Lecture 1: Review Flashcards
What is the EBP process?
Formulate a question based on a clinical problem
Identify the relevant evidence
Evaluate the evidence
Implement useful findings
Evaluate the outcomes
What are the 5 types of evidence-based questions?
Efficacy (real-world) of an intervention (PICO)
Usefulness of an assessment
Description of a condition
Prediction of an outcome
Lived experience of a client
Describe common research designs/methods used to form questions regarding the efficacy of an intervention.
Randomized controlled trial
Nonrandomized controlled trials
Pretest/posttest without a control group
Single subject design
Describe common research designs/methods used to form questions regarding the usefulness of assessments.
Psychometric methods
Reliability studies
Validity studies
Sensitivity and specificity studies
Describe common research designs/methods used to form questions regarding the description of a condition.
Incidence and prevalence studies
Group comparisons (of existing groups)
Surveys and interviews
Describe common research designs/methods used to form questions regarding the prediction of an outcome.
Correlational and regression studies
Cohort studies
Describe common research designs/methods used to form questions regarding the lived experience of a client.
Qualitative studies
Ethnography - social and behavioral sciences
Phenomenology - seeks to explain the nature of things through the way people experience them
Narrative
What are the different types of research?
Experimental
Nonexperimental
Quantitative
Qualitative
Cross-sectional
Longitudinal
Basic
Applied
What is experimental research?
Examines cause and effect relationships
- RCT (True experiment Level II)
- Nonrandomized controlled trial (Quasi experimental Level III)
- Pretest/posttest (Pre-experiment Level IV)
In a typical experiment, participants are assigned to one of two groups, and the groups are manipulated.
- One group receives intervention of interest
- Other group is the control group (they may or may not receive an intervention)
EX: Using CIMT on one client and traditional methods on another.
What is nonexperimental research?
Cannot determine causal relationships
Can answer descriptive, relationship, and qualitative questions
- Descriptive question: group comparison or incidence/prevalence design
- Relationship question: correlational or predictive design
Common approaches to collect and analyze data include surveys, interviews, observation of behavior, standardized measures, and existing data from medical records
Observational in nature
What is quantitative research?
Uses statistics
Describes outcomes in terms of numbers
Deductive reasoning; begins with hypothesis and works down to determine if evidence supports the hypothesis
Centered on testing a hypothesis
- Hypothesis is either directional or nondirectional
- Directional: researcher has an assumption or belief in a particular outcome
- Nondirectional: exploratory, no prior notion about the study results but assumes a difference or relationship exists
What is qualitative research?
Answers questions about meaning and experience
Uses inductive reasoning; moves from the specific to the general
Provides a more personal and in-depth perspective of the person or situation being studied
Data collected may include photographs, diagrams, etc.
Encompasses several different designs: ethnography, grounded theory, phenomenology, and participatory action research
Compare quantitative and qualitative research.
Quantitative:
- Tests theory and/or hypothesis; focus is on confirmation
- Outside and objective
- Deductive reasoning
- Quantifiable, typically standardized measures w/many participants
- Descriptive and inferential statistics
- Reliability and validity; Data accurate? Consistent?
Qualitative:
- Builds theory and/or explores phenomenon; focus is on discovery
- Insider, subjective
- Inductive
- Interviews and observations of a few individuals in their natural environments
- Identification of themes using text or pictures
- Trustworthiness: Data believable?
What is cross sectional research?
Data are collected at a single point in time; uses nonexperimental methods; can be observational in nature; descriptive and correlational studies frequently use cross-sectional research
What is longitudinal research?
requires that data be collected over at least two time points and typically covers an extended time period (several years or decades); intended to examine the effects of time (ex: development, aging, or recovery) on some phenomenon (ex: cognition, independent living, or language); many longitudinal studies examine naturalistic changes making them observational
What is basic research?
Used to investigate fundamental questions that are directed at better understanding individual concepts
- background questions
What studies are typically neither cross-sectional nor longitudinal?
intervention studies
A simple pretest/posttest is not considered ___.
longitudinal study
What is applied research?
Has direct application to health care practices
What is translational research?
Both basic and applied together; findings form the laboratory are used to generate clinical research
The National Institutes of Health’s (NIH) National Center for Advancing Translational Sciences has a mission to promote more translational research
What is hypothesis testing?
Researcher has to decide on whether to accept or reject the research hypothesis based on the p value obtained from the statistical analysis.
When is a hypothesis accepted? rejected?
If p is less than or equal to 0.05, the hypothesis is accepted.
If p value is greater that 0.05, the hypothesis is rejected.
What mistakes occur when interpreting the results?
Type I errors: when the hypothesis is accepted, yet the hypothesis is FALSE
Type II errors: when the hypothesis is rejected, yet the hypothesis is TRUE; sometimes sample size is too small
What is a variable?
Characteristics of people, activities, situations, or environments that are identified and/or measured in a study and have more than one value
What is an independent variable?
These are manipulated or compared in a study; with more than one independent variable is included in a study, the study is described as a factorial design
What is a dependent variable?
These are observed and are intended to measure the result of the manipulation (also know as the outcome or outcome variable)
What are control variables?
Those variables that remain constant; the more control in place, the more confidence that the independent variable caused the change in the dependent variable
What are extraneous variables?
