Historical context and process of publication

i. 854-931AD � Ishaq bin Ali al-Rahwi wrote Ethics of the Physician describing how physician's patient notes were reviewed by a local medical council
ii. 1620� Francis Bacon wrote Novum Organum which described a universal method for generating new science

confidence intervals

i. It means that ____% of the time we expect the true mean to be found within the upper and lower bounds of the CI
ii. If the experiment is repeated 20 times, then 19 of those times we will get a mean that is within that same CI.
iii. If the point estimat

positively skewed distribution

tail to the right
-high mean

negatively skewed distribution

tail to the left
-low mean
-median and mode are closest together

Null Hypothesis (H0)

there is no difference between groups
statistical significance= reject null
not statistically significant= fail to reject

Alternative Hypothesis (Ha)

there is a difference between groups

1 tailed test

directional
(looking for differences on one side of the distribution curve)
a. Specific direction (one group will do better than the other)

2 tailed

non-directional (looking for differences on either end)
a. predicts relationship between variables but not the direction of the relationship
b. One group will be different than the other

normal data distribution

test difference between mean values

skewed data distribution

test differences between medians

bimodal distribution

test differences between modes

P value

probability that the observed difference between groups (means, variances, proportions) occurred by chance and is not a true difference between groups

P> alpha level

fail to reject null

P< alpha level

reject null
-difference is true and not due to chance

Type 1 error

in reality there is no difference
-should not have rejected Ho
-probability = alpha level
-increasing alpha = decrease beta
-studies with large samples may detect small differences that have no clinical relevance

type 2 error

there is actually a difference
-should have rejected Ho
-probability is the beta level
-studies with small samples may fail to detect large differences

statistical power

probability that the statistical test can detect differences in the population when the difference truly exists
-common threshold is 80%

independent variable

the maneuver or exposure under investigation (experimental intervention)

dependent variable

The outcome factor; the variable that may change in response to manipulations of the independent variable.

continuous variable

-numbers or digits within a specific range
-Defined mathematical relationship between consecutive values
Weight, BMI, age (could be discrete or ordinal if asking for a range)

categorical variable

-have distinct categories which are mutually exclusive with no mathematical relationship to each other. Can be made dichotomous by assigning a "cut-off" or threshold
1. yes or no
2.presence or absence of injury
3.normal or abnormal
4.ethnicity, race, gend

discrete variables

-numerical data with values represented by whole numbers only
1. Visit counts
2. Number of treatment sessions

nominal

1. Categorical variables assigned to describe a unique attribute
2. No specific numerical order or mathematical relationship between values
3. Also known as a dichotomous variable if it has only 2 values
yes or no
presence or absence of injury

ordinal

1. Values ranked in ascending or descending order
2. No definite measurable distance between consecutive variables
3. Distances between each variable may not be equidistant
-Likert scales on surveys � strongly disagree, disagree, neutral, agree, or strong

interval data

1. Definite measureable distance between each consecutive value
2. Distance or interval between each consecutive value is equidistant
3. No absolute zero point (arbitrary zero) and relationship between numbers are not directly proportional to each other
4

ratio data

1. Interval data with an absolute or true zero (absence of value)
2. No negative values possible
3. Numbers are directly proportional to each other. Allows for interpretation of absolute or relative changes
4. Most biological measurements that are numeric

precision

1. you get the same value each time the measurement is taken
2. Smaller the error --> more precise

accuracy

1. tells you how close the measured value is to the true or actual value that which you are measuring

