research final exam study guide!!!

what is the concept of dualism as introduced by rene descartes?

Dualism, by René Descartes, is the idea that the mind and body are two separate substances:

The mind is non-physical and responsible for thinking and consciousness.

The body is physical and operates like a machine.

He believed they interact but are fundamentally different — this is called mind-body dualism.

the relationship between metaphysics and the physical worlds:

Descartes’ dualism connects metaphysics (mind) and physical science (body), trying to explain how the immaterial mind can influence the material body

how do we determine article credibility?

by ensuring:

peer reviewed:

journal editorial board and review process, its been evaluated and approved by experts before publication to ensure credibility!

journal quality:

reputation of the article measured by impact factor- which is measured by how often the article is cited and referenced by other researchers. influential and respected in the field!

author reputation:

credentials of the 1st primary author

funding source:

NIH, or other major private philanthropic orgs enhance credibility because they have rigorous review processes that are competitive and means it is ethical credible and high quality!

how do we create a research question?

identifying the problem vs research rationale!

identifying the problem: Start with something you’ve noticed or questioned in practice — what needs to be explored or solved? Comes from your curiosity, clinical experience, or observation

clinical experience:

challenges and experiences in clinic influence areas of curiosity and knowledge Your hands-on experience can spark curiosity—maybe a pattern, challenge, or success you’ve seen with patients that needs deeper understanding.

clinical theory

whats the theory at the source of the question? Ask: What existing theory or concept is behind your question? It helps guide the purpose and expected relationships.

professional literature:

Check if previous research has:

Gaps (what’s missing?)

Conflicts (do studies disagree?)

Limitations (was something not fully explored?)

research rationale: Why is this research important? What will it add to the field or improve in practice?

This is your justification

You’re checking the literature to see:

Is this already well-studied?

Are there gaps, contradictions, or limitations?

Is it worth studying for real-world impact?

review of literature:

Look for:

Systematic reviews or meta-analyses (high-level summaries of evidence)

Is your target population included in those studies?

theoretical framework:

What models or theories have previous studies used?

Are they relevant to your patient population or condition?

purpose

Define what your study will do — explain the aim clearly and tie it to your question and rationale.

what are the components of the PICO question?

P – Population/Patient/Problem

Who are you studying?🧠 Example: Adults with stroke

I – Intervention

What treatment, exposure, or approach are you looking at?🧠 Example: Balance training

C – Comparison (optional)

What is the intervention being compared to?🧠 Example: Standard physical therapy or no intervention

O – Outcome

What result are you measuring?🧠 Example: Improved walking speed or quality of life

inductive vs deductive reasoning:

Inductive reasoning:

Starts with specific observations and moves toward a general conclusion.

begins w experiences

Observing!

specific observations to general conclusions

Deductive reasoning:

acceptance of a general premise upon which inferences can be drawn to reach a specific conclusion.

testable hypothesis

general observation to specific conclusion

EBP: how do we know things as a profession?

Tradition-inherited knowledge and accepted precedent

“That's the way it’s always been done!”

Authority- statements or beliefs or influential leaders

“That’s what the experts say!”

Experience- trial and error in your own practice

“It’s worked for me before!”

* Experience, Authority, and Tradition

ALSO:

scientific evidence

deductive and inductive reasoning

experience- trial and error in own practice

authority- statements or beliefs or influential leaders

tradition- inherited knowledge and accepted precedent

what are the 3 components of EBP?

1. Best available research evidence

2. Patients values and preferences

3. Clinical expertise

what is the process of EBP decision making?

The 5 A’s! Ask, Acquire, Appraise, Apply, Assess

1. ASK a clinical question relevant to patient problem

a. background and foreground

2. ACQUIRE relevant literature guided by your question

3. APPRAISE the literature to determine if it meets standards & if findings are relevant

4. APPLY the evidence to your clinical situation

5. ASSESS the effectiveness of the evidence

Willowbrook Study:

Willowbrook Study 1956: Hepatitis Study at State School

At the Willowbrook State School researchers deliberately infected children with intellectual disabilities with hepatitis to study the disease.

