PA 614 EPI QUIZ 1
3 main types of variables continuous 0-infinity with all the decimals in between ie height of family members dichotomous- yes/no, this/that binary ie biologic sex (not including intersex) discrete: ie whole numbers ie number of dice in a bag
3 main types of variables (not iv/ dv)
status quo- how it is at point in time ex post facto- reactive data is post event and due to event longitudinal- across time cross sectional- several snapshots to predict a course of events ie freshman, sophomore, junior and seniors all asked about sat stress. all asked at same time, and predict stress goes up as your grade level does
4 types of non-experimental survey research
hawthorne effect
A change in a subject's behavior caused simply by the awareness of being studied
post test only
A group of 20 PA students arrive at HPB for their 8:00 class. Ten of them are given a cup of free coffee. One hour later, all 20 students are administered the Epworth Sleepiness Scale. Posttest-Only Pretest-Posttest Solomon Four Time Series
case series
A hospital does a study on all of its kidney transplant patients who suffered GI complications after surgery. Cohort study Cross-sectional study Case series Case-control study
coefficient of stability
A produce vendor at the farmer's market charges their customers by the weight of the food they want to buy. They've been using the same scale for six years and have never calibrated it. Which of the following would be most appropriate to analyze the reliability of the scale? a) Coefficient of internal consistency b) Coefficient of equivalence c) Coefficient of stability
Action Research Model
A strategy of Organizational Development (OD) that typically involves a structured processes :learn by doing 1 problem identification 2 develop an intervention to address the problem (action plan) 3 collect data 4 analyze data 5 adjust theory and try again till success or definitive failure 6 report findings generally a loca level thing
hypothesis
A testable prediction, often implied by a theory - a statement that predicts the outcome of the experiment "if-then"
poor intra-rater reliability
An Olympic judge from Russia gives the figure skating team from her own country a better score than the team from Canada, even though the Canadian team performed a very similar routine. Does this represent which of the following? a. Poor inter-rater reliability b. Poor intra-rater reliability c. Poor face validity d. Poor external validity
construct validity
An employer wants to measure job satisfaction in their employees. They decide to do this by having their employees fill out a questionnaire. How well the answers on the questionnaire truly reflect the employees' job satisfaction would be a measure of_______. Predictive validity Construct validity Concurrent validity
Non-exp study types
Case-series/Case report Case-control Cross-sectional Cohort Important to remember: observational studies only observe, they do not intervene.
Coefficient of internal consistency (Cronbach's alpha)
Consider the following two items on the Myers-Briggs test: i) You regularly make new friends. (Agree/Disagree) ii) You feel comfortable just walking up to someone interesting and striking up a conversation. (Agree/Disagree) You want to find out whether these two items truly both measure the same thing. What reliability measure could you use? Coefficient of internal consistency (Cronbach's alpha) Coefficient of equivalence Coefficient of stability
interval
Data can be put into meaningful order, AND the distances between data points are EQUAL but they don't start at 0 Examples: temperature in Celsius or Fahrenheit, GRE scores
ratio
Data can be put into meaningful order, AND the distances between data points are equal, AND there is a "true zero" value. Examples: speed in miles per hour, height in inches, temperature in Kelvin
ordinal
Data can be put into some kind of meaningful order, but the distances between the data points vary. (no true 0) Examples: finishing order in a race (1st, 2nd, 3rd) an education level (high school, college, graduate degree) Like scales
concurrent validity
Elizabeth Holmes attempted to invent a new type of blood analyzer which could run multiple laboratory tests with a single sample of capillary blood. When tested against established laboratory equipment, the new analyzer almost never gave accurate results. This is an example of poor________ Content validity Predictive validity Concurrent validity
True
Experimental designs will always have both a control group and a treatment group. True False
experimental/environmental threat
Hawthorne Effect: When people know they're being watched, they change their behavior. Rosenthal Effect: The investigator's expectations about the study affect the outcome.
