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;
73 Cards in this Set
- Front
- Back
BELL-SHAPED CURVE
|
A symmetrical, numerical curve representing the normal distribution of a population. Biological processes tend to have bell-shaped curves.
|
|
BIAS
|
Term used to describe a tendency or preference towards a particular perspective, ideology, or result, especially when the tendency interferes with the ability to be impartial, unprejudiced, or objective.
|
|
BLIND
|
A study in which the experimenter is unaware of which group is subject to which procedure.
|
|
SIX TYPES OF BIAS
|
Non-response/non-selection
chronology susceptibility compliance skill/performance observation |
|
CASE STUDY
|
Qualitative descriptive research that is used to look at individuals, a small group of participants, or a group as a whole. Researchers collect data about participants using participant and direct observations, interviews, protocols, tests, examinations of records, and collections of writing samples.
|
|
CLINICAL STUDY
|
A scientific study of how a new medicine or treatment works in a population. Through clinical studies, the investigator may determine alternative and potentially improved methods to prevent, detect, diagnose, control, and treat disease.
|
|
COHORT
|
An observational study with a prospective design that is longitudinal, controlled, and uses no manipulation. This type of study is less susceptible to biases and demonstrates better control of quality of collected data than case-control, case-series, and case reports; but more susceptible to bias than randomized controlled studies.
|
|
CONFIDENCE INTERVALS
|
A range of values for a variable of interest, constructed so that this range has a specified probability of including the true value of the variable.
|
|
DOUBLE-BLIND
|
Investigator and administrator are blind to treatment.
|
|
EXPERIMENT
|
A method of investigating casual relationships among variables or to test a hypothesis. Experiments can be used to help solve practical problems and to support or negate theoretical assumptions. Experiments are usually performed under controlled conditions.
|
|
GENERALIZABILITY
|
The ability to take study conclusions from sample populations and apply them across a population at large.
|
|
MEAN
|
Quantitative average of a given data set; sum of values divided by the number of observations
Easily influenced by extreme values |
|
MEDIAN
|
Middle of a given data set; the number of observations above equals the number below (i.e. the middle number)
Not easily influenced by extreme values |
|
NORMAL DISTRIBUTION
|
Quantitative data that is distributed in a way that symmetrically clusters around the mean (bell shaped curve). Normally-distributed data can be described by the mean and standard deviation.
|
|
NULL HYPOTHESIS
|
The assumption in inferential statistics that there is no difference between the observations in two or more treatments.
Rejected if p < 0.05 Upheld if p > 0.05 |
|
OBSERVATIONAL STUDY
|
Conducting a study in which individuals are observed and outcomes are measured and the results are independent of the researchers (i.e. researchers do not allocate individuals to treatment and do not give (or induce) a treatment)
|
|
PLACEBO
|
The treatment given to the control group in an experiment that is as similar as possible to the actual treatment without containing the intervention of interest
* the preferred method to provide masking/blinding in experimental studies |
|
PROBABILITY
|
The chance that something will happen
|
|
PROSPECTIVE STUDY
|
Following the patient forward over time whereas retrospectively is through [written] records looking back over time
|
|
RANDOMIZATION
|
A method of sampling in which every time an item is drawn from the population, every item in the population had an equal chance of being selected.
|
|
RESEARCH HYPOTHESIS
|
Specific testable predictions made about the independent and dependent variables in the study
|
|
RETROSPECTIVE STUDY
|
Observations made from records
|
|
STANDARD DEVIATION
|
Average difference of individual values from the means.
It is the square root of the variance. |
|
STATISTICALLY SIGNIFICANT EFFECT
|
the probability of obtaining this effect by chance is very low.
A p value of 0.05 or less. |
|
TREATMENT EFFECT
|
The observed difference in the treated animals from the controls (which you have to prove could not have been due to chance).
|
|
VALIDITY
|
The strength of our conclusions- how much one is confident that the results were obtained without bias and can be applied to the clinical population in question
|
|
INTERNAL VALIDITY
|
Study results represent a testable relationship and are verifiable within the study parameters.
Methods to reduce the risk of bias such as randomization and blinding improve internal validity |
|
EXTERNAL VALIDITY
|
Results can be extrapolated to the real world due to similarity of subjects, housing, and exposure to disease agents between the experimental settings and real-life settings.
|
|
CONSTRUCT VALIDITY
|
The outcome assessment is a theoretical concept such as health, anxiety (these concepts must be closely defined)
|
|
VARIANCE
|
How much the data varies from the mean
Large = lots of variance among the data Small = not a lot of variance among the data |
|
STANDARD DEVIATION IN A NORMAL DISTRIBUTION
|
1 SD encompasses 68% of the observations
2 SDs encompass 95% of the observations 3 SDs encompass 99% of the observations |
|
STATISTICAL TESTS
|
Act to test the mathematically probability that your results happened by chance. (no treatment effect)
|
|
THREE TYPES OF OBSERVATIONAL STUDIES
|
Cohort
Case Control Case Series |
|
EXPERIMENTAL DESIGN
|
Primarily concerned with the allocation of subjects to treatments and with what we can conclude from them.
