Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
92 Cards in this Set
- Front
- Back
Conceptualization
|
Defining concepts including levels of abstraction (concept of age would have a low abstraction, the concept of emotion would have high abstraction.
|
|
Inductive (Qualitative)
|
How was the quality of parent-child relationships expressed
|
|
Deductive (Quantitative)
|
decide before hand what the definition is.
|
|
process of operationalization
|
how you are going measure your concepts. (eg using surveys, collect unobtrusively, make observation, ask a question )
|
|
operational indicator
|
what are you going to look for when making observations that represent the concept you have defined.
|
|
What is a variable
|
discrete or continuous (fixed or infinite)
|
|
attributes of a variable
|
ex. male or female, married or single
|
|
mutually exclusive
|
fits into one and only one attribute of a variable
|
|
mutually exhaustive
|
fit into more than one attribute of a variable
|
|
ordinal
|
greater than or less than
|
|
nominal
|
variation in kind, type or quality (no ordering)
|
|
continuous
|
fixed measurement of units (eg age, years of education)
|
|
Index
|
use index to measure concepts with high levels of abstraction
|
|
measurement error
|
the degree to which the true score on the variable deviates from the measured score.
|
|
sources of measurement error
|
method factors: researcher error
ex. measurement operation, bad operational indicator, etc *can be controlled |
|
Generic factors
|
common problems
ex. social desirability, sensitive topic, recall *can be controlled |
|
Idiosyncratic factors
|
a few cases random problem
ex. participant unable to focus, etc. |
|
measurement validity
|
accuracy
how well the empirical indicators match what we intend to measure. |
|
Face validity
|
Does the measure make sense?
Logically represents the intended concept more than other concepts |
|
Content Validity
|
Is the measure comprehensive?
full range of the concept's meaning |
|
Concurrent criterion validity
|
empirical association with same concept that was measured in a different way
|
|
criterion validity
|
is the measure comparable to a criterion?
-criterion= external source 1) a more direct measure of the same concept (eg. observation rather than survey) 2) a previously validated measure of the same concept 3) a different concept that should empirically correlate with the concept |
|
Predictive criterion validity
|
-empirical association with different concepts
-does the measure behave as hypothesized? -should have a strong (positive or negative) association |
|
Measurement Reliablity
|
Does the measure consistently yield the same result (given the concept didn't change)?
-assessing reliability is always based on empirical analysis of data (such as an empirical associations) |
|
test-retest reliability
|
comparing two time points of the measured concept within a person.
-or within an observer = intra-observer reliability |
|
inter-observer reliability
|
comparing multiple observers measuring the same concept
|
|
inter-item reliablity
|
internal consistency of an index
-used for multiple indicator indices only to assess unidimensionality -empirical coefficient=Cronbach's alpha (.75 or ^) |
|
sampling
|
the process of selecting cases from the population to be elements in the sample
|
|
Population
|
Is the whole group you want to know about
|
|
Sample
|
the group that has been selected from the population
|
|
Cases
|
selected by chance
ex. every 5th person through the door |
|
Elements
|
number of people in the sample
|
|
Sampling frame
|
a list of all cases in your population
|
|
probability procedures
|
known likelihood of selection into the sample (p=n/N, p>0)
--no systematic bias |
|
non-probability procedures
|
-unknown likelihood of selection into the sample.
-cases are not selected by chance --systematic bias is present |
|
sufficiently large for probability cases
|
generally 1000 elements is sufficiently large for probability cases.
|
|
Where is the common site of Cylindroma?
whats the histology? |
Skin of scalp and forehead
Nest of epithelial cells that are surrounded by thickened basement membrane |
|
sample generalizability
|
the degree to which research findings, such as empirical associations, from the sample hold true for the population
|
|
what is representative
|
Assessing sample representativeness is often based on demographic characteristics (e.g. race, gender, age, religion, social class, etc)
|
|
representative sample
|
will have demographic characteristics that are very similar to the population.