Tuskegee Study:

Tuskegee Study 1932: US Public Health Syphilis Study

The U.S. Public Health Service studied untreated syphilis in Black men in Tuskegee, Alabama, without telling them they had the disease.

the nuremburg code:

why was it created? and what does it address?

1949

Why was it created?

Created after Nazi doctors were tried for horrific human experiments during WWII on lives deemed unworthy.

What does it address?

Emphasis on voluntary consent given by an adequately informed participant.

Researchers must be scientifically qualified!

what are the ethical principles of the Belmont Report?

Respect for persons:

Respecting personal autonomy and protecting those with diminished autonomy.

Beneficence:

Risk-benefit analysis, maximizing benefits and minimizing harm.

Justice:

Equitable distribution of benefits and burdens

what are the significant elements of informed consent?

Before the signature:

Researchers must ensure the participants understand what they're signing & can ask questions.

Free to Withdraw:

As volunteers, subjects are free to discontinue participation at any time without penalty and may request that their data is discarded.

Vulnerable Participants:

Consent without intimidation or coercion due to influence in populations. Special care with those who cannot consent for themselves, i.e. children & cognitively impaired.

define: fabrication, falsification, and plagiarism:

Fabrication:

Making up data or results and reporting on them

Falsification:

Manipulating data or processes, changing or omitting data or results- tweaking numbers slightly in favor of research, or getting rid of certain data.

Plagiarism:

Using another person's ideas, process, results, or words, without giving them credit.

independent vs dependent variables:

Independent:

Manipulated or controlled by the researcher

Presumed to explain or influence the dependent variable

“Variable that is changed”

Dependent:

Response variable

Outcome/variable affected by the change

Independent variable: What you change

Dependent variable: What you measure

what are the numerical variables?

and define each! give examples!

diff between ratio and interval

Continuous:

infinite number of values along a continuum, not limited to a whole number. I.e. weight: 25.5lbs, 25.487lbs or distance

Ratio variable: true zero!!!!

measured from an absolute zero

Has an absolute (true) zero meaning zero means none

Example: 0 height = no height at all. weight age force

Interval variable: no true zero!!!!!

rank order with equal intervals, zero does not indicate absence

Has no true zero; zero is just a point on the scale, not the absence of the thing.

Example: 0 degrees Celsius doesn’t mean no temperature, just a reference point (cold).

SAT scores

Discrete:

measured in separate units, cannot be measured in intervals less than 1= limited to a whole number. It means no decimals or fractions are used; values jump in full units.

I.e. how many children you have! number of visits or heart rate 84 bpm

ratio- equal intervals measured from true zero

interval- equal intervals, no true zero

what are the categorical variables?

and define them!

Nominal- categorical labels

Gender, blood type, label associated

Ordinal- rank order

MMT, pain scale

Nominal = Name only

Ordinal = Order matters

define reliability:

Predictable & reliable under steady conditions across researchers and pts.

random vs systematic error:

Random error:

Happens by chance

Causes inconsistent results

Systematic error:

Causes biased results in the same direction

❌ Not fixed by repeating—needs correction of the source🔁

what are the sources of measurement error:

The Rater:

skill, not following protocol, inaccurate recording, distraction, bias

The Instrument:

Imprecise, affected by environmental changes, background noise

The Measured Characteristic:

Physiological response, Changes in performance, personal characteristics (motivation, fatigue, cooperation)

what is relative reliability?

Measures consistency of ranking between scores

Answers: Do people stay in the same order across tests?

🔧 Measured with Intraclass Correlation Coefficient ICC

0-1

1= completely reliable

0= no reliability completely random

When it says relative reliability is the proportion of total variance in a set, it means:

👉 It looks at how much of the total variation in scores is due to true differences between people, not measurement error.

Relative reliability = how much of the total variance is due to real differences between people (consistency in ranking)

what is absolute reliability?

Measures the amount of measurement error

how much measurement error there is

expressed in original units, how much of the measured value is likely due to error

Answers: How close are the actual scores to each other?

✅ Example: If someone scores 90 and 91, the difference is small = good absolute reliability

🔧 Measured with standard error of measurement (SEM)

Absolute reliability = how much variance is caused by measurement error (how much scores wiggle because of mistakes)

what is test retest reliability?