Important to consider : selection/allocation of subjects and experimental procedures |
|
EXPERIMENTAL HYPOTHESIS
|
H(a)
Investigational objectives |
|
CONTROLS
|
A standard against which other conditions can be compared in a scientific experiment
|
|
PLACEBO
|
Usually saline treatment
Is preferred to no treatment |
|
ALTERNATIVE TREATMENTS
|
Standard of care: treatment that experts agree is appropriate, accepted and widely used
|
|
CROSS SECTIONAL MONITORING
|
Observations are made at one time
|
|
LONGITUDINAL MONITORING
|
Observations follow a patient over time either prospectively or retrospectively with records
|
|
CONFOUNDED EFFECTS
|
Effects of treatment are interrelated with the effects of something else
|
|
IT IS BELIEVED THAT...
|
= "I think..."
|
|
IT IS GENERALLY BELIEVED THAT...
|
= "I and a few others think..."
|
|
IT HAS LONG BEEN KNOWN THAT...
|
= "I couldn't find a reference..."
|
|
WHILE IT HAS NOT BEEN POSSIBLE TO PROVIDE DEFINITE ANSWERS TO THE QUESTIONS...
|
= "The experiment didn't work out, but I figured that I could get a paper out of it anyway."
|
|
THANKS ARE DUE TO JOHN DOE FOR ASSISTANCE WITH THE EXPERIMENT AND TO JOANNA DOE FOR VALUABLE DISCUSSION...
|
= "John did the work and Joanna explained what it meant."
|
|
CONSTRUCT VALIDITY
|
Outcome is a theoretical construct
- "well-being", "success" and "return to function" |
|
NON-RESPONSE BIAS
|
A systemic difference between results of subjects who stayed in and who dropped out
|
|
NON-SELECTION BIAS
|
The "frame" doesn't fit -- the population selected doesn't fit the greater population
|
|
CHRONOLOGY BIAS
|
Results are compared from two populations treated at two different times
|
|
SUSCEPTIBILITY BIAS
|
Results are compared for two prognostically different groups.
Often occurs from non-random grouping |
|
COMPLIANCE BIAS
|
Test groups are not treated identically
|
|
SKILL/PERFORMANCE BIAS
|
Dissimilar levels of skill are used in technical procedures
ex. Senior Clinician vs. Senior Student |
|
OBSERVATION BIAS
|
Outcomes are not measured in a comparable fashion
Ex. subjective evaluation, poor equipment |
|
HAWTHORNE EFFECT
|
Knowledge of being observed changes behavior
|
|
HOW TO RANDOMIZE
|
Randomize, check predictable characteristics, re-randomize if needed.
|
|
RANDOM SUBJECT ASSIGNMENT
|
All individuals have an equal chance of being selected
Knowing something about one tells us nothing about another NOT the same as haphazard selection/systemic selection |
|
MINIMIZE NONRESPONSE/NONSELECTION BIAS
|
Improve response rate rather than increase study size
Careful selection of frame Measure non-respondants |
|
DESCRIPTIVE STUDIES
|
Animals serve as their own controls
Measurements not compared to other groups Often found in surgical papers NO COMPARISON OF TX NOT ALWAYS CAUSE/EFFECT RELATIONSHIP |
|
CROSS-SECTIONAL STUDIES
|
Sampling and classification of a group of animals according to disease status and analysis carried out to identify relationships between cause and disease
A fishing expedition All observations made at one point in time Cannot infer causality Need to make sure sample is representative of population and is truly random |
|
OBSERVATIONAL CASE-CONTROLLED DESIGN
|
A study comparing a group of animals that have or have had the disease being studied and a similar group of animals that do not have the disease.
Groups are compared in light of how they differ in numbers of animals have hypothesized "cause". No experimenter manipulation, but semi-controlled. Used to investigate rare diseases Low in cost - makes use of preexisting data Highly susceptible to bias No random allocation No control of sources Lots of bias is possible |
|
COHORT OBSERVATIONAL STUDY
|
A study comparing 2 groups of animals
1 with the hypothesized cause, but free of disease of interest 1 group of healthy control animals w/o cause These two groups are as identical As possible when considering weight, sex, age, breed, etc. Less susceptible to biases Better control over quality of collected data Large time periods required for future assessment Some problems may arise with: cohort selection methods and determining what is "disease-free" Are the outcome assessments blinded? Length of outcome Loss-to-followup Bias can be eliminated with careful selection of cohorts |
|
NON RANDOMIZED GROUPS
|
Differently treated groups, but groups may not be uniform
Very susceptible to selection bias Conclusion on as internally valid as groups were similar |
|
RANDOM GROUPS
|
Treatment groups are randomized to assure uniformity
Important to evaluate internal validity Important observations can be tested at start to establish equivalence of pre-treatment groups |
|
RANDOMIZED BLOCKS
|
Grouping of subjects with similar characteristics into treatment subgroups
The entire population is blocked or sorted into categories of interest Then equal numbers from each category are randomly assigned to the various treatment and/or control groups Allows examination of subpopulations of subjects and assures internal validity of the groups for that purpose |
|
CROSS-OVER DESIGN
|
Usually a drug study or tx w/ relapse of signs. Repeated measures analysis needed.
Subject is sequentially exposed to each of study tx, usually w/ a washout period between Ideally, some tx may be repeats Subjects become their own controls. Decreased animal purchase costs Have to be wary of carryover effects |
|
LATIN SQUARE
|
An n by n array containing symbols from some alphabet of size n, arranged so that each symbol appears exactly once in each row and exactly once in each column
Advantages: decreases experimental time and subject number compared to classic crossover Disadvantages: Potential increase in drug interaction crossovers |
|
DISTRIBUTION OF RESEARCH DESIGN
|
7% true experimental
93% clinical epidemiology |
|
QUESTIONS TO BE ANSWERED IN RESEARCH
|
44% cause
24% therapy 8% frequency 7% diagnosis 2% prognosis |
|
STATISTICS
|
Used when making generalizations about the whole population after measuring a sample.
Mathematical comparison of the outcome obtained in an experiment to outcomes that could occur "by chance" Probability Needed to understand the physical world |
|
MODE
|
Most frequently occurring value
One way of describing "center" of data with no numerical meaning |
|
RANGE
|
From lowest to highest value
Greatly influenced by extreme values |
|
PERCENTILE
|
The proportion of all observations falling between specific values
|