Chapter 31 - 33
approach uses data and prior beliefs (the priors) to predict the likelihood of a particular outcome (the posterior)
Bayesian
set of conditions that provide support for the existence of a causal relationship between an exposure and an outcome
Bradford Hill criteria
assesses the homogeneity of stratum specific measures of association
Breslow Day test
type of regression model that estimates a hazard ratio
Cox proportional hazards
time graph that displays cumulative survival rates in a study population
Kaplan Meier plot
weighting method can adjust odds ratio or other measures of association for an exposure variable and an outcome variable after using a third variable to stratify the data
Mantel Haenszel
methods are stochastic processes that use algorithms to take samples from stimulated probability distributions
Markov chain Mote Carlo (MCMC)
tests for spatial autocorrelation, which is a measurement of how similar one location is to nearby places
Moran's I coefficient
compartmental mathematical model of infection transmission that describes how the susceptible (S) individuals in a population may become infected (I) and then eventually recover (R) with immunity
SIR model
developed before the start of data analysis
a priori codes
makes inferences based on limited observation and minor premises, so the conclusions are assumed to be best guesses that are merely probable
abduction
uses computers to stimulate the actions and interaction of various individuals (agents) in a population
agent based modeling/individual based modeling
relationship between two variables
association
pattern in which a variable measured over times has values influenced by its own past values as per a Durbin Watson test or another test statistic or, in spatial analysis, a measurement of how similar one location is to nearby places
autocorrelation
identifies one core category or core phenomenon and several related categories that express the major and minor themes of the analysis
axial coding
deletes predictor variables from the model until deleting a variable significantly reduces the overall fit of the model
backward stepwise method
describes the adverse impact of a particular health condition on a population
burden of disease
identifies the set of ratio/interval and/or nominal variables that most accurately predicts group membership in a model with two ratio/interval and/or nominal dependent variables
canonical analysis
group of related codes
category
exposure that has been scientifically tested and shown to occur before the disease outcome and to contribute directly to its occurrence
causal factor
relationship in which an exposure directly causes an outcome
causation
identifies groups of similar observations using an algorithm that seeks to minimize the variations among observations within each group
cluster analysis
label attached to a word or phrase
code
use of words or short phrases to briefly summarize the contents, attitudes, processes, or other aspects of each item in a transcript or other qualitative document
coding/indexing
mathematical model in which each "individual" in the stimulated population exists in only one of several states at one time, but over time these individuals can move between states
compartmental model
probability of an event occurring given that some prior event has already occurred
conditional probability
indicator of neutrality that is present when the results of a study are shown not to be due to researcher bias
confirmability
third variable that is associated with both the exposure variable and the outcome variable and distorts the apparent relationship between the exposure and outcome
confounder
process in which qualitative data are collected and analyzed simultaneously, rather than waiting to begin analysis after all data have been gathered
constant comparison
process of categorizing textual data
content analysis
economic analysis that compares the health gains from an intervention to the financial costs of that intervention
cost effectiveness analysis (CEA)
present when the interpretation of the data accurately reflects the studied groups or texts
credibility
probability of an event occurring by the end of a particular observation period
cumulative probability
refers to raw or unprocessed facts, figures, symbols, or signs
data
interdisciplinary field that uses statistics, machine learning, and other types of computational tools to generate information and knowledge from various types of data
data science
makes logical inferences based on facts or widely accepted premises, and the conclusions are assumed to be certain
deduction
indicator of consistency that is demonstrated through transparency about data collection, analysis, and interpretation methods
dependability
variable in a statistical model that represents the output or outcome for which the variation is being studied
dependent/outcome variable
one in which the outcomes of the model are the same every time the model is run with the same inputs
deterministic model
burden of disease metric that is quantified as the sum of YLDs and YLLs in a population
disability adjusted life year (DALY)
uses the tools of linguistics to evaluate the ordinary use of written and spoken language
discourse analysis
statistical method that identifies the set of ratio/interval and/or nominal variables that most accurately predicts group membership in a model with a nominal dependent variable
discriminant analysis/discriminant function analysis
ability of a model to distinguish between independent groups
discrimination
derived variables created by recoding one variable with n categorical responses into a set of n - 1 dichotomous (0/1) variables
dummy variables
third variable that defines groups of individuals who experience different biological responses to various exposures
effect modifier
represent the proportion of variance accounted for by the correlation between each pair of canonical variates
eigenvalues
concepts that are identified during the early stages of qualitative analysis and assigned a label or code that describes them
emergent codes
cause of a disease or other health disorder
etiology
third variable that produces an apparent but false association between two other variables that are not causally related
extraneous variable
seeks to identify the most frequent and important categories
focused coding
instructs the computer to add the best predictor variables to the model one at a time until adding an additional variable does not significantly improve the overall fit of the model
forward stepwise method
approach to probability that is based on the expected frequency of an event occurring over a long time period or if an occurring over a long time period or if an experiment is repeated many times
frequentist
examines how well real data match the values predicted by a model
goodness of fit test
equation describing the conditional probability of an individual having an event at a particular time given that the person has survived to that time
hazard function
compares durations of time to an event in two populations
hazard ratio
study of the interpretation of texts
hermeneutics