Determines if the instrument measures performance consistently

- intervals

- carryover

- testing effects

Checks if a measurement tool gives consistent results over time

Same test, same person, different time

✅ High test-retest reliability = stable and repeatable results

🧠 It tells you: “If nothing changes, will I get the same score again?”

Test-retest reliability is mainly about the instrument or tool and how consistently it measures over time.

what is rater reliability?

intra vs inter:

Inter-rater reliability: ✅ Consistency between different raters

variance between two or more raters testing the same subjects

Intra-rater reliability: ✅ Consistency of the same rater over time

stability of data from a single rater across trials

what are the ways we can improve/maximize reliability?

- Standardize measurement protocols

- Train Raters

- Calibrate and improve instruments

- Take multiple measurements

what is norm referencing?

Compare and rank an individual within a given population- percentile ranks

Norm referencing is when a person’s test score is compared to others’ scores, not a fixed standard.

🧠 Think: “Compared to the norm”

what is criterion referencing?

A fixed standard that represents an acceptable level of performance

Criterion referencing means comparing a test score to a fixed standard or specific goal, not to other people’s scores.

🧠 Think: “Meet the criteria”

what is minimal detectable change MDC?

The amount of change that must occur to be confident that it is actual change and not attributable to error.

Confident in saying that what we are seeing is actual change, not within error amount.

Based on standard measurement of error SEM

Smallest change that shows a real difference, not just measurement error

what is minimal clinically important difference MCID?

The smallest difference that signifies an important rather than trivial difference in measurement.

Smallest change that is meaningful to the patient

Focuses on clinical importance

✅ Answers: “Does this change matter to the patient?”

descriptive statistics:

what is distribution vs distribution frequency?

distribution:

total set of scores for a particular variable

Shows the overall pattern of the scores: "Scores ranged from 6–10, most were around 8."

Frequency Distribution:

A table of ranked order scores that shows how many times the value occurred

Shows how often each score occurred:

6 → 2 times

name the skew:

negative skew

L tail = negative skew

name the skew:

positive skew

R tail = positive skew

measures of central tendency: what is it?

what is mean, median, and mode?

Mean: the “average”

Median: the middle number when scores are in rank-order

Mode: the score that occurs most frequently in a distribution

Central tendency = a way to describe the center or typical value of a data set.

👉 Includes: mean, median, mode. It's how we summarize what’s “normal” or “average” in the data.

what is standard deviation? and how is it interpreted?

- Represents variability or dispersion of scores around the mean

- Smaller standard deviation means the scores are less spread out from the mean

Standard deviation = how much the data spreads out from the mean.

Low SD → scores are close to the mean

High SD → scores are spread out wide

what is inferential statistics?

Inferential statistics are concerned with testing hypotheses and using sample data to make generalizations concerning populations.

statistical inference:

probability vs sampling error

Probability:

the likelihood that an event will occur, given all possible events

Sampling error:

the difference between values observed in the sample and in the population

What we observe vs what is actually true of the population

Probability = chance that a result happens by random chance

Sampling error = the difference between a sample’s result and the true population value because the sample isn’t perfect

inferential stats:

what is confidence interval? how is it interpreted?

A range that should contain the population mean

It’s a range of values calculated from your sample data

You’re “confident” (usually 95% confident) that the true population mean lies somewhere within this range

It accounts for sampling error and variability

For example, a 95% CI means: "We are 95% confident the true value is between X and Y"

A narrow CI means more precise estimate

A wide CI means less precision, more uncertainty

expressed as 95% or 99% 95% of the time the CI would contain the true population mean

what is alternative hypothesis vs null hypothesis?

Null hypothesis (H₀) =

a statement that there is no effect or no difference no relationship between variables.

Alternative hypothesis (H₁) =

a statement that there is an effect or a difference between groups or variables.

Hypothesis stating the expected relationship between independent and dependent variables

alternative hypotheses:

directional vs non-directional:

Non-directional Hypothesis:

does not specify which mean is expected to be higher (uses a two-tailed test)

Directional Hypothesis:

indicating an expected direction in the difference between means (use a one-tailed test)

Directional hypothesis: predicts the specific direction of the effect

Example: "Dance therapy improves quality of life" (expecting increase)

Nondirectional hypothesis: predicts a difference, but no specific direction

Example: "Dance therapy affects quality of life" (could improve or worsen)

what is a type I error?

