Term
|
Definition
| A condition, intervention, or characteristic that will predict or cause a given outcome |
|
|
Term
|
Definition
| The outcome variable which is either caused by or related to the independent variable |
|
|
Term
|
Definition
| Conceptualizing a variable in a form that explicitly states how it is going to be measure quantitatively |
|
|
Term
|
Definition
-Labels, names, identifies
-Mutually exclusive and exhaustive categories
-No arithmetic manipulation other than counting can be performed
Examples: 1=male 0=female, numbers for football players, blood types
Weakest
Appropriate stat test: Frequency/chi square |
|
|
Term
|
Definition
-Rank order points on a scale (directionally)
-Numeric values are limited
-Intervals between items are not known, and are not equal
Ex: Educational attainment (0=less than HS, 1=some HS, 2=HS degree, 3=some college, 4=college degree, 5=post college)
Appropriate stat test: frequency/non-parametric tests |
|
|
Term
|
Definition
-Has numeric properties
-Assume equal amounts between the numbers on the scale- equal intervals
-No true 'zero' amount
Example: Temperature on the Farenheit scale, standard numbering of calendar years
Appropriate stat test: parametric/non-parametric tests |
|
|
Term
|
Definition
-Has all the properties of an interval scale
-In addition, it has a true zero point indicating complete absence of the variable
-Can form ratios: 10 lbs is twice as heavy as 5 lbs
Examples: Weight, temp on Kelvin scale, number of pts visiting a pharmacy
Appropriate stat test: parametric/non-parametric tests |
|
|
Term
|
Definition
-Numeric values assigned to a variable consist of limited categories
-No way to expand your answer
-Yes or no questions
-Better, worse, the same |
|
|
Term
|
Definition
-A larger range of numeric values are assigned to a variable
Continuous data can be converted to categorical data but not vice versa |
|
|
Term
|
Definition
Precision (tight grouping)/reproducibility
-Extent to which a repeated measurement of a stable phenomenon by different people and instruments at different times and places produces similar results
-Range from 0.00-1.00
-Closer to 1.00 = most reliable, closer to 0.00 = most unreliable |
|
|
Term
|
Definition
Accuracy
-Degree to which the data measure correspond to the true state of the phenomenon being measured (does the instrument measure what it is supposed to measure?) |
|
|
Term
|
Definition
-Measure of the strength and direction of a relationship between two variables
-Range from -1 to 1
-Farther from 0 = stronger the correlation
-0= random event, no correlation |
|
|
Term
|
Definition
-Determine reliability by testing & retesting (blood glucose meter).
-Use two instruments that assess the same concept
-Rater reliability: assessed when data is collected for a study by an observer or tester/rater (Used with objective assessments)
-Intra-rater reliability: stability of data recorded by one indicividual
-Inter-rater reliability: variability between two or more raters who measure same group of subjects
-Internal consistency: used with instrument/surveys that use multiple questions/items to measure a single construct |
|
|
Term
|
Definition
-Face validity: instrument appears to test what it is supposed to test (qualitative). Weak
-Content validity: Method of measurement includes all dimensions of the construct one intends to measure (and nothing more)
-Construct validity: Degree to which inferences can legitimately be made from the instruments or measures to theoretical constructs that the instrument was supposed to measure |
|
|
Term
|
Definition
-Highly correlated with an already validated scale (gold standard)
2 types:
-concurrent validity: instrument to be validated and gold standard are administered concurrently
-predictive validity: relationship btwn the new instrument & the outcomes or gold standard is examined to determine if the new instrument is a good predictor of the outcome/gold standard |
|
|
Term
|
Definition
-Convergent validity: high positive correlation between scores on measurement and another measure reflecting underlying phenomenon
-Discriminate validity: indicates a low correlation will exist between measure that assess different characteristics |
|
|
Term
| Major concept of validity (accuracy) |
|
Definition
-Sensitivity: the proportion of people with disease who have positive test
-Specificity: proportion of people without disease who have a negative test |
|
|
Term
| Positive predictive value |
|
Definition
-Estimates the likelihood that a person who tests positive actually has the disease
-PPV= a/a+b |
|
|
Term
| Negative predictive value |
|
Definition
| -Estimates the likelihood that a person who tests negative is actually disease free |
|
|
Term
|
Definition
Manipulation of the variable?
-Yes: Experimental design
-No: Observational
Is the comparison group assigned randomly?
