Chapter 3 Flashcards
A variable is …
something that varies. Thus, it must have more than one level (i.e., exist in more than one state).
Any given variable may be either
measured or manipulated.
T/F: some variables may be measured OR
manipulated, depending on the
nature of the study.
T
Any variable can be considered in two different ways:
- As a conceptual variable, or construct: the variable defined in terms of its definition, i.e., as an abstract concept.
Stress: “an adaptive response to a perceived danger or threat that involves physiological, cognitive, affective, and behavioral - As an operational variable: the variable define in a more specific way that allows for manipulation / measurement.
So, to operationalize a variable is to decide exactly how that variable will be manipulated or measured.
E.g., How should we operationalize stress?
components.”
One method with animals to manipulate stress
restraint stress
One way with humans to manipulate stress
The Trier social stress test (TSST)
The Trier social stress test (TSST)
mainly consists of an anticipation period (10 min) and a test period (10 min) in which the subjects have to deliver a free speech and perform mental arithmetic in front of an audience
Measuring stress
Questionnaire, interview or diary, or physiological measures (Skin conductance, heart rate, blood pressure)
What’s a claim?
Any sort of argument being made
Three types of claims
- Frequency
- Association
- Causal
Frequency claims
the claim regards the prevalence (or frequency) of a particular behaviour or phenomenon. Note that there is only one variable, and it is being measured.
Association claims
In such cases there are two variables, both measured.
The claim involves the association between them.
association claims involve
correlations between two variables, often plotted on scatterplots (one variable on each axis)
Causal claims
again involve two variables, with the key difference that
one variable is manipulated, while the other is measured.
This enables the researcher to test a causal hypothesis.
Association claim verbs
linked to, associated with, correlated with, prefers, may predict, tied to, goes with
Causal verbs
Causes, affects, promotes, reduces, increases, decreases, makes, changes, leads to, adds, etc.
T/F: Interrogating a claim involves evaluating it based on its validity: the appropriateness of its conclusion or decision. That is, whether the claim is reasonable, accurate and justifiable.
True
four types of validity
- Construct validity: How well is/are the variable(s) operationalized?
- External validity: How generalizable are the findings (the claim)? To what degree do the findings generalize to or represent people and contexts beyond the study?
- Statistical validity: Are the statistical conclusions accurate and reasonable? That is, do the statistical results support the claim?
- Internal validity … is part of a larger question related to the criteria for causation and will be discussed a little later.
Interrogating frequency claims, stat. validity
For frequency claims, it usually comes down to the
margin of error of the estimate , which indicates how close your
results are to the overall population.
Sample size’s effect on margin of error
The bigger the sample size, the smaller (narrower) the margin of error of the estimate
Construct validity info will almost always be in the
methods section
statistical validity will be in the
results section
To visualize an association
scatterplot data
To statistically analyze an association
we typically calculate an r value
The r value , or coefficient of correlation , describes the
association in two ways:
- Its direction : A positive r value represents a positive association (as one thing increases, so does the other); negative represents negative.
- Its magnitude : How strongly are the variables
How large is the number (ranges from 0 to 1)
To infer whether that association is
statistically significant, we can calculate a
p value and r-value (correlation)
p-value
represents the likelihood that the relationship seen is due to random chance
The threshold for statistical significance is
0.05. p < 0.05 means that there is
less than a 5% chance that what we see is just due to random chance. In this case, we
call the relationship statistically significant.
The p value is affected by sample size. For a given r value , the resulting p value will go down as sample size goes up.
value will go down as sample size goes up.
External validity of association claims has
Lower expectations than frequency claims.
As with all causal claims, there are two variables:
1) An independent variable that is manipulated
The manipulated variable is often easy to operationalize.
2)dependent variable that is measured:
Interrogating causal claims: statistical validity
p-value
establishing causation between two variables.
That is, not just that A and B are related, but that A causes B.
There are three criteria for causation:
- Covariance do the two variables change/vary together? As one changes, does the other change in a predictable manner?
- Temporal precedence does the “causative” variable (and its manipulation) come before the other variable?
- Internal validity Are there any alternative explanations for the observed relationship? Confounds? Maybe a “third variable” influencing both variables being examined? is less cut and dry. If we think long
enough, we can probably come up with a possible confound or two for any study, but did the researchers do a reasonable job?
Ways to assess internal validity
Having inclusion criteria and using random assignment