- include descriptions of alpha level and p value

Type I Error: rejecting the null when it was true

You think your alternative hypothesis is true (reject null), but it actually isn’t true

🔹 Alpha level (α)

You set this before the study (commonly 0.05)

It’s the maximum risk you’re willing to take of making a Type I error (rejecting a true null)

🔹 P-value

You get this from your data

It tells you the actual probability that your results happened by chance if the null is true probability that the observed difference occurred due to chance

If p ≤ α → Statistically significant → Reject the null

If p > α → Not significant → Fail to reject the null

👉 Alpha = your cutoff

👉 P-value = your result

👉 You check: “Did the result beat the cutoff?”

what is a type II error?

include descriptions of beta and power:

Type II Error:

Failing to reject the null when it is incorrect

You think your alternative hypothesis is not true (fail to reject null), but it actually is true

Power= 1-B

Power is probability that a test will lead to rejection of the null Power = detects true differences

β = 0.20 → 20% chance of missing the truth reasonable to protect against type 2 error

Power = 0.80 → 80% chance of catching the truth if it’s there

Beta = the risk of missing true differences

Type II error is the mistake

β is how likely that mistake is

Power is how good your test is at avoiding that mistake

If it has high power (usually ≥ 0.80), you're more likely to find real differences

Power (1−β) helps you design your study well enough so that if a true effect exists, you’ll be able to find it

what is effect size? and how is it interpreted?

The degree to which the null hypothesis is false

the magnitude of the difference between groups

The null hypothesis says: "There is no effect or no difference."

Effect size measures how much your data deviates from this “no effect” idea.

A larger effect size means the null hypothesis is more strongly false — there is a bigger difference or stronger relationship.

A small effect size means the null is only slightly false — differences or effects are minor.

A big effect size means the difference or relationship you found is strong and meaningful.

It suggests the null hypothesis is more clearly false — your intervention or variable really made an impact.

Measures how big or meaningful the difference or relationship is in your data

Cohen’s d (difference between two means)

0.20 = small effect size

0.50 = medium effect size

0.80 = large effect size

what is parametric vs non-parametric?

Parametric: Used to estimate population parameters

Must meet certain assumptions to be valid!

Random sampling

Interval or ratio (cont. data)

Normal distribution

Equality of variance (homogeneity)

Nonparametric: less powerful analogs "less powerful version"

Valid alternative when assumptions are not met

Use parametric tests when your data meets assumptions (normality, equal variance). Use nonparametric tests when it doesn’t or when data is ordinal/ranked.

comparing means:

paired t-test:

compares means from the same group at 2 times (pre and post)

Used when subjects serve as their own control

Compares differences in scores for each pair so the subject is only compared with themself

Reported as t statistic:

t [df] = t statistic, p=p value, CI95 [lower, upper]

Compares means within the same group at two different times

wilcoxon test

Example: Compare balance scores before and after dance therapy in the same patients

comparing means:

independent t-test

compares means between 2 different groups

Each group consists of a different set of subjects

Reports as same t statistic as above^

assumptions include equality of variance (levenes test)

Mann whitney U

Independent t-test

Compares means between two different groups

Example: Compare balance scores between Group A (dance therapy) and Group B (no therapy)

define when you would use ANOVA?

define:

one way ANOVA

two way ANOVA

ANOVA: comparing more than two means, 3 or more treatment groups or conditions, between groups and within groups

based on F statistic

parametric requires assumptions to be met

non parametric- kruskal wallis test

One-Way ANOVA:

one independent variable with 3 or more levels (factors)

You want to see if three different exercise programs (A, B, and C) lead to different average improvements in patients’ strength.

Two-Way ANOVA:

two or more independent variables (factors)

You want to see the effect of exercise program type (A, B) and gender (male, female) on patients’ strength improvement.

ANOVA:

repeated measures ANOVA

mixed design ANOVA

Repeated Measures ANOVA:

within subjects design, use with same subject under multiple conditions

You want to test if patients’ pain levels change across three time points: before treatment, after 2 weeks, and after 4 weeks.