Reliability

1. The stability and repeatability of an outcome measure/ assessment tool
2. Reliability of a measure is population specific (limitation can depend on what pop they are performed on)

test-rest reliability (reproducibility)

a. given that a measure is stable in an individual or a group of individuals do we get the same results if the rest is repeated more than once

inter-rater or inter-observer reliability

a. for a stable measure in an individual, do we get the same results if different raters/observers are used

validity

the extent to which a test measures or predicts what it is supposed to

Responsiveness

1. The ability of the assessment tool to accurately detect change when it has occurred
2. Examined by administering the assessment tool before and after a treatment of known effectiveness
3. Ceiling effect of an instrument or assessment tool will affect r

selection bias

1. Sample is not representative of a population
2. Occurs during recruitment or selection of study subjects
3. Narrow or stringent eligibility criteria can result in a non-representative sample

spectrum bias

1. Type of selection bias that refers to recruiting patients who only have severe or classical symptoms
2. May overestimate the accuracy of a screening test

volunteer bias

1. Participants who volunteer for research tend to be characteristically different than general population

non-response bias

1. Bias introduced due to sample size
2. Typically occurs in observational study

Recall bias

1. Occurs in case control studies or retrospective studies
2. Cases are more likely to report higher exposure status compared to controls

social desirability bias

1. When individuals over report socially good behaviors and under report socially bad behaviors

bias introduced due to study methods

By virtue of study design (retrospective, non-randomized trials)

observer or assessor bias

1. When measurements require the judgement of an assessor

information bias

1. Due to poor quality data
2. Usually occurs in retrospective or case control
3. Data can be misinterpreted or hard to validate

misclassification bias

1. Research subjects are diagnosed inaccurately or outcomes are incorrectly identified
2. Occurs commonly in observational studies (case control, cohort, cross sectional)
3. To reduce in case control: diluting the control with a larger sample size

miscellaneous source of bias

1. Unequal distribution of confounding variable between groups

measurement error

1. May occur due to chance or due to the nature of the instrument

Confounding bias

1. Presence of extraneous variables that may influence the relationship between cause and effect

internal validity

i. study design and scientific methodology (from the methods)
1. Prioritized more in smaller, early stage efficacy trials that are concerned with actual therapeutic effects of a treatment

external validity

i. how generalizable the study is to real world situations
1. Prioritized in larger, confirmatory or effectiveness trials that are concerned with whether the treatment works

Qualitative studies

-meaning of a lived experience or social phenomenon
-purposeful sampling (reach out to specific individuals who have experienced the phenomenon)
-iterative, evolving design
-collect data through interviews, focus groups, observations, pre-existing documen

quantitative studies

-magnitude or effect or extent of a phenomenon
-sampling is random (need to fill eligibility criteria)
-predetermined rigid design
-data collection via surveys or questionnaires, clinical tools
-calculating the aggregate or average
-data analysis via data

Discourse analysis

Method that analyzes language and its role in understanding social context

Ethnography

1. Study of humans in their natural environment
2. Involves being immersed or as a distant observer over a prolonged period
3. Participant usually don't know they are being observed

Grounded theory

1. Understand the process, relationships and to develop a theoretical explanation behind social phenomenon
2. Theory is developed from data (inductive process)
3. Observe participants or conduct interviews

Phenomenology

1. Explores how individuals make sense of the world and that aims to provide insightful accounts into the subjective experience of individuals

Thematic analysis

1. Identify and categorize data into themes and categories to describe a phenomenon of interest or patterns
2. No statistical analysis

Mixed-methods

i. Combines both quantitative and qualitative methods to answer its research question

Ancillary Studies

1. Off shoot of a clinical study that has a different research objective from the parent clinical study
2.Collecting additional data in supplement to the clinical study

secondary studies

1. Off shoot of a parent clinical study but no new data is collected

Plagiarism

1. Appropriation of another person's ideas, words, results or predictions without giving them due credit

fabrication

1. Making up data or results and recording or reporting them as true
2. Making data to fit one's hypothesis

falsification

1. Manipulating research material, processes or equipment, altering or omitting data such that research is not accurately represented in the research records
2. altering a figure in a paper or removing outliers

deductive reasoning

i. Already have a question/theory that is a gap in knowledge
ii. Start with a clear observation or idea
iii. Tunnel vision (know what you want to research)
iv. Usually quantitative
theory-->hypothesis-->pattern--> observation

inductive reasoning

i. Begins with general observations (infer conclusions from data)
ii. Usually qualitative
iii. Already have data set
theory --> hypothesis --> pattern --> observation