Variables that can influence the outcome of a study; they are tracked and then later examined to determine its influence
Can be natural characteristics of the participant, such as age or gender, or they could be features of the environment such as noise or lighting.
Includes: demand characteristics, experimenter effects, situational variables, and participant variables
What is a confounding variable?
It’s a type of extraneous variable that not only affects the dependent variable, but is also related to the dependent variable.
What are descriptive statistics?
Describe the data in a study
Provide an analysis of data that helps describe, show, or summarize it in a meaningful way that patterns might emerge from it
Include:
- Frequencies (actual number or count along with a percentage)
- Frequency distribution (how often something occurs within a given interval) on line graphs, histograms, etc.
- Measure of central tendency (mode, mean, median)
What is normal distribution?
Data points are distributed in a symmetrical, bell-shaped curve; two halves are mirror images
What is skewed distribution?
lack of symmetry in the spread of scores, such that the curve is longer than the others
What does positively skewed distribution look like?
skewed right
What does negatively skewed distribution look like?
skewed left
Symbol for mean of a sample
M, x̄ (X Bar)
Symbol for mean and standard deviation of a sample
M (sd)
Symbol for standard deviation of a sample
S, sd, σ
Symbol for sample variance
s^2
Symbol for number of participants in a study or number of participants in a group in a study
N, n
What is inferential statistics?
The are the tests of difference. This type of statistic “infers” which means it concludes/suggests from evidence and reasoning , something about a larger population based on a sample used in a study.
Inferential statistics are often divided into what two categories?
Tests of difference (t-tests and analysis of variance [ANOVA])
Tests of relationships (correlations and regressions)
What are the 2 subcategories of inferential statistics?
parametric and nonparametric statistics
What is parametric statistics?
assumes that the distribution of scores in the sample are relatively normally distributed; uses mean and standard deviations
Nominal
What is nonparametric statistics?
these are considered distribution free; compares categorical and rank-ordered data
Ordinal, Interval, Ratio
Compare/contrast p-values and critical values.
They both do the same thing: enable you to support or reject the null hypothesis in a test. But they differ in how you get to make that decision. In other words,they are two different approaches to the same result.
What is statistical significance?
This is a number that expresses the probability that the result of a given experiment/study could have occurred purely by chance.
Statistical significance is highly dependent on sample size.
What is level of significance?
Level of significance (alpha level = α) – is the amount of risk you are willing to assume; typically the standard level of significance is 0.05. This means there is a 5% risk that the difference between two groups, or the differences in scores from pretest to posttest, is NOT a true difference but instead occurred by chance.
When is α level determined?
The α level is determined BEFORE the study begins and is typically set at 0.05. When statistical significance is reported in a study, it is reported as a p value.
What are the most common statistical tests used for comparing differences in parametric statistics?
t-test and analysis of variance (ANOVA)
What is an independent sample t-test (unpaired sample)?
Compares differences between 2 groups
Ex: Attention span of boys vs attention span of girls
What is a dependent sample t-test (paired sample)?
Compares differences within a group at 2 time points on a single dependent measure
Ex: Attention span of boys vs attention span of girls before PE and after PE.
What is an ANOVA?
Compares differences between 3 or more groups on a single dependent measure
What is a repeated measures ANOVA?
Compares differences within 1 group at 3 or more time points
What is a mixed-model ANOVA?
compares both between-group and within-group differences simultaneously (interaction effect); provides separate results of the between-group differences and within-group differences (main effects)
What is an ANCOVA?
Compares differences between and/or within groups while statistically controlling (equalizes groups) a variable (the covariate)
What is a chi-square?
Nonparametric test; compares frequencies of 2 or more groups
What is a Mann-Whitney?
Nonparametric test; compares 2 groups using rank-order data
What is a Kruskal-Wallis?
Nonparametric test; compares 3 or more groups using rank-ordered data
What is a p-value?
Probability is how likely something is to happen.
The p value is related to Type I error in that it indicates the probability that a Type I error occurred.
Describe how p-values are used.
In a research article, a researcher may report an exact p value as calculated, for example p = 0.035 or they may report p ≤ 0.05. In both cases, the p value indicates statistical significance.
- In the case of p = 0.035, they are saying the probability that the results occurred by chance is 3.5%
- In the p ≤ 0.05 reporting, they are saying that the results occurring by chance is less than or equal to 5%
Statistical significance (does/does not) equate to clinical significance.
Does not
In addition to statistical significance, what other types of statistics are useful in interpreting results of a study?
Effect size (ES)
Confidence intervals (CI)
Power
What 3 things affect the power of a study?
The alpha level set for the test of statistical significance
The magnitude of the effect
The sample size of the study
Symbol for alpha level
Level of significance that is used to determine if an anlysis is statistically significant
α
Symbol for probability value
The calculated likelihood of making a type I error
p
Symbol for degrees of freedom
Number of values that are free to vary in a statistical analysis based on the number of perticipants and number of groups
df
Symbol for critical value in a t-test
t
Symbol for the critical value in an ANOVA test
F
Symbol for the critical value in a chi-square
x^2
Symbol for effect size
ES
Symbol for cohen’s d
Effect size statistic
d
Symbol for Eta squared
Effect size statistic
η2
Symbol for omega squared
Effect size statistic
ω2
Symbol for confidence interval
CI