Mixed Design ANOVA:

Combines both between-subjects and within-subjects factors in one study

Tests effects of different groups (between-subjects) and repeated measures (within-subjects) simultaneously

Study looks at the effect of two different diets (Diet A vs. Diet B — between-subjects) on weight loss over 3 months (measured monthly — within-subjects).

what is a correlation? how is it measured?

Correlation: The degree of association between 2 variables!

a statistical technique used to measure and describe the relationship between two variables

Correlation Coefficient (r): expresses a degree of relationship

Measures the strength and direction of a linear relationship between two variables

Values range from -1 to +1

+1 = perfect positive correlation (both go up together)

-1 = perfect negative correlation (one goes up, other goes down)

0 = no correlation

the extent to which x and y are "co-related"

THE CLOSER THE VALUE IS TO PLUS OR MINUS 1, THE STRONGER THE RELATIONSHIP

what does it mean for two factors to be correlated?

They change together in some way — as one goes up or down, the other tends to go up or down too

The strength and direction of their relationship is measured by the correlation coefficient (r)

Correlation does not mean causation — just that there’s a relationship, not that one causes the other

draw a positive and negative scatter plot

correlations:

nonparametric vs parametric

If your data is normal & continuous → use Pearson (parametric)

If your data is non-normal or ordinal → use Spearman or Kendall (nonparametric)

Parametric Statistic:

Continuous variables

Normal distributions

Nonparametric options:

Spearman rank correlation coefficient

Kendall’s Tau-b

what is a regression? how is it measured?

A statistical method used to predict the value of one variable (dependent) based on one or more other variables (independent)

Coefficient of Determination (R²): proportion of variance shared by two variables

Tells you how much of the variation in the dependent variable Y can be explained by the independent variable X

Ranges from 0 to 1 (or 0% to 100%)

Higher R² = better prediction

Tells us how much of a variance in the dependent variable Y can be explained by the independent variable of X.

regression:

simple vs multiple regression:

- what is MR used for?

Simple Regression

A statistical method that uses one independent variable (X) to predict one dependent variable (Y).

Predicting a person’s test score (Y) based on the number of study hours (X).

Multiple Regression: multiple IVs, 1 DV

A statistical method that uses two or more independent variables (X₁, X₂, X₃...) to predict one dependent variable (Y).

Predicting blood pressure (Y) based on:

age (X₁)

weight (X₂)

exercise level (X₃)

MR is used for applied prediction, decision making, theoretical explanation.

why are regressions used?

ESTIMATION AND PREDICTION!

write out the research designs levels of evidence:

strongest: clinical practice guidelines and health tech assessment

systematic review and meta analyses

randomized controlled trials

controlled clinical study

retrospective and prospective cohort

case report/case series

weakest: expert opinion

research design:

inclusion vs exclusion criteria:

Inclusion Criteria:

primary traits of a target population “who we want!”

May include factors such as:

Demographic

Clinical presentation

Geographic location

Exclusion Criteria:

factors that preclude participation- don't want anything that confounds findings!

May include potentially confounding factors such as:

Comorbidities

Contraindications to intervention

Inability to follow study instructions

elements of a "true experiment" / requirements of an RCT!

1. Manipulation of IV’s

Researcher imposes specific experimental conditions

2. Random Assignment

Each subject has an equal opportunity of being placed in any group

3. Use of a control group

active vs inactive control groups:

Inactive Controls:

does not receive a comparison treatment

Placebo: treatment w/o an active component

Sham: essentially a placebo for a physical intervention i.e. providing ultrasound w/o turning on the machine

Active Controls:

conventional or standard care is provided

Can compare against a new treatment approach

Still receives some sort of treatment

blinding:

- single-blind, double-blind, triple-blind:

Single-blind: Participant doesn’t know

Double-blind: Participant and researcher don’t know

Triple-blind: Participant, researcher, and analyst don’t know

Each level helps reduce bias more!

experimental design:

pretest-posttest design

Basic RCT with 1 IV

A pre-posttest control group design is a true experimental design where:

There are two or more groups (usually an experimental group and a control group).

Both groups are tested before the intervention (pre-test)

One group gets the treatment, the other does not (or gets placebo/standard care)

Both groups are tested again afterward (post-test)

Strong internal validity & strengthens evidence that effects of the treatment account for the group differences in posttest scores!

independent t test for 2 groups one way anova for 3 or more groups

mann whitney U 2 groups

kruskal wallis 2 or more groups

experimental design:

posttest only design

- why would we choose this?