False
Internal validity concerns how well the results of a study or experiment can be generalized to the wider population. True False
Ratio
Length in centimeters Nominal Ordinal Interval Ratio
observed only
Non-experimental research is usually observation --- no variables manipulated
content validity
PA students have been busily preparing for an exam at the end of their cardiology module, which they expect will test their knowledge of cardiology concepts. However, there has been a mix-up, and their test contains several questions on infectious disease! This represents a threat to what type of validity? Content validity Concurrent validity Construct validity Predictive validity
predictive validity
Refers to the function of a test in predicting a particular behavior or trait. - How well does the score on a test predict some other criterion measure?
case-control study
Researchers match patients with ovarian cancer to patients of the same age, SES, and geographical area who do NOT have ovarian cancer, in an attempt to identify risk factors for developing the disease. Cohort study Cross-sectional study Case series Case-control study
intra-rater reliability
Results are consistent with the same person rating the test more than once - Does a single rater score similar events comparably?
pretest-post test
Step 1 - randomize sample into 2 groups Step 2 - Both groups get the pretest (should perform the same) Step 3 - The experimental gets the intervention (control gets nothing) Step 4 - Both groups get the post test
Post Test only
Step 1. Randomize the sample into 2 groups Step 2. The experimental group gets the intervention (control gets nothing) Step 3. Both groups get the post test **NO BASELINE TEST to compare ***
temporal
Study participants moving away, changing their contact information, passing away, or otherwise being lost to follow-up are examples of what type of threat to internal validity? Measurement Sampling Temporal
Correlational (non-experimental) closer to 1 is closer to likely causation but does not prove cause closer to -1 is likely to preventative but does not prove
Tests relationship between 2 or more variables. Researcher is NOT testing if one variable causes another, or is different from another. **If one variable changes, does related change happen to the other? **Concerned about strength of relationship between variables.
predictive validity
The PACKRAT is a test administered by the PAEA, which is meant to predict how well the test-taker will do on the PANCE. How well it actually does this would be a demonstration of its: Content validity Concurrent validity Construct validity Predictive validity
Alternate Hypothesis (Ha or H1)
This type of prediction states if we do x, it will change the outcome compared to the control
Null Hypothesis (H0)
This type of prediction states if we do x, it will not change the outcome compared to the control
pretest post test
Two randomly-assigned groups of students take an anatomy quiz. One of the groups then takes a year-long course on gross human anatomy. (The other group does not take the course.) At the conclusion of the course, both groups take the same anatomy quiz again. Posttest-Only Pretest-Posttest Solomon Four Time Series
concurrent validity
You want to develop a new questionnaire to measure symptoms of depression. You conduct a pilot study in which participants are asked to fill out your questionnaire, as well as the PHQ-9 (a well-established and widely used depression questionnaire). You are testing your questionnaire for which type of validity? Content validity Concurrent validity Construct validity Predictive validity
Measurement
You want to perform a study that involves running complete blood counts on all of your participants. One day, your study assistant forgets to calibrate the CBC analyzer before running the day's samples. Your study now faces what threat to internal validity? Temporal Measurement Sampling
quantitative test
a cholesterol panel is what type of test hint: it's about levels/ amount
cronbach's alpha
a correlation-based statistic that measures a scale's internal reliability
Time Series
a forecasting technique that uses a series of past data points to make a forecast Step 1: Assign participants to the experimental and control groups Step 2: Take your initial observations (pretest). Step 3: Experimental group gets the intervention. Step 4: Multiple post tests are administered over time. Very similar to a longitudinal study. example: 2 groups of randomized diabetics both get baseline A1C group 1 gets diabetic consult group 2 gets nothing 4 months later check A1C THEN keep checking A1c every 4 months to see if intervention makes LASTING difference
cohort study
a group of people who share a defining characteristic are identified and followed over time, with measurements taken at regular intervals -- prospective and longitudinal
qualitative test
a pregnancy test OTC is what type of test "yes" or "no"
cross-sectional study
a study in which people of different ages are compared with one another - a snapshot of a population at the moment in time - cannot make any definite claims about how the exposure and outcome are linked
external validity
an attribute of an experiment in which variables have been defined in a normal, typical, or realistic way -can the results be generalized to the wider population "limitations"
random error
an error that occurs when the selected sample is an imperfect representation of the overall population
cohort why
can theoretically go on indefinitely and yield rich longitudinal data
content validity
consistency between the test items or problems and the kinds of situations or problems that occur on the job - How fairly do the items on a test represent what the test seeks to measure?
