PLCY Exam 2

conceptions

mental images we use as summary devices to bring things together

concepts

social constructs, agreed upon meaning

conceptualization

process of clearly defining an abstract ideas for research purposes

dimensions 

manifestations of the concept

variables

properties of an event, person, thing that are measurable

operationalization

identifies a plan for the concrete and systematic measurement of a variable

ecological fallacy

a conclusion on the micro-level based on macro-level analysis

reductionism 

a conclusion on the macro-level based on micro-level analysis

reliability

measure reproduce the same results consistently

validity

refers to how accurate the measure is

Internal validity

is there a cause and effect of the independent variable on the dependent variable?

Face validity

Does this measure look good on the face of it? Does it make sense?

Construct validity

How well do multiple indicators of a variable correlate with each other?

External validity

How representative is the group being studied? How real is the study?

nominal

categories with no distinction

ordinal 

categories with rankings 

interval

distance between points are equivalent, and there is no zero

ratio

same as interval but true zero exsists

binary

questions with the most basic responses possible

non-ordered categories  

questions that require sorting into nominal categories that cannot be ranked

frequency (ordinal)

questions that require an assessment of timing or duration in ranked categories

frequency (interval/ratio)

questions that require an assessment of timing or duration in terms of units of time

degree

questions that require an assessment of the extent to which something is true or of the magnitude of something

agreement

questions that require subjects to assess how true or untrue some statement about their lives or the world (Likert scales)

single item

questions that work well for simple concepts

composite

measures that include multiple items like index

index

sums responses to survey items that captures key element of particular concept

scale 

averages responses to a series of related items at capture a single concept 

mutually exclusive

categories do not overlap with each other

exhaustive

respondents must have at least one accurate response to them

screening

eliminates people that aren’t being studied or are ineligible

order effects 

earlier questions may influence how participant respond to latter

priming effects

order of your questions can also “prime” people to answer a certain way

use existing questions…

already have high validity and reliability

face-to-face interview

respondent understand questions, collect paradata
physical appearance may bias, social desirability bias

telephone interview 

cost effective, conducted faster
interviewer effects from the person’s voice, unlikely to answer phone, respondent fatigue, data on location

mail surveys

more likely to disclose bad behaviors, cost-effective
no explanation for confusion, low response rate

online surveys

cost-effective, easy to program, reach greater number of participants
certain people respond more online, digital divide, ethical concerns

mixed modes

different types of information from respondents

cautious of mode effects

mode effects

study results may vary based on how the survey is administered

CAPI software

software for in-person interviews, helps with skip pattern

nonresponse 

some people may choose not to participate in a survey at all or not respond to specific questions

measurement error

approach used to measure a variables affects responses

sampling error

any difference between the characteristics of the survey sample and the characteristics of the population from which the sample is drawn

coverage error

the sampling frame does not capture all members of the target population

response rates

proportion of the people who completed a survey compared to total contacted

frequency

number of observations with particular value of the variable

frequency distributions

different values of a variable occur using actual counts

mean

continuous variable average but is sensitive to outliers

median

continuous variable measure best when there are outliers

average for binary

proportion of one value over the total

average for ordinal

can compute the mean, only useful in relation to the scale rather than on its own

average for nominal

frequency is best because values are meaningless

effect size

value measuring the strength of the relationship between two variables

chi-square

relationship between two categorical

correlations

relationship between two interval/ratio variables

t tests

binary nominal variable and a continuous variable

ANOVA (analysis of variance)

continuous dependent variable and categorical independent variable