There are two or more groups (experimental and control)

Only a post-test is given — no pre-test

One group gets the treatment, the other does not

Then you compare outcomes after the intervention

Identical to pretest-posttest design EXCEPT no pretest for either group!

Why a posttest only?

- If pretest is impractical or potentially reactive

- Decreases bias if prettiest could influence posttest scores

- Increases external validity

- Best done w large sample size to balance out interpersonal characteristics

experimental design:

factorial design

A factorial design is a research setup that studies two or more independent variables (factors) at the same time to see:

Individual effects of each factor (main effects)

Combined effects (interactions) between the factors

Used to account for the interaction of several variables

2 or more IVs

3×3 Factorial Design

2 independent variables

Each has 3 levels

Example:

Factor A: Exercise (yoga, strength, swimming)

Factor B: Diet (low, medium, high protein)

→ 9 total groups (3×3)

analyzed with ANOVA 2 way and 3 way most common

what is a randomized block design?

A research design where participants are grouped (blocked) by a characteristic (like age, gender, severity), and then randomly assigned to treatment groups within each block

experimental design:

repeated measures

ONE GROUP is tested UNDER ALL CONDITIONS

Subject serves as their own control

Advantage: control for individual differences

what is a carryover effect?

A carryover effect occurs when the effect of a previous treatment “carries over” into the next treatment period — and influences the results of later conditions

experimental design:

crossover/carryover design

Two levels of IVs are repeated

● Control for order effects- counterbalancing the treatment conditions

Means you change the order for different participants to cancel out order effects.

● Half receive intervention A followed by intervention B, the other half receives intervention B followed by intervention A.

● Should only be used if the patient’s condition/disease will not change

much over time

● When the treatment has cumulative effects, a washout period is needed

You’re testing 2 pain meds (A and B) on the same group of patients:

Group 1: Gets drug A → then drug B

Group 2: Gets drug B → then drug A

paired t test or wilcoxon

what is a quasi-experimental design?

Similar structure to experimental but lack:

● Random assignment

● Comparison groups

● Both

Used when randomization and/or control groups are not possible or unethical

Threats to internal validity

quasi-experimental:

interrupted time series

An interrupted time series design is a quasi-experimental study where:

series of measurements that are interrupted by one or more treatment occasions

You measure the same group repeatedly over time, then "interrupt" that timeline with an intervention, and keep measuring to see if there's a noticeable change in the trend.

Strengthens causal inference if trends clearly change after the intervention

It splits time into “before” and “after” segments. Pre and Post!

Benefit is to show trends over extended periods

IV is time!

Series of measurements that are interrupted by one or more treatment occasions

One group studied

Most effective at evenly distributed intervals

quasi-experimental:

non-equivalent groups design

It’s a type of quasi-experimental design where:

You have two or more groups (like experimental and control)

But the groups are not randomly assigned

You still compare pre-test and post-test scores between them

🧪 Example:

You compare math scores between:

One class that received a new teaching method

Another class that used the traditional methodBut students were already assigned to their classes — you didn’t randomly place them. ➡ ️ The groups are non-equivalent

Similar to a pretest-posttest design except subjects are not randomly assigned

subjects can self select!

Limitation and potential bias due to lack of random assignment

what is a single subject design?

A single-subject design (SSD) is a type of research where one person (or a very small group) is studied intensively over time, and acts as their own control.

improve decision making in clinical setting

no one size fits all, allows for individualized care because RCTs may not be adequate since they generalize to all people

single subject design:

whats the design follow?

baseline intervention phases!

A = Baseline (no treatment) control condition!!!!!

B = Intervention

A-B design!

Measuring the response of the target behavior at frequent and regular intervals

2 testing periods: A-B Design

SSD:

N of 1 Trials (crossover)

One person, randomly switches between treatments (like A-B-A or B-A-B)

Used as an active decision making tool

● Which treatment should I use for this patient?

● Most commonly used to determine individual effectiveness of drugs

Crossover design:

Treatment and placebo are altered until the pt and clinician reach a decision on preference for one.

