CJ 292 Exam 2

Concepts

words/symbols used to communicate mental images

Conceptualization

• Specify precisely what we mean when we use particular terms

  • Dimensions

  • Indicators

• More sophisticated understanding

• Conceptual definitions

  • Working definition

  • Uses words and symbols

  • So we know what to observe

  • Everyone understands

• Operational definition

• Spells out how the concept will be measured

• Absolutely specific; no confusion

Operationalization

The process of developing these definitions

first method of operation

• Using available data

  • Statistics already collected 

  • Definition and accuracy

second method of operation

• Asking questions

  • Open and closed-ended questions

third method of operation

• Observing behavior

  • Measure characteristics of individuals, places, or events

fourth method of operation

• Using unobtrusive means

  • Physical trace evidence

  • Archives

Nominal

• Categorical or qualitative

• Names or labels for characteristics

Ordinal

• Attributes may be logically rank ordered

• Greater than/less than

• can put them in order

Interval

• Fixed measurement unit, but no absolute or fixed zero point

• Distance that separates unit has meaning

Ratio

Fixed measuring units with absolute zero points

Measurement validity

Does it measure what we want it to measure

Face validity

On face value, does it make sense

Content validity

Full range of values or criteria

Criterion validity

• Another item to measure it against

• College kids drinking and BAC

Construct validity

Using theoretical constructs

Reliability

• Consistency 

• Exp: speedgun (reliable method of measuring speed)

Test-Retest

Redoing or retaking a test to see if there is consistency. Consistency would show reliability.

Interitem Reliability

Asking similar, but different, questions. Should get similar answers each time.

Alternate-Forms Reliability

Two groups get asked the same question, but in different ways. Answers should be similar. The questions must be given in random assignment

Interobserver Reliability

Two or more observers look at the same phenomenon and you see if they come to the same conclusion independently.

Population

the entire set of elements (individuals or other entities)

Sample

•  a subset of elements from the population 

  • Should ideally represent my population so we can be more confident with results 

  • Exp: campus safety (talk to a few students to get a feel on how MSU students as a whole feel about campus safety) 

  • Exp: every country (195) (population), select sample (madagascar, japan, chile, spain, australia)

Sampling error

• The distance in between the target population and the sample

  • How not equal they are to each other

  • The higher the sampling error, the less generalizable your study is

Census

study the entire population instead of drawing a sample

• Avoids problems of generalizability and representativeness

• Expensive and time consuming

• U.S. Decennial Census

Probability sampling

• You know in advance the probability of someone being selected for a sample 

• Random selection: key point differing from nonprobability 

• Sample is representative if population

Random selection

key point differing from nonprobability

Simple Random Sampling

• Identifies cases strictly on the basis of chance

  • Flipping a coin

  • Rolling a die

  • Random number table

  • Random digit dialing

• Equal Probability of Selection Method (EPSEM) 

• Replacement sampling

Systematic Random Sampling

• If dont have list but numbers 

  • Sorted sequentially by a random number

• Select the first item randomly, then select every nth item

• Elements that have to be arranged sequentially

• Three steps

  • Total in population/total needed in sample 

  • Exp: 1000 cases in pop./50 needed in sample = every 20th case selected

• Select a # from 1-20 at random. This is the first case. 

• After the first case selected, every nth case

  • Every 20th case would be selected

• Lists cannot be sorted in any meaningful way

Stratified Random Sampling

• Strata - Layers, levels, groups

• Purpose: to ensure that various groups will be included in the sample 

1. Distinguish all elements in the population according to their value on some relevant characteristic

   1. That characteristic forms the sampling strata 

   2. Each element must belong to one and only one stratum 

2. Sample elements randomly from within each strata

3. Proportionate Stratified Sampling

   • Population: n=10,000

   • Sample: n=500

4. Disproportionate Stratified Sampling

   • Population: n=10,000

   • Sample: n=500

     • Oversample those groups in small (equal number from each stratum)

Cluster Sampling

• Use when entire list is not available 

  • List of all kids in elementary school in the U.S. is not available

  • List of all kids in all East Lansing elementary schools is available

• Cluster

  • Naturally occurring aggregate of elements of the population

  • Can only appear in one cluster 

• Two stage process

  • Random sample of clusters (exp: elementary schools) 

  • Sample elements from within each cluster

Purposive sampling

• Each sample element is selected for a purpose

• Researcher selects who participates

  • Entire population of a limited group or subset of a population

  • Or

  • “Key informant survey”

    • Knowledgeable about subject

    • Willing to talk

    • Represents a range of view points

Units of analysis

• Who or what (entity) you want to learn about

• Individuals

  • Goal

    • Learn about population made of individuals

  • Objects of study

    • Attitudes, behaviors, beliefs

• Groups

  • Goal

    • Learn about the group, not individuals in it

  • Objects of study

    • Social groups, non-human groups (agencies, hot spots)

• Aggregation

  • Taking separate responses then grouping them together

Ecological fallacy

If you collect data from groups, you cannot make conclusions about individuals

Reductionist fallacy

If you collect data from individuals, you cannot make conclusions from groups

first criteria for causation

• Empirical association

  • Have to be related to one another

second criteria for causation

• Time order

  • X has to come before y

third criteria for causation

• Nonspuriousness

  • What really causes variables to increase if there is third variable

fourth criteria for causation

• Causal mechanism

  • Why does this make sense?

fifth criteria for causation

• Context

  • Larger picture

Criteria of a true experiment

• Two comparison groups

  • Treatment

  • Control

• Random assignment to each group

• Assessment of change

first threat to internal validity

• Selection bias

second threat to internal validity

• Endogenous change

third threat to internal validity

• External events

fourth threat to internal validity

Contamination

fifth threat to internal validity

• Treatment misidentification

Cross-sectional vs. longitudinal

• Cross-sectional

  • One data collection point

• Longitudinal

  • More than once

  • Follow people over time

Attractive features of survey research

• Versatile

• Efficiency

• Generalizability

Omnibus survey

Large-scale survey with a lot of topics and different ideas

Questionnaire

Survey instrument containing the questions in a self-administered survey

Interview Schedule

Survey instrument containing the questions asked by the interviewer in an in-person or phone survey