module 4 - MIDTERM 2 Flashcards
(42 cards)
internal validity
- how much you can trust that the findings of the study are due to the factors explored in the study
- confounding variables/lurking variables
- control groups and randomization
- clinical sample vs. non-clinical sample vs. analogue sample
confounding variables/lurking variables
- factors that you did not explore
- can potentially affect findings/results but you did not include them in your study (“oopsies”)
control groups and randomization
- control group can be used as a comparison group
- randomization involves randomly assigning individuals to groups and it is used to eliminate systematic differences
clinical sample
involved in a clinic study where participants may be diagnosed as a part of the study
non-clinical sample
everyday folk, general public, undergrad students
analogie sample
- experimental design where the procedures or participants used are similar but not identical to the situation of interest
- e.g. recreating situations as similar as possible; simulated scenarios; mundane realism
external validity
- do these results actually apply to the real world? Are they generalizable?
- high external validity = highly applicable to the real world
- statistical and clinical significance
- completer analysis vs. intent-to-treat analyses
- effect sizes
statistical significance
- a result or difference is meaningful enough/mathematically meaningful enough between variables that wasn’t caused solely by chance
- e.g. Does this treatment work more than that treatment?
- affected by sample size and type of statistical test used
completer analysis
- any one dropped it is not included at all; only include those who completed treatment are included in the analysis
- assumed people who dropped out wouldn’t have responded to treatment anyway
- often times overestimates the efficacy of a treatment because it only included people who stayed throughout the study
intent-to-treat analysis
- the people who dropped out would not have changed anymore since they dropped out
- assume that the person didn’t change since dropping out so whatever scores they had when dropping out will be used at the end for the analysis
effect sizes
- pearson’s correlation
- looking at the relationship between to variables
- if one variable increases what will happen to the other?
- a perfect correlation is 1; as one variable goes up by one so does the other variable
interpretation of correlations
- 0 → .10 = no relationship
- R = .10 → .30 = weak correlation
- R = .30 → .50 = moderate correlation
- R = .50+ = strong correlation
clinical significance
- if expert in the field believes a statistically significant finding is large enough/meaningful enough to be clinically important and should therefore direct the course of patient care
- we decide the difference is great enough we should change the way we treat
- just because something is statistically significant doesnt mean its going to be clinically significant/clinically meaningful or important
correlational research
data from how things naturally occur
cross-sectional designs
data from one time point
cohort effects
- gathering information from different groups
- e.g. gathering information of marijuana use from 22 year olds and 80 year olds today
longitudinal designs
- data at multiple time points over a period of time from the same group (typically years)
- make inferences about cause and effect
- expensive, take a long time and are rare
cross-generational effects
- cohort effects but in longitudinal research
- e.g. if you followed a group of 20 year olds for 20 years would the next group of 20 year olds be the same
epidemiological research
- correlational research that involves studying the prevalence, distribution and consequences of disorders in populations
- prevalence and incidence
prevalence
- percentage of the population that have the disorder
- 12 month prevalence: % of population that have had the disorder in the past year
- lifetime prevalence: % that has ever had the disorder in their life
incidence
- number of new cases of the disorder in a specific time
- typically in the past year
- incidence is normally lower than prevalence
treatment outcome research
- experimental research involves manipulation
- outcome research is interested in the effects of research and treatment
open label trials
participants know what they are signing up for and what treatment they are getting
randomized control
randomly assign people to a control trial; treatment or no treatment, they don’t know what they will get