Midterm Flashcards
Measurement Validity
When the measure measures what we think it measures.
Generalizability
The extent to which info gleaned can be used to inform about population as whole (External validity)
Causal validity
Internal Validity - when hypothesis that A causes B is correct.
3 Stages of forming a good research question
- identifying 2. refining 3. feasible?
Theory
logical interraled set of proportions about empirical reality
Inductive Reasoning
observed data –> generalization which explains relationships between objects observed
(QUALITATIVE)
Deductive Reasoning
theory –> hypothesis –> define variables and operations –> implement measurements and observations to see if they confirm/fail hypothesis
(QUANTITATIVE)
Dependent Variable
variable that DEPENDS on another variable
Positive Association/Correlation
going in the same direction
Negative Association/Correlation
going in different directions
Theoretical Statement
social networks positively influence psychological wellbeing.
Moderating Variable
Variable that influences relationship between variables
Mediating Variable
Variable that explains the relationship between variables
Problem Formulation
- Problem area (something you are interested in). 2. Idea (what are concepts and possible variations)
- Theory (how are concepts related to each other)
Operationalization
How we measure variables included in a study
Direct Measurements
Visual, physical symptoms, interviews, self-administered questionnaire, written records
Indirect Measurements
Unobtrusive/Indirect observations, content analysis
Nominal Measurement
categories: types of vanilla, milk, etc. (when you can’t order other things) (gender, ethnicity, religion)
Ordinal Measurement
No measureable distance between values, Taste Test: rating of 1-5 (social class, racism, sexism)
Interval/Ratio
measureable distance between values, based on absolute zero, degrees, etc. (meaningful intervals)
Discrete variable
cannot take on all values within the limits (ex: 1, 2, 3)
Continuous variable
variable can take on any variable (ex: 1, 1.25, 1.45, 2)
Measurement Error
- Systematic (social desirability, acquiescence bias, leading questions)
- Random (why respondents feel that way that day, regression to the mean, multiple rating behr’s, inadequate training)
Reliability
The measurement is consistent
Validity
The results are accurate
Test/Re-Test Reliability
same result after multiple trials
Inter-rater reliability vs. Intra-rater reliability
correlation between 2 raters (inter) single observer of 2 pts in time (intra)
Sampling
Random (probability) and Non-random (non-probability)
Population
The entire set of individuals or other entities to which study findings are generalized
Sample
Subset of population used to study whole
Sampling Frame
Structure for choosing sample (ex: list of Loyola students)
Sampling Units
units listed at each stage of multi-stage sampling design
Units of analysis
from whom or what are you collecting data (ex: individuals, households, orgs)
Ecological Fallacy
making a generalization based on a small sample
Representative Sample
looks like population from which it was selected
Unrepresentative Sample
doesn’t look like the population from which it was selected
Census
a study of the entire population (assumes perfect response)
Sample population generalizability
if sample is generalizable to the target population
Cross-population generalizability
If sample cannot be generalized to target population
Sampling Error
Any diff between characteristics of a sample and characteristics of population from which it was drawn
Recruitment Strategies for Diverse Populations
- involve key members of community/orgs
- demonstrate benefits to community
- understand cultural barriers
- train interviewers
- go where potential participants are
Simple Random Sampling
identifies cases on basis of chance
Systematic Random Sampling
random sampling where first element randomly chosen and every nth after
Stratified Random Sampling
elements distinguished according to value (ex: pie chart on race) - can be proportionate or disproportionate
Cluster Sampling
select one group, then another group within that group, etc. (useful when sampling frame unavailable)
Availability Sampling
Non-random: convenience sampling - whoever is there
Quota Sampling
Non-random: preset # based on elements of population (purposely picking people out)
Purposive Sampling
Non-random: each element selected for a purpose usually b/c of unique position of a sample elements (ex: survivors of sibling violence)
Snowball sampling
Non-random: identify one member who identifies another member, etc.
