Research Terminology

From FK Wiki
Jump to navigation Jump to search

Block B6 - Research


This page requires organization. For now, you can use your browser to search for words on this page. ⌘-F on Mac, CTRL-F on Windows, and menu>Find-in-page on Android.


Analytical study
designed to test a hypothesis. Usually deals with "why and how" questions. Often considers exposure and outcome. Uses control group. Used to help identify the cause of the disease. Analytical studies are a type of observational study.
Descriptive study
usually no hypothesis. Usually deals with "who, what, when, and where" questions, such as incidence and prevalence studies. Don't have control group. Descriptive studies are a type of observational study.
Association
(aka dependence) any statistical relationship, whether causal or not, between two random variables or two sets of data.
Correlation
when two variables have a linear relationship with each other, they're said to be correlated with each other. Correlation is not sufficient to demonstrate the presence of a causal relationship (i.e., correlation does not imply causation). Though since often causality exists between two correlated variables, correlation can give a hint at potential causality.
Causation
when changing one variable predictably causes another variable to change. This is also called the cause-effect relationship.
Accuracy
in the laboratory, accuracy of a test is determined when possible by comparing results from the test in question with results generated using reference standards or an established reference method.
Reliability
the reproducibility of an experimental result; the extent to which the same test or procedure will yield the same result either over time or with different observers.
Repeatability
variations in the repeat measurements on the same subject under the same conditions. e.g. the blood pressure was checked within 3 minutes using the same sphygmomanometer in the same location by the same doctor.
Reproducibility
variations in the repeat measurements on the same subject under different conditions. e.g. a different thermometer was used; the blood-pressure was checked in the left arm rather than the right arm and by a different doctor.
Validity
the degree to which conclusions about the relationship among variables based on the data are correct or ‘reasonable’. Validity involves ensuring the use of adequate sampling procedures, appropriate statistical tests, and reliable measurement procedures.


Confounding variable
a third variable that correlates with both the dependent and independent variable. In other words, confounding is interference by a third variable so as to distort the association being studied between two other variables. e.g. "age" might be a confounding factor when studying differences of strength between males and females.


Dichotomous
also known as "binary", dichotomous simply means two possible values such as "on" and "off". Dichotomous variables belong to the categorical types of variables.


Efficacy
efficacy answers the question, "does it work?" Efficacy is usually how well a treatment works in clinical trials or laboratory studies. e.g. a drug has good efficacy if it kills the cancer cells (like it was designed to).
Effectiveness
effectiveness answers the question, "does it benefit the patient?" For example, a drug may have good efficacy if it works well in vitro by killing the cancer cells, but poor effectiveness if it because it also kills the patient.
Efficiency
answers the question, "is it worth it?" For example, a drug may have good efficacy and good effectiveness but poor efficiency because it is too expensive to produce.


Experimental
exposure status is assigned to subjects by the researcher for the study. e.g. RCT.
Observational
exposure status is not assigned. e.g. cohort (population grouped by exposure status and followed to see if they develop outcome), case-control (cases have outcome, controls don't have outcome, and past exposure status is then determined), and cross-sectional (exposure and outcome determined at the same time).


Good Laboratory Practices
(GLP) are guidelines for non-clinical studies (in vitro, animal, and analysis of these). GLP includes guidelines for resources (organization, persons, facilities, equipment), characteristics (test items and test systems), rules (protocols, SOPs), results (raw data, final report), and QA (independent monitoring of research process).
Good Clinical Practices
(GCP) are guidelines for clinical studies (i.e. studies involving humans). GCP includes guidelines about rights, safety, and well-being.
Good Manufacturing Practices
(GMP)


In vitro
In a sample taken from a person or animal, such as bone marrow extracted and then placed inside a test-tube. "In vitro" comes from Latin where "vitro" means "glass".
In vivo
In a living person or animal. For example, when a living person is injected with a drug to test the effects of the medication. "Vivo" means "living" in Latin.


