Ch 3: Interrogation tools Flashcards
Variable
- something that changes/ varies
- must have at least two levels
levels
values within a variable
constant
something that doesn’t vary- has only one level (in this case)
measured variable
variable is observed and recorded by researcher an as it occurs naturally
example of a measured variable
age, IQ, gender
How do you measure abstract variables?
using sets of questions to represent different levels
- ex: stress, depression
manipulated variable
- a variable that a researcher controls
- usually by assigning participants to levels
example of a manipulated variable
assign some to take a test in a full room vs alone
can all variables be manipulated or measured?
- some can only be measured
( ex: age) - some unethical to manipulate
( ex: assigning low quality vs high quality schooling) - some can be manipulated or measured
(ex: measure kids taking music or drama, or assign kids to music or drama)
Conceptual Variable (or Constructs)
- abstract theoretical concepts
example of a conceptual variable
- infant temperment
- anxiety
conceptual definitions
defining conceptual variables at a theoretical level
operationalizations (or operational variable)
- turning a conceptual definition into a measured or manipulated variable so it can be tested
operational definition
- operationalizing/ defining a conceptual variable in the terms of the exact procedures used to measure or manipulate it
what are some ways “school achievement” could be operationalized?
- self-report questionnaire
- checking school records
- teacher observations
claim
an argument someone is trying to make (about variables)
what are the 3 types of claims?
- frequency
- association
- causal
frequency claim
- describes a particular rate/ degree of a single variable of interest
- describes how frequent/ common something is
is the variable in frequency claims measured or manipulated?
in frequency claims, the variable is always measured
association claim
- involve two variables and their relationship to each other
- one level of a variable is likely associated to a level of another variable
- both measured variables
3 types of associations
- positive
- negative
- zero
positive association
high (scores on variables) goes with high, low with low
- ex: partners who express gratitude 3x more likely to stay together
Negative association
High scores on variables goes with low of another, low with high
- ex: people who multitask most are the worst at it
zero association
- no association btwn variables
- no trend to data points
- ex: childhood obesity not linked to autism
how can association claims be useful
useful because they help make predictions
- if we know a level of a variable we can predict the level of the other
- the stronger the association between the variables, the more likely the prediction is accurate
predicting
using associations to make an estimate more accurate
Causal claims
argues one variable is responsible for changing another
- one of the variables is measured and the other manipulated
3 criteria for causal claims
- they start as association claim
- temporal precedence (causal variable must come before the outcome variable)
- no other explanations (confounds- internal validity)
where else can claims come from (non research)?
- popular press
- intuition
- experience
- authority
what does it mean to say something is valid?
- reasonable
- accurate
- justifiable
What are the 4 validities?
- construct validity
- external validity
- statistical validity
- internal validity
what is construct validity?
- how well conceptual variables are operationalized
- how well the variables are measured or manipulated
External Validity
- How well do the participants represent the intended population?
- what context?
Statistical Validity
Extent to which the data supports the conclusions
Internal Validity
- Only relevant in in causal claims
- How confident are we that the first variable causes the second, and that that is the only explanation ( no confounds)
How is Construct Validity used in frequency claims?
ask how was the conceptual variable measured?
ex: 80% college students depressed last year
- how was depression operationally defined?
How is External Validity used in Frequency claims?
ask how the participants were chosen, how well they represent the intended population
ex: 72% of the world smiled yesterday
- only urban areas? how were participants chosen
How is statistical Validity used in frequency claims?
percentage is usually accompanied by a margin of error
- ex: 41% teenagers text while driving- report included “+/- 2.6% margin of error
margin of error
- statistical figure based on sample size that indicates where the true value in the population probably lies
- helps describe how well the sample estimates the true percentage
Correlational Studies
- studies to support association claims
- measures 2 variables
How is construct validity used in association claims? (+ example)
- how well were the two conceptual variables operationalized to be measured?
- ex: people who multitask more are worse at it
- how is frequency of multitasking measured? - self report or observation?
- how is ability to multitask measured? - self report or scored exercise?
How is external validity used in association claims?
can the claim generalize to other populations, contexts, times, places?
How is statistical validity used in association claims?
- how strong is the relationship between the two variables?
- how statistically significant is the relationship?
statistical significance
- means result is probably not due to chance of sampling error
- if you had access to whole population of interest we would probably see same pattern of results
what are the mistaken conclusions that can come about in association claims?
- Type I error (false positive): study might mistakenly conclude from sample an association between variables when there is no association in the population
- type II error (miss): study might mistakenly conclude from sample no association when there is an association in the full population
Sampling error:
natural discrepancy that happens between population and a sample used to represent the population
what are the 3 criteria for causation?
- covariance
- temporal precedence
- internal validity
covariance
two variables are related/ associated
temporal precedence
one variable must come before the other in time
internal validity/ third variable criteria:
study should be able to eliminate external alternate explanations
- want to be confident that variable A leads to change in variable B, and nothing else changed
experiment
study in which one variable is manipulated and the second is measured
independent variable
a manipulated variable that we believe will cause a change in the dependent variable
dependent variable
in a causal claim- the measured variable which we believe will be effected by the independent variable
random assignment
method used to assign participants to different levels or conditions of the independent variable in a way that each person is equally likely to be assigned to any condition
how does random assignment increase internal validity
it controls for potential alternate explanations due to variables within participants
Music lessons enhance IQ experiment- what are the variables, and the levels?
independent variable: music lessons
dependent variable: IQ
variable levels of music lessons:
- keyboard
- voice
- drama
- none
results: kids with music lessons gained
Music lessons enhance IQ experiment- what are the results?
- kids with music lessons gain avg. 3.7 IQ points
- statistically significant
- establishes covariance
- method establishes temporal precedence and internal validity
How is Construct Validity used in Causal claims?
need to see if manipulated and measured variables were operationalized well
Relationship between internal and external validities
Internal Validity is emphasized in experiments- as experimental control increases, externally validity decreases
How is statistical validity used in causal claims?
- what is the strength of the relationship between the independent and dependent variable?
- statistical significance
How do researchers prioritize validities?
- best research programs employ multiple types of methods to address their research questions to emphasize different validities in each of their studies