Sampling Error Rule
larger sample size = less sampling error, except homogenous population can have smaller sample
Replication
duplicating a study to see if the same evidence and conclusions are produced
Empirical
valuation of observation-based evidence
Straw person argument
someone attacks a particular position by distorting it in a way that makes it easier to attack
Ad Hominem Attack
tries to discredit the person making the argument rather than the argument itself
Bandwagon appeal
when new interventions promoted on their new-ness
Evidence-Based Practice
a process in which practioners make practice decisions in light of the best research evidence available
Process of EBP
- Formulate a Question
- Search for Evidence
- Critically Appraise the Relevant Studies
- Determine Which EBP Intervention is most appropriate
- Apply Evidence-based interventions
- Evaluation and Feedback
Randomized Clinical Trials
experiments that use random means (coin toss) to assign clients who share similar problems or diagnoses into groups to receive different interventions
Paradigm
a fundamental model or scheme that organizes our observations and makes sense of them
Postmodernism
paradigm that rejects the notion of an objective reality and of objective standards of truth and logical reasoning associated with scientific methods
Contemporary Positivism
paradigm that there is an objective answer to the question of what really happened
Interpretivism
paradigm that doesn’t focus on isolating and objectively measuring causes or on developing generalizations
Critical social science
paradigm that views social life as a struggle among competing individuals and groups
Independent Variable
variable that explains or causes something
Idiographic model
when trying to explain a person’s behavior by enumerating the many reasons for it
Nomothetic model
Instead of understand person as fully as possible, try to understand general phenomenon partially
Anchor Points
pieces of information about the various places you might be able to find a particular participant (phone tracking, mail tracking, agency tracking, field tracking)
Linguistic Equivalence
translation equivalence - when an instrument has been translated into another language successfully
Back-translation
When bilingual person translates instrument and instructions into target language
Forward-translation
several bilingual people examine both version to ascertain whether both versions contain same conceptual meanings
Conceptual equivalence
instrument and observed behr’s have same meanings across cultures
Metric equivalence
psychometric equivalence/scalar equivalence - scores on a measure are comparable across cultures
Cross-sectional studies
research studies that examine some phenomenon by taking a cross section of it at one time and analyzing that cross section carefully
Longitudinal studies
studies intended to describe processes occurring over time and conduct observations over an extended period
Trend Studies
studies that study changes within some general population over time
panel attrition
some respondents who are studied in first wave may not participate later
Reductionism
overly strict limitation on the kinds of concepts and variables to be considered as the causes in examining a broad range of human behavior
Conceptualization
refining and specifying abstract concepts
attributes
concepts that make up a broader concept
curvilinear relationship
one in which the nature of the relationship changes at certain levels of the variables
extraneous variables
alternative explanations for relationships that are observed between independent and dependent variables
Moderating variables
variables that affect the strength or direction of the relationship between the independent and dependent variables.
Spurious relationship
one that no longer exists when third variable introduced
Categories of operationalizing variables
- self-reports
- direct observation
- available records
Systematic error
when information collected reflects a false picture of the concept we seek to measure because of way data is collected or dynamics of those who are providing the data
Triangulation
using several different research methods to collect data in order to decrease systematic error
Internal Consistency Reliability
a method that assumes that the instrument contains multiple items
Face validity
if the measure pertains to concepts being measured
Content validity
does the measure cover the full range of the concept’s meaning?
Known-groups validity
Criterion validity: when scores on measurement are similar to already validated or other measures
Construct validity
measure is related to other measurements as specified in the theory
Convergent validity
exists when you can show a relationships between two measures of the same construct
Discriminant validity
when measures correspond highly to similar measures of similar constructs
Factorial validity
how many different constructs a scale measures and whether the number of constructs and the items that make up those constructs are what the researcher intends
Likert Scale
strongly agree, agree, disagree
Semantic Differential
Asking about two opposite positions
Univariate Analysis
the examination of the distribution of cases on only one variable at a time
frequency distribution
the number of times the various attributes are observed in a sample
Standard deviation
bell-curve, how similar to the curve
Bivariate analysis
an analysis that examines the relationship between two variables
Contingency tables
values of the dependent variable are contingent on values of the independent variable
Descriptive statistics
when statistics are reported to DESCRIBE the relationships among the variables in a sample
Inferential statistics
when statistics go beyond just describing a set of sample observations and attempt to make INFERENCES about causes processes or about a larger population
Mail Surveys - ADV/DISADV
ADV: cheaper and quicker that interview, large samples, anonymous
DISADV: expensive, time consuming, lower response rate
Online Surveys - ADV/DISADV
ADV: inexpensive, not time-consuming, large samples, automatic data entry, anonymity, fast
DISADV: under-representative of poor, elders, lower response than face-to-face
Face to Face Interviews - ADV/DISADV
ADV: high response rate, more info collected - fewer missing items, opportunity to clarify, observe, probe
DISADV: more expensive, time-consuming, physical appearance bias, lack of anonymity
Telephone Surveys - ADV/DISADV
ADV: less expensive, less time-consuming, no physical appearance bias, fewer missing items, opportunity to clarify, interviewer safety, computer-assistance possible
DISADV: unlisted numbers, cell phone regularity, easy of hanging up, caller ID