Threats to Research Flashcards
Reliability
If I measure the same person a second time, I should get similar results. Cannot trust data from an inconsistent measure.
Validity
The extent to which your measure measures what it’s supposed to measure.
Face validity
Items appear to assess the construct of interest, no items that clearly don’t assess the construct. (based on logic)
Content validity
Like face validity, except purpose is to ensure adequate measurement of the construct. (also based on logic)
Predictive validity
The measure correlates with related future outcomes (GRE)
Can data be used if your data collection measures are not reliable and valid?
No
Extraneous variables
- Confounding variables that change along with levels of IV. ex: one group with an upbeat researcher, the other with a solemn researcher.
- Nuisance variable: introduces variability to DV but not a deal breaker
What is a subjective construct?
Something less easy to measure, like locus of control or self confidence.
What does construct sensitivity refer to?
A measure that is not very sensitive to changes in behavior- require large changes in behavior before changes in measurement scores will be observed. (Beck Depression Inventory)
Restriction of range
A threat to research in that it overly limits how people can respond ie if on teacher reviews, you are only given positive options for responses.
Reactivity
People behave differently when they know they are being observed.
Experimenter bias
We can be inconsistent when we want participants to behave a certain way.
Ways to control threats
- plan carefully
- many threats can be eliminated by study design
- randomly sample participants
- operationalize variables, have reliable and valid measures, and have standard measurement procedures.
- blind studies
What is the general rule for “enough” participants for valid data?
30
Skew
Skew is when the mean of the distribution is shifted left or right (positive or negative)
- “tail of the whale”
- side the tail is on indicates + or -
- if the long tail of the curve points to the right side, the distribution is positively skewed
Kurtosis
when too many or too few participants are at the mean
Frequency
number of times that you got a certain score within a data set
Relative Frequency
f/N
Percentile rank
Cumulative frequency/n- tell us how a person scored relative to others, based on their ranking within the distribution.
Cumulative Frequency
Running total of all scores observed
Sources of data
- direct: gathered from individuals through interview, physical measures, test administration.
- indirect: data gathered from existing sources-archival data like school records, hospital records, info on websites
IRB
International Review Board: governing body that reviews research proposals using human participants and decides if research meets ethical standards. Overarching goal is to protect participants.
Mesokurtic
a normal distribution that looks like a bell curve
Negatively skewed
when the bulk of subjects are on the right and the long tail is pointing to the left
Positively skewed
when the bulk of subjects fall to the left side of the curve and the long tail of the curve points to the right side
Construct robustness
some constructs are hard to measure with any degree of power (such as birth order or locus of control)
Internal validity
does the study control extraneous variables that might have explained results?
External validity
do the results generalize to the population at large?
How is probability expressed?
p=.05 indicates 95% probability, meaning that chance is operating only 5% of the time (5 times out of 100). This is an accepted standard in psychology research: if a finding is influenced by chance no more than 5% of the time, researchers are happy.