heterogeneity of variance among the variables in a linear regression model that is demonstrated when the distribution of residuals from a regression model across the length of the best fit line is uneven
heteroscedasticity
multivariable regression model that adjusts for different levels of exposure, such as for both census tract and county
hierarchical model/multilevel model
homogeneity of variance among the variables in a linear regression model that is demonstrated by the even distribution of residuals from a regression model across the length of the best fit line
homoscedasticity
variable in a statistical model that predicts the value of some outcome variable
independent/predictor variable
variable used to measure performance, achievement, or change
indicator
makes inferences based on observations, and the conclusions are assumed to be likely
induction
refers to data that have been processed and presented in a format usable for understanding a situation and making decisions
information
occurs when the effect of one predictor variable on an outcome variable depends on the presence or absence of a second predictor variable
interaction
describes a repetitive process
iteration
occurs when a screening test that enables early detection of an adverse health condition is incorrectly interpreted as prolonging survival with the condition
lead time bias
actuarial table that displays conditional and cumulative survival probability in a population
life table
model is used when the outcome variable is a ratio or interval variable
linear regression
statistical test that determines whether survival rates are longer in one population than another
log rank test
visualization of the hypothesized causal pathways that lead to an outcome of interest
logic model
model is a probability based regression model used when the outcome variable is binomial
logistic regression
model that uses a logit link function of ln(p/1-p), also written as logit(p), as their outcome
logit regression
third variable that was not measured in a study but is affecting the apparent association between an exposure variable and an outcome variable
lurking variable
method of data analysis derived from artificial intelligence
machine learning
value of a coefficient in a logistic regression model that gives the model the greatest probability of matching the observed data
maximum likelihood estimate (MLE)
documenting personal reflections an impressions about observations, participants, experiences, codes, categories, and themes
memoing
composite indicator derived from two or more other measures
metric
causal pathway in which many different risk factors or combinations of risk factors contribute to a disease occurring
multicausality
problem that occurs when two or more predictor variables in a multiple regression model are highly correlated, and that redundancy makes the coefficients for one or more of those variables highly inaccurate
multicollinearity
model examines the relationships between several ratio/interval and/or nominal predictor variables and one ratio/interval outcome variable when there is a linear relationship between the independent and dependent variables
multiple linear regression model
examines the relationships between several ratio/interval and/or nominal predictor variables and the value of one nominal outcome variable
multiple logistic regression
qualitative analysis method that seeks to understand personal stories
narrative analysis
machine learning algorithm that is used in the analysis of qualitative and social media data to examine how people speak and write in real life situations
natural language processing
assumes that causal pathways can be bidirectional
nonrecursive model
equation that includes one or more functions of one independent variable along with the derivatives of those functions
ordinary differential equation (ODE)
linear regression modeling approach that finds the line that minimizes the average vertical distance from each point in a data set to the fitted line
ordinary least squares (OLS)
recursive causal analysis strategy that uses regression models to examine causal patterns among variables
path analysis
seeks to group codes into a limited number of categories
pattern coding
predicts the probability group membership while adjusting for covariates
propensity score matching
metric used in health economics to estimate the additional duration of life and quality of life conferred to populations by successful public health interventions
quality adjusted life year (QALY)
construct that captures an individual's perceived position in life in the context of that person's expectations, goals, values, and concerns
quality of life (QOL)
assumes that all causal pathways are undirection
recursive model
statistical model that seeks to understand the relationship between one or more independent variables and one dependent variable
regression model
the difference between the observed value in a data set and the value predicted by a regression model
residual
study of sign and symbols
semiotics
the process of examining the robustness of statistical methods and the results of models
sensitivity analysis
examines whether there is a linear relationship between one ratio or interval predictor variable and one ratio or interval outcome variable
simple linear regression model
process of compiling and analyzing data from social networking services like instagram and twitter
social media analytics
describes results that are false or invalid, such as the spurious associations that are generated by extraneous variables
spurious
the inputs vary according to a probability distribution, so the outcomes differ slightly every time the model is run
stochastic model
nonrecursive causal analysis strategy that can be used to examine complexities in the directionalities of the path diagram
structural equation modeling
statistical evaluation of the distribution of the durations of time that individuals in a study population experience from an initial time point until some well defined event, such as death, discharge from a hospital, or some other outcome
survival analysis
describes the timing of events
temporality
concept that encompasses one or several categories
theme
construct that provides a systematic explanation about phenomenon
theory
variable that is associated with an exposure variable and an outcome variable but is not part of the causal pathway from an exposure to an outcome
third variable
inverse of the variance inflation factor (1/VIF), should not be too close to 0
tolerance
present when the interpretation of qualitative data is likely to be applicable in other circumstances
transferability
should be small enough (typically < 10) to show that the independent variables in a regression model have reasonably independent errors and are not intercorrelated (values cannot be lower than 1)
variance inflation factor (VIF)
burden of disease metric used to quantify the population level reduction in health status attributable to nonfatal conditions
years lived with disability (YLDs)
burden of disease metric used to quantify the population level reduction in health status due to premature mortality
years of life lost (YLLs)
