BAPS unit 2

-patterns in time: does a peak in explanatory variable occur in concert w peak in response variable, with appropriate time separation?
-prevalance: is a rare exposure followed by rare effect?
-context: is there a plausible connection between the two?

proving causation when experiment is unethical

case report (observational)
-new issues and ideas are introduced, can be beginning of better studies

a detailed description of the management of a patient/client that may serve as a basis for future research

ecological study (observational)
-subjects are groups of people (ex households, nations)
-compares health of populations in diff places or times
-can't say anything about individual level

a study in which the units of analysis are populations or groups of people rather than individuals

cross-sectional study (observational)
-researchers look at group of people that fit their description of subjects they're interested in
-1 time measurement of exposure and outcome

A study in which a representative cross section of the population is tested or surveyed at one specific time

Case-control study (observational, retrospective)
-compares people w disease (cases) w similar group of healthy people (controls)
-ask subjects to remember habits to see which subjects in each group had exposure to factor of interest, compare frequency of

study where a group of individuals with the diseases, referred to as cases, are compared to individuals without the disease, referred to as controls

cohort study (observational, prospective)
-begins w people who have been exposed to factor of interest
-subjects are followed and evaluated to see if develop condition of interest

study that measures variables of a group of people over time

randomized controlled trial (experiment)
-subjects randomly assigned to treatment to see effects on response variable
-other variables are controlled or blocked
-can demonstrate cause and effect

An experimental study in which researchers randomly assign individuals to either an experimental or a control group and expose the experimental group to the manipulated variable of interest

systematic review
-identifies all relevant studies and synthesizes all of the conclusions

summarized findings from multiple studies of a specific clinical practice question or topic that recommend practice changes and future directions for research; one of the strongest sources of evidence for evidence-based practice

-combines data from multiple similarly designed studies to perform singular statistical anlaysis
-increases power of the test, but can be difficult to find studies w similar designs

a procedure for statistically combining the results of many different research studies

Yi ~ iid N(�, ?)
-all Y are independently and normally distributed w mean � and SD ?
-Yi = response variable for subject i
-iid = independent and identically distributed (identically distributed- the distribution doesn't fluctuate and all items in the sam

general model for response variable

Yi = � + ?i, where ?i ~ iid N(0, ?)
-decompose response variable into overall mean and variability about the mean (true mean + some error term --> response variable for subject i varies around the true mean)
-i = index counting for subjects from i=1 to i=

improved model w Yi


the thing we want to detect


sources of variability that are unaccounted for

Yi,j = � + ? j + ? i,j, where ? i,j ~ iid N(0, ?)
-taking error term, saying some of it isn't due to random variability, but can be due to something we can account for?, here pulled treatment effect out of error term
-? would be smaller than before (varia

Yi,j model

Yi,j,k = � + ? j + ? k + ? i,j,k, where ? i,j,k ~ iid N(0, ?)
-now taking out blocking effect from error term ? ? would be smaller
-more terms in model ? ? gets smaller in error term
-Yi,j,k = response variable for subject i in treatment group j in block

Yi,j,k model

-reduces bias and effects of confounding variables, makes groups comparable
-allows for cause and effect conclusions
-allows for use of probability models

random assignment of treatments is good to...

experimental unit
-smallest part of experimental material that is assigned to a factor level combination within a study

a person, object, or some other well-defined item upon which a treatment is applied

factors of interest
-explanatory variables
-could be observational or experimental (randomly assigned)
-levels are like diff options, if multiple factors of interest, multiply levels to get treatment options

factors included in a study to determine if diff levels of them impact the results

nuisance factors
-extraneous variables
-but its variability in response needs to be minimized or explained
-BLOCK on nusiance factors

factor that has an effect on the response but is of no interest to the experimenter

crossed factor design/two-way factorial design
-two factors are crossed if all possibly combos of levels are randomly assigned to groups of subjects
-balanced design: when there are the same number of observations for each treatment
-pros: two-way design

a study in which researchers cross two or more independent variables, or factors, and study each possible combination of the levels of the variables

Yi,j,k = � + ? j + ? k + (??) j,k + ? i,j,k, where ? i,j,k ~ iid N(0, ?)
-? k = treatment effect of factor 2
-(??) j,k = interaction effect

model w 2 variables of interest

repeated measures design
-each subject is a block of time slots
-experimental unit is a time slot (NOT subject)
-basically multiple measures on same subject, bc it's same subject, the experimental unit is the diff measures on the subject?

An experiment using a within-groups design in which participants respond to a dependent variable more than once, after exposure to each level of the independent variable

randomized block design
-blocked on a characteristic that could affect response
-subjects are experimental units

First, units are sorted into subgroups or blocks, and then treatments are randomly assigned within the blocks

longitudinal experiment
-each subject receives 1 treatment, same variable is measured on same subject at diff points in time
-pros: each subject has own pre-existing conditions that can affect response, variability is reduced when compare subjects to them

a study in which subjects are followed and periodically reassessed over a period of time

crossover experiment (counterbalanced)
-subjects randomly assigned to 1 of several orders for receiving treatment levels
-pros: randomizing order of treatment reduces confounding

an experiment in which subjects get exposed to one condition, then cross over to another condition

split plot design
-randomly assign subjects to 1 of several levels of the 1st variable, then randomly determine order of the presentation of levels of 2nd variable
-experimental units receive all levels of 2nd variables, but only 1 level of 1st variable

involves assessing a general intervention on the whole plot and assessing other treatments to subplots within the whole plot