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;
67 Cards in this Set
- Front
- Back
2 major categories of study designs
|
parallel design & cross over design
|
|
pre-experimental study designs
|
case study
one group pre- & post- test static group comparison limited usefulness & are usually preliminary studies that lead to more rigorous study designs |
|
true experiemntal study designs
|
pre- post- test w/ control group
post-test only w/ control group solomon four group design cross over design |
|
quasi experimental study designs
|
time series
non-equivalent control group are excellent designs if selection problem can be controlled |
|
epidemiologic study designs
|
descriptive or case series
case-control cross-sectional studies cohort studies |
|
*primitive research design
*intervention occurs & observation is made *doesn't control for threats to internal validity *advantage-good for identifying problems for future research *disadvantages-no control for internal validity |
one-shot case study
|
|
*this design impllies an increased sample size (more than one subject)
*collects data before & after an intervention or experiment no control group & results are worthless *advantages-before measures allows for established baseline to be established; comparison can be made w/ 1st measure *disadvantages-little control so internal validity still prob |
one group pre- post-test
|
|
*differences is 2 different groups vs same group in prev design
*assumes both groups are same at outset *no randomization *advantages-both maturation & history effects are controlled for IF the groups were the same at outset, there is no guarantee they were; testing & reactive effects are controlled *disadvantages-selection bias & mortality are major concerns |
static group comparison
|
|
*randomization
*2 groups w/ pre & post measures *we are able to better attribute any change shoud to intervention *advantages- controls for history, maturation, & testing effects, but you have not removed this concern *disadvantages-doesn't control for reactive effect to testing; history problems |
pre- post- test w/ control group
|
|
similar to pre- post- test w/ control group w/ addition of 2 groups added that are not pre tested
* controls for reactive effect to testing *allows researchers to tease out the differences due to intervention X *ANOVA - a powerful parametric test, is used to partition the between group & within group variances; adds costs to studies & controls for everything *very stron test design, but not always practical |
solomon four group design
|
|
*also very strong design
*testing effect & reactive effects are not a problem *since we have no initial score, we can't be sure that changes are due to intervention X |
post test only w/ control group
|
|
*every subject gets every treatment & data is collected
*order in nwhhich they receive treatment varies *must have wash-out period between treatments *each subject becomes their own control *eliminates concern over equality of experimental & control subjects & can have same # of subects as # of treatments |
latin square or cross over design
|
|
*expansion of latin square
*will not see this very often b/c large sample sizes are required = expensive *pts leave trial *strong design that minimizes variance component |
counter balanced design
|
|
*most commonly used w/ retrospective data
*no randomization, but samples are enerally so large, we can elimante this concern *problems occur when extrapolating to an even larger group *seasonality is a concern |
time series
|
|
*better than time series
*provides a control group *concern over equality of control group *still no randomization *typically analyzed using ANOVA w/ trend analysis component |
multiple time series
|
|
*same design as pre- post- test w/ control group but no randomization
*focus on how groups are selected *self selection is a problem |
non-equivalent control group
|
|
study of the distribution & determinants of diseases & injuries in human populations
concerned w/ frequencies, groups, & factors |
epidemiology
|
|
study of effects of drugs in humans
|
clinical pharmacology
|
|
study of use & effects of drugs in large numbers of people
|
pharmacoepidemiology
|
|
the investigator studies the impact of varying some factor that s/he controls
|
experimental studies
|
|
contrasts in outcomes between the study groups are observed & analyzed, not created expeirmentally
|
observational studies
|
|
rates that measure how many people will develop a disease, its the number of new cases
|
incidence
|
|
rates that measure the number of people in a population who have the diseas
|
prevalence
|
|
people diagnosed as having the disease are compared w/ people who do not have the disease
compares cases & controls w/ regard to some element in their past |
retrospective studies
|
|
starts with group of people (cohort) all considered to be free of a given disease, but who vary in their exposure to a supposed noxious factor
|
prospective studies
|
|
-similar to one-shot case study
-report of single pts -exposed to drug & experienced a particular, usually adverse, outcome -not typically useful to document causation -can be very useful in challenge-rechallenge situations (adverse events) |
case report
|
|
-collection of pts, all have single exposure, whose clinical outcomes are then evaluated & described
-can be collections of pts w/ single outcome, whose antecedent exposures are then examined -similar to one-shot case study -simple, descriptive account of interesting characteristics observed in group of pts -generally involves subjects seen over short period of time in single setting -no control subjects -useful for quantifyng the incidence of an adverse reaction |
case series
|
|
-epidemiologic study where subjects are selected on the basis of whether they do or don't have particular disease
-subject selection -usually retrospective in nature -good study design for rare disease -relatively efficient in time & costs comared to clinical trials -has the largest # of possible biases or errors |
case-control
|
|
-analyze data collected on a group of subjects at one time rater than over a period of time
-designed to determine "what is happening" right now -think survey -focus on point in time (prevalence studies) |
cross-sectional studies
|
|
-epidemiologic study in which an exposed group to a drug or disease is compared to an unexposed group, & both gropus are followed over a period of time, to determine the unintended drug effect on the incidence rate for both groups
-think forward looking |
cohort studies
|
|
the target to which investigators generate study results
|
population
|
|
the object about which the researcher wishes to draw conclusions based on the study
|
unit of analysis
|
|
method used to select the sampling units
|
sampling design
|
|