|
|
sampling error
|
the degree to which the sample characteristics deviate from population characteristics
|
|
sources of sampling error
|
-random
-small sample "n" or small sub-group "n" -incomplete sampling frame --a list of cases in the population used for selecting elements in the sample -refusal to participate/failure to locate element -sample attrition (longitudinal studies -non-response to operational indicator (70% response rate is good) |
|
non-respondents
|
individuals selected for the sample but for whom the researcher has no information (i.e. data)
sampling error occurs if non-respondents systematically differ from respondents (i.e. people who provide information) on demographics and other concepts of research interest. |
|
Simple Random Sampling
|
all cases in the population have an equal probability of selection (n/N)
-number cases in the sampling frame and then randomly select cases --lottery, random tables of numbers, etc --random digit dialing |
|
Stratified random Sampling
|
1. put all cases in one and only one stratum
2. randomly select elements within each stratum |
|
strata
|
are based on relevant person characteristics
|
|
proportionate
|
-equal probability of selection for all cases
-% of relevant sample characteristic = % in population |
|
disproportionate
|
(balancing out population groups)
-probability of selection varies by strata --higher probability among cases within "small" stratum --lower probability among cases within "large" stratum --% of relevant sample characteristic is not equal to % in population |
|
Cluster
|
naturally occuring aggregates
-write list of all clusters 1. draw random sample of clusters 2. randomly select elements within each cluster -multi-stage cluster sampling |
|
cause
|
A cause is an explanation for some characteristic, attitude or behavior of individuals, groups (such as families or organizations) or for events.
|
|
nomothetic causation
|
-causal explanation (based on empirical evidence)
--Variation in one variable (independent), on average, leads to variation in another variable (dependent), all other variables being equal --probabilistic tendencies: average experiences |
|
causal validity
|
the degree to which we can trust our conclusion that the independent variable (x) causes the dependent variable (y)
--three criteria |
|
1. emperical association (ie found in data)
|
x and y must vary together
x is associated with y ex. positive associate x:education ---> y: $$$ higher levels of education are associated w/ higher levels of annual income |
|
Simple Random Sampling
|
all cases in the population have an equal probability of selection (n/N)
-number cases in the sampling frame and then randomly select cases --lottery, random tables of numbers, etc --random digit dialing |
|
Stratified random Sampling
|
1. put all cases in one and only one stratum
2. randomly select elements within each stratum |
|
strata
|
are based on relevant person characteristics
|
|
proportionate
|
-equal probability of selection for all cases
-% of relevant sample characteristic = % in population |
|
disproportionate
|
(balancing out population groups)
-probability of selection varies by strata --higher probability among cases within "small" stratum --lower probability among cases within "large" stratum --% of relevant sample characteristic is not equal to % in population |
|
Cluster
|
naturally occuring aggregates
-write list of all clusters 1. draw random sample of clusters 2. randomly select elements within each cluster -multi-stage cluster sampling |
|
cause
|
A cause is an explanation for some characteristic, attitude or behavior of individuals, groups (such as families or organizations) or for events.
|
|
nomothetic causation
|
-causal explanation (based on empirical evidence)
--Variation in one variable (independent), on average, leads to variation in another variable (dependent), all other variables being equal --probabilistic tendencies: average experiences |
|
causal validity
|
the degree to which we can trust our conclusion that the independent variable (x) causes the dependent variable (y)
--three criteria |
|
1. emperical association (ie found in data)
|
x and y must vary together
x is associated with y ex. positive associate x:education ---> y: $$$ higher levels of education are associated w/ higher levels of annual income |
|
Time order
|
variation in x must occur before variation in y
ex x:edu --> y:$$$ grad in 07 in 09 |
|
nonspuriousness
|
an association is spurious when it is due to variation in a third variable
ex x ---> y ^z^ there is no causal link between x and y z is an extraneous variable - also called control variables or confounding covariates |
|
causal validity is increased when...