Validity
how well a test measures what it is purported to measure.
External validity
the extent to which study findings can be generalized beyond the sample used in the study.
Internal validity
the extent to which the effects detected in a study are truly caused by the treatment or exposure in the study sample, rather than being due to other biasing effects of extraneous variables.


Incidence
the number of new cases in a particular period. Incidence is often expressed as a ratio, in which the number of cases is the numerator and the population at risk is the denominator.
Prevalence
the number of all new and old cases of a disease or occurrences of an event during a particular period. Prevalence is expressed as a ratio in which the number of events is the numerator and the population at risk is the denominator. The total number of cases of a disease in a given population at a specific time.


Intra-observer error
the differences between interpretations of an individual making observations of the same phenomenon at different times.
Inter-observer error
the differences between interpretations of two or more individuals making observations of the same phenomenon.


Positive predictive value
Negative predictive value
Normal distribution
Normal distribution
when data is distributed into a symmetrical bell curve.


Null-hypothesis (H0)
The default hypothesis. The hypothesis that is widely accepted as true and which the researcher is trying to disprove. For example, in a clinical trial of a new drug, the null hypothesis might be that the new drug is no better than the current drug.
Alternative-hypothesis (H1)
The hypothesis that is being presented as an alternative to the null-hypothesis. The researcher will try to prove that this is true as well as prove that the null-hypothesis is false. For example, the new drug is better than the old drug.
Type-I Error
When both the null-hypothesis and the alternative-hypothesis are rejected.
Type-II Error
When both the null-hypothesis and the alternative-hypothesis are accepted as true.
Power of a test
probability of a test correctly rejecting the null-hypothesis.


Relative Risk
(RR) risk of disease in exposed group divided by risk of disease in non-exposed group. Used in cohort studies.
Odds ratio
(OR) a measure of association between an exposure and an outcome. Used in case-control studies because

Both RR and OR are interpreted as follows:

=1 no association.
>1 positive association.
<1 negative association.

Some examples:

  • RR = 5 means that people that were exposed are 5 times more likely to have an outcome when compared to those that were not not exposed.
  • RR = 0.5 means that people were exposed are only half as likely to have an outcome when compared to those that were not not exposed.
  • RR = 1 means that people were exposed have the same likelihood to have an outcome when compared to those that were not not exposed.


Random error
errors in experimental measurements that are caused by unknown and unpredictable changes in the experiment. Can be minimized by using appropriate statistical tests.
Systematic error
errors in experimental observations that usually come from the measuring instruments. Can be minimized by using appropriate research methods.


Multivariate statistical analysis:Researchers use multivariate procedures in studies that involve more than one dependent variable (also known as the outcome or phenomenon of interest), more than one independent variable (also known as a predictor) or both.

Tests of significance
p-value and confidence interval indicate the reliability of an association that was observed. In other words, they help identify if an association is truly an association or if it was only observed due to chance.
p-value
a measure of how likely an observed association be if it was purely due to chance in the absence of an exposure. That means, the smaller the p-value, the higher the chance that the observed association is a true association and not due to chance. A p-value of 0.05 indicates only a 5% chance that the RR or OR was observed by chance alone.
95% Confidence Interval
(CI) the range of values of a measure of association (using either OR or RR) that has a 95% chance of containing the true RR or OR. In other words, 1.2-3.5 95% CI means we are 95% sure that the true value falls within the range of 1.2 to 3.5. If the range of values contains 1.0, then we can say that there is no significant association between exposure and outcome. A 95% CI below 1 indicates less risk of outcome in exposed population. A 95% CI above 1 indicates a higher risk.


Variance
measures how far a set of numbers is spread out. A variance of zero indicates that all the values are identical. A small variance indicates that the data points tend to be very close to the mean, while a high variance means that the data points are very spread out from the mean.


Variate
for example, if you stand on a weigh scale, the variable is body weight, and the variate is whatever the measured weight is.