list of sampling units from which the sample is selected i.e. phone books, pt records membership lists
|
sampling frame
|
|
how study participants are selected to enter the study; has more to do wtih external validity issues
|
random selection
|
|
how study participants are assigned to each treatment in study
|
random assignment
|
|
probability sampling
|
*all sampling units have known, non-zero, probability of being selected
*sampling designs are best methods to obtain an unbiased sample that can be generalized to a larger population *random procedures to select a sample *random sampling doens't guarantee a representative sample (more true for small sample..as samples get larger they are more representative of the population) *simple random sampling *systemic sampling *cluster sampling *stratified sampling |
|
non probability sampling
|
*probability of selecting sample unit is not known
*not everyone in the study's population has an equal opportunity to be selected *more likely to yield a biased sample & may not represent the population of interest & thus not generalizable *more convenient, less expensive, less time consuming *convenience sampling *snowball sampling *quota sampling *purposive sampling |
|
population characteristics, analysis
precision of estimates availabl resources response rate of a survey study design |
sample size determination
|
|
-every unit in population has equal probability of being included in sample
-used when population is homogeneous -assign number to each unit in population from 1 to N -desired sample size N -select (n) different random #s between 1 & N using a random number table |
simple random sampling
|
|
-selects every kth unit after random start
-easier than simple random sampling -be cautious of cyclical patterns that may coincide w/ sampling interval -randomly choose interval (k) independent of sampling frame -randomly choose starting point between 0 & k & then choose every kth unit (ie k=10, randomly choose a # between 0 and 9. that will be starting point & from there you choose every 10th unit) |
systematic sampling
|
|
-breaks population into groups or clusters
-cluster can be natural groupings such as neighborhoods -randomly select sample of clusters using simple random of systematic sampling technique |
cluster sampling
|
|
-population is divided into non-overlapping groups or categores (strata)
-independent simple random samples are drqen for each stratum -increases the representation of populations that are small proportions of total -can be proportional stratified sample or disproportional stratified sample |
stratified sampling
|
|
selects from each strata a sample size from each strata proportional to that strata's composition in population. this allows for direct generalization from the entire sample to the population
|
proportional stratified sample
|
|
this sample does not take into account the proportion of each strata in the population. this can lead to one or more strata within the population to be underrepresented. unlike a proportioned sample, a disproportionate stratified samples doesnot allow researchers to make direct generalizations .
|
disproportioanl stratified sample
|
|
-relies on availabe subjects
-people met on street, volunteers, captive audiences -quick & easy but does get whole representative of population |
convenience sampling
|
|
-specifies desired characteristics & selects for a sample appropriate ratios of population that fit the desired characteristics
-similar to stratified sampling, but NO random selection -may use any methods to select a sample to fulfill assigned quota |
quota sampling
|
|
-relies on information from other subjects to identify additional subjects for study inclusion (ie referrals)
-those subjects provide leads to more subjects -useful for studying populations that are difficult to identify or access |
snowball sampling
|
|
-elements are selected based on expert judgment
-used when sample size is small & simple random sampling may not represent the most representative elements -used to identify pts for clinical trails that have appropriate disease state for trial |
purposive sampling
|
|
two types of surveys
|
cross sectional & longitudinal survey
|
|
surveys can...
|
-describe single variables
-show relationships between variables -study effects of treatments -study longitudinal changes by collecting data at different points of tiime -make comparisons between groups |
|
cross sectional survey
|
most common
-standardized data is collected from a predetermined population (short period of time) -importnat to identify appropriate subjects to suppply desired information -generally difficult to survey the entire target population |
|
longitudinal survey
|
multipurpose survey
-follows a single sample w repeated surveys over period of time -in addition to the core questions, non-core questions may be asked to account for latest developments & iinterests -trend studies |
|
type of question:
how satisfied are you w/ PHA 444? |
attitude
|
|
type of question:
true/false: vitamins help with your health |
knowledge
|
|
type of question:
how often do your kids take their vitamins? |
behavior questions
|
|
type of question:
IQ, age, race |
attributes
|
|
type of question:
what should be done in order to educate consumers about generic drug products? |
open ended question
|
|
type of question:
in your opinion what is the biggest barrier to etting consumers to use generic drug products? |
open ended question
|
|
type of question:
how did you find out you received generic drug? a-pharmaicst told me b-MD told me c-both rx& md told me d-read label |
close ended question
|
|
type of question:
how likely are you to ask for generic drug? a-very likely b-somewhat likely c-unsure d-somewhat unlikely e-very unlikely |
close ended w/ ordered choices example
|
|
type of question:
how did you find out you received generic drug a-pharmacist told me b-MD told me c-both rx & MD told me d-read label e-other_____- |
partially closed example
|
|
type of question:
please rank following from most likely to least in terms of who you would ask about an over the counter medication? (1 = most likely, 5 = least likely) __pharmacist __MD __nurse __family member __good friend |
rank ordered example
|
|
survey type of question:
if you have used generic drugs, how satisfied are you w/ their performance? a-highly satisfied b-mildly satisfied c-neither satisfied nor dissatisfied d-midly dissatisfied e-highly dissatisfied |
attitude example
|
|
survey type of question:
true/false: generic drugs are as safe as trade name drugs |
knowledge example
|
|
if you pay for your own prescriptions out of your own pocket, do you ...
a-request trade name product b-request generic product c-have no preference |
behavior example
|
|
type of question:
are you currently married? a-yes b-no |
attribute example
|