|
-x is empirically associated with y
-x occurs before y (time order is established) -the association between x and y is shown to be nonspurious |
|
meeting causal criteria in research
|
-ensuring association
--bivariate statistical analysis of x and y |
|
ensuring proper time order
|
cross-sectional studies(tricky)
--x is an attribute assigned at birth, a prior circumstance or a discrete event --panel studies (longitudinal) (better) ---measure x at time 1 and y at time 2 |
|
ensuring nonspuriousness
|
first researcher must design their studies to measure a large # of and the appropriate kind of extraneous variables
-then multivariate statistical analyses with x,y and z |
|
statistical control
|
an extraneous variable is held constant and the association between x and y remains the same (effect size and statistical significance)
|
|
Writing Good Survey Questions: Avoid Confusion
|
-ask very specific and clear questions
-avoid very long questions and sentence fragments -avoid double-barreled questions |
|
Writing Good Survey Questions: Avoid bias
|
-avoid "emotional" language
--eg feminist, terrorist, prejudice -avoid status associations --eg "Obama-care", prestigious or otherwise -avoid one-sided or leading questions --To what extent do you oppose a tax increase? --How mad does your mother-in-law make you? -Avoid overlapping or unbalanced response choices |
|
Writing Good Survey Questions: Additional consideration
|
-Sensitive or threatening questions
--illegal behavior, sex, money, etc -Expecting too much of the respondent --memory ---use specific and reasonable time periods --Knowledge ---floaters: include a "don't know" response choice --relevance ---use skip pattern/contingency questions ---confusing beliefs with reality Respondents cannot give valid and reliable responses to confusing, biased or otherwise problematic |
|
Method of Data Collection
|
-nuts & bolts
-strengths & weaknesses -ethical concerns |
|
Surveys
|
-question: brief response
-secondary data |
|
Participant Observation
|
-observe: social setting
|
|
Content Analysis
|
-unobtrusive: books, newspapers, etc
|
|
In-depth Interviews
|
question: elicit long detailed responses
|
|
Experiments
|
-expose to stimulus then observe or question. (all are qualitative data)
|
|
Focus
|
-question group: elicit long group discussion
-observe: group discussion |
|
Surveys -Research Design
|
Ex. 2000 presidential election survey design during the Bush/Gore election
|
|
Surveys- Research Questions
|
-prevalence and predictions: describing large populations
-self-reported beliefs, behaviors, characteristics, etc |
|
Surveys- Conceptualization
|
-define concepts prior to writing survey questions
|
|
Surveys- Operationalization
|
-operational indicators: asking survey questions determine, validity and reliability of the measure
|
|
Surveys- Sampling
|
Probability procedure
-generalizability is very strong |
|
Surveys- Causal Validity enhanced by:
|
-longitudinal panel designs compared to cross sectional designs
--to establish time order -(number) and quality of non-focal in survey --to establish non-spuriousness |
|
Survey Instrument: Interview Schedule
|
-administered by trained interviewer
--in person or by telephone |
|
Survey Instrument: Questionaire
|
-self-administered
--computer/web or paper & pencil |
|
Response Choices to Survey Questions
|
Open ended: no response choices provided to the interviewee
-used when unknown range or many possible responses --survey questions must be very specific Close ended -mutually exclusive (not overlapping and exhaustive --exception: checks all that apply *** Respondents cannot find on appropriate answer they will either skip the question or provide an innaccurate response*** |
|
Designing the Survey Instrument
|
-assurance of anonymity or confidentiality
-directions for --answering questions(questionaire) --asking questions (interview schedule) -question order --starting with interesting questions -group questions by theme --general questions first then specific questions -sensitive topics asked towards the end of the survey |
|
Questionaire formatting
|
-attractiveness
--plenty of white space --vertical (not horizontal) response choices -clarity --formatting of directions differs from question stem --skip patterns: use of graphics and provide instructions |
|
Indices
|
-provide instructions
-response choices for each question must be the same --matrix format response choices |