case series
describes a group of patients with something in common; often a rare or unusual condition - usually done over a short period of time strictly descriptive - no comparisons made to patients wo condition usefulness - to describe a new, rare, or unsual phenomenon, which can then be used as a basis for further research
validity
does the instrument measure what it is supposed to measure
systematic error
errors of measurement that are due to a characteristic of the examinee of the instrument
what are the 3 major classifications of research and how do they differ:
experimental (randomization [of either sample into groups or just random people from population] and control group and independent variable) quasi-experimental (may or may not have control group but no randomization) non-experimental (generally just observation. case study, case control, cohort, cross sectional. nothing is manipulated by researcher. no randomization)
Solomon Four
experimental design which minimizes effects of pretesting. Use this when you are afraid for the pre-testing effects. Step 1: Randomize! Into four groups this time. Step 2: Groups 1 and 2 get the pretest. Step 3: Groups 1 and 3 get the intervention. Step 4: All four groups get the posttest. helps to ensure the pre-test is not a confounding variable
coefficient of stability
extent to which scores are consistent over time - how well does the test hold up over time? Do the measurements stay true, or is there drift? **an easier way to think of this is "test-retest" reliability
reliability
is the instrument consitent
inter-rater reliability
measure of agreement among observers on how they record and classify a particular event - two or more raters observe the same event -- are they consistent with each other?
historical, ethnography, case study and other data is classified as what type of data historical- looking back at historical documents and accounts ethnography- live among the subjects and learn their ways then report findings case study- all participants have the condition being studied and we report about it and other-other
qualitative non-experimental study data comes in 4 types (describe)
Survey and Correlational data is classified as what type of data
quantitative non-experimental study data comes in 2 types:
case control
similar to case-series but now they have a "control group" cases: patients with a disease, condition, or outcome of interest controls: patients who are similar to the cases in terms of gender, age, geo, BUT DO NOT HAVE DISEASE usually retrospective: looking backwards and identifying potentially significant events with the help of hindsight
nominal
strictly descriptive; usually a category or descriptor cannot be ordered Examples: favorite color, undergrad major, place of birth. Usually qualitative data!
coefficient of equivalence
the degree of the relationship between various forms of a test - if the two separate tests are supposed to measure the same thing? Do they?
internal validity
the degree to which the effects observed in an experiment are due to the independent variable and not confounds - is the independent variable truly the only thing affecting the dependent variable (controlled for)
face validity
the extent to which a test item appears to fit the particular trait it is measuring - Does the measurement seem to be appropriate to an outside observer?
concurrent validity
the extent to which two measures of the same trait or ability agree - How well do the test results correlate with the results of an established test measuring the same thing?
construct validity
the extent to which variables measure what they are supposed to measure - How well does the test measure what it claims to be measuring?
randomized clinical trials
the participants are randomized into experimental and control groups - the experimental group gets the intervention, and the outcomes are then compared
rosenthal effect
the result when an experimenter's preconceived idea of appropriate responding influences the treatment of participants and their behavior
qualitative
think "descriptive" - involving words or descriptions You can usually find this with studies more likely to involve interviews, questionnaires etc
quantitative
think "quantity" - almost always involves numbers You can usually do math with this type of data. - experiments and studies are more likely to involve scientific/mathematical measurements
case control usefulness
this case type would be used to identify risk factors and/or exposures - what might have happened to the cases, that didn't happen to the controls?
temporal threat
this internal validity threat - history, maturation, and attrition/morality "longitudinal" - Are things likely to change as the study goes on over time
sampling threat
this internal validity threat has to do with regression to the mean and differential selection of subjects - how much is left up to chance?
measurement threat
this internal validity threat has to do with the testing and instrumentation - Are your instruments valid?
population related
this is a threat to external validity - is your sample truly representative of the wider population
co-efficient of internal consistency
this is if two items on the same test are supposed to measure the same thing, do they? - measures with Cronbach's Alpha
cross-sectional why
this is used to determine how prevalent a certain condition is in the population of interest (prevalence study) answers the question - what?
clinical trial with crossover
this type of study begins with a RCT, then continues after a "wash out period" - the groups are swapped and the intervention is repeated
meta analysis
this type of study uses existing study data from smaller but similar studies to re-analyze the data using stats to make it like one big study and report results. not the same as a lit review
experimental
what broad class of research has a control group, an experimental group, a dependent and independent variable and has randomization
operational definition
what do you call a statement of the procedures used to define research variables - important to define what you are studying
independent variable "Independant: I change"
what do you call the experimental factor that is manipulated; the variable whose effect is being studied. what is manipulated
dependent variable "depends on the independent variable"
what do you call the variable measured in an experimental study ie effect, outcome, or response in which the research is interested.