-Yes: True experimental design
-No: Quasi-experimental design |
|
|
Term
|
Definition
| -Is the independent variable really associated with the dependent variable? |
|
|
Term
|
Definition
| -Are the results of our study generalizable? Can our conclusions be applied to the population of interest? |
|
|
Term
| Threats to internal validity |
|
Definition
| History, maturation, attrition, testing, instrumentation, statistical regression, selection bias, diffusion of treatment, compensatory equalization of treatments |
|
|
Term
| Threats to external validity |
|
Definition
-Interaction of treatment and selection
-Interaction of treatment and setting
-Interaction of treatment and history |
|
|
Term
| Limiting threats to validity |
|
Definition
-Can be controlled or reduced through appropriate study design, appropriate sample selection, and use of specific inclusion/exclusion criteria
-Some threats can be further controlled via statistical methods
-Studies that rely on sound research design to minimize threats to validity are considered more robust |
|
|
Term
|
Definition
-Subjects are randomly assigned to at least two comparison groups
-Separated by how subjects are assigned to groups |
|
|
Term
| Quasi-experimental design |
|
Definition
-Does not meet requirements of true exp design
-Lacks either randomization or comparison groups (or both)
-Does have a control measure or multiple measuresn (if not, it is a "non-experiment") |
|
|
Term
|
Definition
O1 X O2
-Observation of group, intervention, then another observation
-No randomization
Problems: lack of a control group, vulnerable to threats of internal validity |
|
|
Term
| Time-series design ("interrupted") |
|
Definition
O1O2O3 X O4O5O6
-Measures effects of variables over time
-Based on application of multiple measurements before and after treatment
Advantages: The multiple pretests and posttests act as pseudo-control conditions because they would help us to recognize if any of the common threats to internal validity were confounding the study
Disadvantage: the greatest threat to the internal validity of these studies is history |
|
|
Term
| Non-equivalent Pretest-Posttest control group design |
|
Definition
O1 X O2
-------
O1 O2
-Strongest of all quasi-experimental design as it contains a comparison [control] group (assignment is not random) |
|
|
Term
|
Definition
-Open (non-blinded): Both the investigator and the subjects know about the treatment/placebo assignment
-Single blinded: only one of them know about the assignment
-Double blinded: both the investigator and the subject have no knowledge of the assignment |
|
|
Term
| Repeated measures design (Cross-over design) |
|
Definition
-One group of subjects is exposed to all levels of a treatment variable
-Washout period between the exposures/interventions |
|
|
Term
|
Definition
-Do not involve manipulation of the independent variable
-The independent variable is not an active, but an attribute variable
-An observational study may or may not have a control group
-Assignment to treatment/exposure cannot be random |
|
|
Term
|
Definition
| -The study of the distribution of factors that affect health within population groups |
|
|
Term
|
Definition
| -involves the examination of data that has been collected in the past |
|
|
Term
|
Definition
-Involves measurement and data collection starting in the present and going into the future
-More reliable because the researcher has control of data collection
-Major disadvantage is time and expense associated with long periods of data collection |
|
|
Term
|
Definition
-Conducted at a single point in time and they provide a snapshot of the situation
-Used in prevalence studies
-Limitation: temporality problems, can't show that the cause came before the effect |
|
|
Term
|
Definition
-Observe subjects over time
-Researchers make observations over time
-Aids in the distinction between independent and dependent variable |
|
|
Term
|
Definition
-Classifies the occurrence of disease according to person, place and time
-Aid in generating hypotheses that can be explored by analytic and epidemiologic studies |
|
|
Term
|
Definition
-Describe the experience of a patient regarding their disease condition or response to treatment
-Low internal & external validity
-Help in generation of inductive hypothesis
-No control group |
|
|
Term
|
Definition
-A proportion reflecting the number of existing cases of a disorder relative to the total population at a given point in time
-(# of existing cases)/(total population) |
|
|
Term
|
Definition
-Quantifies the new cases in the population during a specified time period
-Represents an estimate of the risk of developing the disease during that time
-(# of new cases)/(total population at risk) |
|
|
Term
|
Definition
-indicates the likelihood that someone who has been exposed to a risk factor will develop the disease, as compared with one who has not been exposed
-(incidence among exposed)/(incidence among non-exposed) |
|
|
Term
|
Definition
-How much disease that occurs that can be attributed to a certain exposure
-The amount or proportion of disease incidence that can be attributed to a specific exposure
-(incidence among exposed)-(incidence among non-exposed) |
|
|
Term
| Number needed to treat (NNT) |
|
Definition
| -The number of people we must treat in order to prevent one adverse event or produce one benefit |
|
|
Term
| Number needed to harm (NNH) |
|
Definition
-the number of people that must be exposed to a risk to cause one adverse event
-Use to examine the effect of a harmful exposure |
|
|