✅ N-of-1 Trial

A study done on just one participant

That one person tries multiple treatments over time

Outcomes are measured during each phase to see which treatment works best for them

🔁 How Crossover Design Fits In:

N-of-1 trials use a crossover design: The same person gets Treatment A, then Treatment B, then maybe back to A

There are “washout” periods in between to clear the effects of the previous treatment

The person is their own control

Great for personalized medicine

Helps find the best treatment for an individual, not just what works on average

what is descriptive designs?

Observational approach

Seeking to document:

● Traits

● Behaviors

● Conditions

case report vs case series vs case study!

Case report:

Detailed description of individuals condition or response to treatment (can also focus on a group)

Helpful for clinical theories, or new approaches to diagnoses

Can provide novel research questions and hypotheses

Can emphasize unusual problems or outcomes

Case series:

Expansion of a case report involving observations in similar cases

Case study:

NOT a case report

Qualitative research method investigating broad questions related to individuals or systems

Deep dive into one case (person, event, organization, etc.) ✅ Use to understand context and all factors around a single unit in detail

what is observational research?

what types?

cohort, case control, longitudinal, cross sectional

Comparing groups without assigning exposure

Researchers observe without manipulating variables ✅ Use when you can’t ethically or practically assign treatment

observational research:

longitudinal studies

Following subjects across time and collecting data at intervals

Prospective - direct recording or measurements

Retrospective - examining previously collected data

Prospective = watch it happen

Retrospective = look back at what happened

Longitudinal = over time, either way

observational research:

cross-sectional studies

taking a population "snapshot" at a single point in time

collects data about exposure and outcome at the same time

more efficient than longitudinal studies

- risk of reverse causation

observational research:

cohort studies

Prospective or retrospective!

Follow a group with shared exposure (e.g., smokers) over time to see outcomes

subject selection is free from target outcome at baseline

susceptible to outcome

challenges- classifying exposure and attrition

Cohort Study = follows groups based on exposure to see who gets the outcome

Prospective cohort: Starts now, follows participants forward in time to see if they develop the outcome.

Retrospective cohort: Looks back at existing records to see who was exposed and who developed the outcome

.

Key: Both look at exposure first, then outcome — but differ in when data is collected (future vs past).

observational research:

case control studies

Compare people with a condition (cases) to those without (controls)

determining if groups differ in terms of exposure history or presence of risk factors by looking backwards

Look backward to identify prior exposures

✅ Use when condition is rare or time/resource-limited

➡️ Retrospective (looking back)

challenges- selection, observation, and recall bias and confounding factors

match controls to minimize bias

You start with the outcome

(cases = have the condition, controls = don’t)

Then you look back in time to compare past exposures or risk factors

Case-control = outcome first, then search for cause (retrospective by design)

what is the value of qualitative research?

Qualitative research methods provide researchers with the ability to:

Explore complex human beliefs

Attitudes

And behaviors

That can generate a deeper understanding of the impairments, functional limitations, and disability of our patients as well as therapeutic interactions

qualitative research:

grounded theory

Builds a theory from the data (theory is "grounded" in participants' words)

seeks to form or discover a theory

theory development that explores the how and the why of the phenomenon through systematic processes

qualitative research:

phenomenology

Explores the lived experience of people

seeks to describe common meaning of individuals of a lived experience of a concept/phenomenon

✅ Use when studying what it's like to go through something (e.g., living with pain, grief, illness)

explore the meaning of something from the vantage point of those who have experienced it

qualitative research:

case study

Deep dive into one case (person, event, organization, etc.)

✅ Use to understand context and all factors around a single unit in detail

study of a real life "bounded system"

- individual, group, or organization

The case being studied is clearly defined in terms of time, place, people, or context — it has boundaries.

detailed in depth data collection across multiple sources that provides case descriptions and themes

comprehensive descriptions of a single patient or organization. Have specific design and analysis

qualitative research:

ethnography

Studies cultures or social groups in their natural environment

focused on "culture sharing group"

- immersion in day to day lives

✅ Use when you want to immerse yourself to understand beliefs, behaviors, and norms

examining social units or cultures through direct presence and observation in the natural setting

qualitative research:

generic

A flexible, non-specific approach that doesn’t fully fit the other types

✅ Use when you just want to describe and interpret experiences without forming theory or doing deep cultural or philosophical analysis

standard interpretive research

qualitative research:

methods of data collection

Qualitative methods of data collection:

Participant observation

Interviews!!!

