in ways that would be expected

Survey

Questionare for all research types (best are individual)

Guidelines

Need to get a true and accurate response from respondents

Open ended vs close ended

Short answer vs multiple choice

Summary 9

Open vs close ended and three types of validity

Best practices

Be specific

No double barled questions (no two in one; keep it simple)

No leading/biased questions

No prior knowledge questions (no speculation)

Avoid social desireability (remind them they are anonymous)

Other guidelines

General format- uncluttered

Contingency questions- not applied to everyone

Matrix questions- scale of 1-10 (is good for space, but may choose for the sake of it)

Question ordering