EBP Exam Review Part 1

Ascertainment Bias

Only performing tests that will confirm your diagnosis

Example: conclusion on the expected outcome based on their expectations from the subjective.

Confirmation Bias

Only looking at data or information of what you think is going on to support diagnosis

Example: clinician believes all individuals with rheumatoid arthritis benefit from aquatic therapy based on data from previous patients who improved

Recency Effect

Judging based on most recent case and ignoring other options

Value Bias

Preferring/making decisions based on the outcome we value

Ethical Principles of PTs

Autonomy: Respect for patient choices

Beneficence: Act in patient's best interest

Non-Malficence: Do no harm

Concept

A general image or abstraction of an observable phenomena described in words

Age, fatigue, pain, ROM, strength

Construct

Non-observable phenomena that is created for research purposes defined by observable measures

Patient satisfaction, QOL, motivation, Pain

Biological Plausibility

Human body could perform in a predicted manner

Null Hypothesis

There will be no statically significant difference or relationship between groups, any observed relationships are due to chance

Alternate Hypothesis

Investigator expects to find a difference

What are the 4 probability sampling methods?

Simple random sample, systematic, stratified and cluster

Simple Random Sample

Each member of a population has equal chance of being selected

Systematic sampling

Every "nth" subject on the sampling frame is selected

Example: A study enrolls every 6th potential subject from a sampling frame. This is referred to as:

Stratified Sampling

the population is divided into subgroups and weighted based on demographic characteristics of the national population

CLuster Sampling

Used when naturally occurring pockets of population are geographically dispersed

What are the 3 non-probabilistic sampling methods?

Convenience, snowball and purposive

Convenience Smapling

Use of readily available subjects

Example: Investigators wish to study the prevalence of work place injuries in migrant farm workers. They are given permission to make announcements about the study at a local church gathering

Snowball Sampling

Recruit subjects who recruit more subjects

A non-probabilistic selection method that involves using a "word-of-mouth" technique to make eligible candidates aware of a study

Purposive Sampling

Uses hand picked subjects that meet the researcher's needs, often used in quantitative research

Individual Random Assignment

Randomly assigning each individual to a group

Block Random Assignment

Pre-determine a number of subjects to be contained in each group

Systematic Assignment

sample members count off or are numbered repetitively according to group numbers

Matched assignment

Subjects are matched into subsets based on important characteristics

Consecutive Assignment

Randomly assign order of group assignments before enrolled in a study

Power

Ability to detect a difference (or relationship) if one exists

Power formula

Power = 1-beta

How do you increase power?

Increase power by controlling extraneous variables, using homogenous groups and increasing the sample size

What error would you commit if if you don't control power?

Type ll error

Quantitative Research

OBJECTIVE, generalizable, causes and effect can be determined

Qualitative Research

Subjective, describe or explain, cannot determine cause and effect

Experimental Design

2 or more groups, random assignment and purposive manipulation of subjects

Quasi Experimental Design

A study with 2 or more naturally occurring groups (non assignment) and purposeful manipulation of an independent variable are all characteristics

Non-Experimental Design

No manipulation of subjects, and non random assignment of groups

Between Subjects (Independent)

Comparison between 2 Independent group of subjects

Within Subjects (Dependent)

Repeated Measures within the same group

Cross Sectional Design

Data collection at a single time point

Longitudinal Study

Repeated data collection over extended period of time

Retrospective Study

Looks backward in time to look at historical data to answer

Prospective Study

Looks forward to data collect in real time to answer

Factorial Design

Number of independent variables or factors included

Single factor: effect of 1 variable

Two factor: effect of 2 variables and their interaction with each other

Independent Variable

The factor you manipulate (e.g., treatment type). Predicts an outcome of interest

Dependent Variable

The outcome you're measuring (e.g., pain level).

Nominal

Categories (e.g., male/female, diagnosis type).

No rank or value placed on it

Ordinal

Ordered but not evenly spaced (e.g., pain scale 0-10, MMT)

Classification with order without equal intervals but can't do math

Ratio

Order, interval, distance and origin known, can do all math

Has true zero point like ROM

Interval

Order and interval known but no true zero point, can do addition and subtraction

Example: Temperature

Content Validity

The content of a test is representative of the meaningful elements of a variable, judged by content experts or people w experience

Construct Validity

Validity of abstract concepts that UNDERLIE the measure

What are the types of construct validity?

convergent and discriminant

Convergent Validity

Comparison of scores between two similar instruments expected to produce similar results

Discriminant Validity

Differentiation among characteristics or levels of the same characteristics

Creators of a new self-report survey for patients with amyotrophic lateral sclerosis indicate that the instrument is capable of distinguishing among three levels of symptom severity.

Criterion Validity

One measure is systematically related to other measures or outcomes

What are the types of criterion validity?

Concurrent and Predictive

Concurrent Validity

Ability of an index measure to capture a similar outcome to another measure

Predictive Validity

Ability of an index measure to predict a future outcome

Floor effect

Right skew or positive skew

Failure of a measure to detect lower scores for patient who's status has declined

Developers of a hypothetical self-report instrument for patients with amyotrophic lateral sclerosis (ALS) indicate that some individuals reported continued decline in their mobility even though they had scored 100 (maximum disability) on the survey.

Ceiling effect

Left skew or negative skew

Failure of a measure to detect higher scores for patients who status has improved

Authors of a study about outcomes of physical therapist management of plantar fasciitis report that one of the instruments used could not detect continued improvement in five of their subjects (total n = 30).

Selection Threat

Poor selection, sample isn't representative of target population

Solve by multiple study sites and probabilistic sampling

Assignment Threat

Poor allocation of subjects may result in groups groups that are different than each other at onset of study

Solve by probabilistic sampling

Attrition Threat

Loss of subjects, decrease power

Solve by replacement, intention to treat analysis

Maturation Threat

Changes over time can alter outcomes

Solve by control/comparison group, scheduling, baseline measures

Rivalry Demorazlization

"I will show them" they know they are in control group and have bias

Solve by separation and blinding

Compensatory Equalization

untreated group demands to receive a treatment that is the same as or equivalent to the treatment received by another group in the research study.

Solve by masking, different locations

Diffusion Treatment

If subjects have contact, there is risk they may find out what is occurring

solve by masking, separation

Statistical Regression

Outliers that regress the mean over time

Solve by trimming outliers and avg repeated baseline measures

History Threat

Times passes, subjects pick up new activities (affecting study results)

Solve by control and scheduling

Instrumentation Threat

Wrong measurement device

Solve by calibration, protocols, selection of rigorous instrument

Testing Threat

Amount of practice prior to testing can alter results and improvement

Solve by practice sessions and repeated measures

What is the most used central tendency?

Mean

What central tendency measure would you use for extreme outliers?

Median

What central tendency is used for nominal data?

Mode

What is the least informative type of variability?

Range

SD

How much on average the scores deviate from the mean of the data set

Interpercentile Ranges

Data is divided into equal portions to determine how the score relies in relation to others

Coefficient of Variation

Allows comparison of variability in different measures of same phenomena

SD/Mean

Standard Error of Measurement

hypothetical estimate of variation in scores if testing were repeated

Standard error of the Mean

the standard deviation of the sampling distribution of the mean

Standard Error of the estimate

a measure of the scatter of points around a regression line

Type 1 Error

Concludes a difference exists when there is no difference resulting in a false positive

Alpha = Probability of making a type 1 error

Type ll Error

Concludes no different exists when there is a difference resulting in a false negative

Beta= Probability of making a type ll error