- Structured

- Semi-structured

- Unstructured

Focus groups

Written documents

qualitative data analysis:

coding types

structural, descriptive, in-vivo, process coding

Coding: a word or phrase that summarizes or captures the essence of a portion of data

TYPES:

Structural: conceptual phrase "Health beliefs"

Descriptive: basic labels "Fear," "family," "transportation"

In-Vivo: using the participants own language "I felt invisible"

Process coding: using gerunds (words ending in -ing) "Coping," "struggling," "seeking help"

searching literature:

truncations

An * asterisk is commonly used as a truncation symbol

Allows for any ending

Provides multiple variations

Not all databases use this

A search technique that uses a symbol (like *) to find all word endings or variations of a root word

Helps get more results in database searches

Searching for child* will find: child, children, childhood, childish, etc.

searching literature:

combining terms using boolean operators

Words that help combine or exclude keywords in database searches to narrow or broaden results.

Common operators:

AND → Narrows search (must include ALL terms)

OR → Broadens search (includes ANY of the terms)

NOT → Eliminates articles that contain the 2ND term NARROWS search

Diabetes AND exercise → finds articles with both words

Diabetes OR obesity → finds articles with either word

Diabetes NOT type 1 → finds articles about diabetes but excludes type 1

searching literature:

MeSH terms

MeSH Terms (Medical Subject Headings)

A controlled vocabulary used in databases like PubMed

Standardized terms that describe the content of articles

Help you find all articles on a topic, even if different words are used

Find content with different terminology for the same concept!!!

evidence for validity:

construct, content, and criterion related

validity: measuring what we say it will measure!

content validity:

make sure its getting all points to support the findings i.e. studying depression just doesn't study fatigue but also mood changes, motivation, fatigue, anger, etc.

criterion related validity:

what we are measuring vs the gold standard

construct validity:

It tells you how well a test or tool measures what it’s supposed to measure — the "construct" or concept.

creating a search strategy:

1. identify terms based on your research question

2. set your limits and exclusions

3. create a term combinations

4. determine appropriate database to search

* run a preliminary search

* adjust terms and limits as needed

components of descriptive stats:

what is Q3-Q1? and what is Q2?

what is variability?

- range

- percentiles

- quartiles

shape (distribution shape, normal or skewed bell curves)

central tendency (mean median mode)

variability shows the differences when the other two factors don't provide a clear picture!

- range: differences between highest and lowest values in distribution

- percentiles: scores position in distribution relative to all other scores

- quartiles: breaks distribution into 4 equal parts

Q2 is median 50%

each quarter is 25% of distribution

Q3-Q1 is interquartile range

-standard deviation represents variability or dispersion of scores around the mean

* smaller SD means the scores are less spread out from the mean

how do we determine statistical power?

power:

desired levels can be determined in planning stages

alpha level:

best level needs to be determined considering impact on type 1 and type 2 error

number of subjects:

larger the sample, the greater the statistical power

effect size:

degree to which the null is false

paired t test:

- how is it reported?

- what is the nonparametric alternative?

reported as t statistic:

t[df]= t statistic, p=p value, CI95 [lower, upper]

nonparametric alternative:

WILCOXON TEST

independent t test:

- assumptions include?

- reported as?

- nonparametric alternative?

assumptions include equality of variances:

LEVENES TEST!

reported as t statistic

nonparametric alternative:

MANN WHITNEY U TEST

a test used to assess the assumption of equality of variance?

levene's test!

ANOVA:

- reported as?

- non parametric alternative?

- effect size for anova intepretation:

based on F statistic

non parametric:

KRUSKAL WALLIS TEST

effect size- eta squared n^2

small: .01

medium: .06

large: .14

CORRELATION INTERPRETATION:

<- 0.25 little or no relationship

.25 to .50 low to fair

.50 to .75 moderate to good

>-.75 strong!

correlation nonparametric options

spearmann rank correlation coefficient

kendalls tau-b