Reasons for Research
Increases professional base of sci knowledge
demonstrates professional credibility
justify cost of services
justify necessity of treatment and treatment outcomes
demonstrates benefits of treatment to patient
Sources of knowledge
includes trial and error, traditions, authority, logical reasoning (deductive and inductive reasoning)
quantitative and qualitative research
deductive reasoning
acceptance of general idea and inferences drawn to a specific case.
a major and minor premise, and conclusion
if premises are true then the conclusion must be true
usefulness fully dependent on truth of premises
inductive reasoning
develops generalizations from specific instances
basis for problem solving and common sense
quality of knowledge derived from inductive reasoning is dependent on representativeness of the specific observations used as the basis for generalizations.
To be
Quantitative Research
a type of research that studies how variance in x cause's variance in y.
involves measurement of outcomes using numerical data under standardized conditions.
Advantage: ability to summarize scales and to subject data to statistical analysis
Qualitative Research
concerned with deep understanding of phenomenon through narrative description.
Measurement = open ended questions, interviews, observations
evidence based practice
evidence from clinical research is emphasized in the decision making process
develop clinical question (PICO)
search for best available research evidence
critically appraise research evidence
integrate research
evaluate
PICO
specific Population
specific Intervention
Comparison of what already exists
Outcomes of Interest
Clinical Question
prompts literature search, focuses on particular patient, determines solutions to clinical problems, usually ends with a solution
Research Question
guides scientific inquiry, focuses on a certain population, obtains knowledge generalizable beyond individual situations, and usually ends with more questions.
Randomized Controlled Trial
Gold standard of research = true experimental designs
Random allocation of subjects
Blinding
Minimize bias
Limitations
Systematic Review
method to access and interpret evidence to inform clinical practice
usually RTC
predetermined method how to do this
Systematic Review process
State study objective > develop protocol > develop search strategy > conduct search > retrieve relevant papers > eval methodological quality of studies > analyze and synthesize finding > determine if statistic data are sufficient > report the results of S
Meta-Analysis
a statistical synthesis to combine results that are similar
Descriptive statistics
uses shape of sample, positive skew, non symmetrical, central tendency and variability to analyze use of data and describe its meaning
sample shape
the distribution of values. Shown in frequency, histograms and stem leaf
visualizes skew and symmetrical stats.
Positive skew
tail goes off the right; means mean is on the left. Left skew tails off to the left.
Skewed means are less stable because the outliers pull the mean up or down and are less likely to represent the true distribution because biased by outliers.
When the mea
Non Symmetrical
means mean, median and mode aren't the same
Central Tendency
how the data is clustered in the sample. Mean, median, mode
used to summarize and describe the typical nature of the data and center of the distribution
Variability
the dispersion of the sample. Variance, standard deviation, range and confidence interval
Frequency Distribution
table or graph of rank ordered scores. Allows for assessment of distribution, range and clustering of data
nominal data
% listing
Mean
? = ?x/N or sum of numbers over number of values
Median
middle value when scores are arranged in order of increasing size. The 50th percentile midpoint
Mode
score that occurs most frequently
Range
Xhighest - Xlowest. Bigger range means skewed data and implies outliers
Standard Deviation
Sx = ?(?(x-x?)2/n-1) square root of variance. Provides a measure of variability of scores
Normal Distribution
data falls in a symmetric bell shaped curve indicating distribution of scores.
W a normal curve the mean, median and mode approximate each other
� 1 SD = 68%
� 2 SD = 95%
� 3 SD = 99%
Confidence Interval
an interval estimate of a population parameter.
estimates are better than point estimates of pop mean due to sampling error
Range of scores w/ boundaries or confidence limits that should contain the population mean. Boundaries of CI based on sample mean a
Confidence Level
a statistical estimate of how close to the population value a sample of particular size is supposed to be.
Expressed as probability %
x = z(s/sqrt n)
Larger samples have smaller interval, smaller samples = larger interval
Parametric Data
used to estimate population parameters with dependency on nature of the data
Random selection
Normal distribution
= sample variance
Interval or ratio data
Non-parametric Data
when statistical conditions do not meet the requirements for parametric.
Small samples
Not normally distributed
Unequal variance (two groups can differ greatly)
Nominal or ordinal data
Independent Variable
manipulated variable or cause. What you change to affect dependent variable.
Dependent Variable
measured variable or effect or outcome.