Search
Create
Log in
Sign up
Log in
Sign up
Measurement, data management, and systematic review
STUDY
Flashcards
Learn
Write
Spell
Test
PLAY
Match
Gravity
Terms in this set (76)
what does it mean to be a healthcare professional?
...
what is evidence based practice?
...
purpose of a systematic review
-to gather together, evaluate and summarize the published research addressing all that we know about a specific topic
-need to be very specific about how the article are selected
what are the steps for a systematic review?
1. problem formulation
2. data collection
3. data evaluation and coding
4. analysis and interpretation
5. reporting the results
reporting the results
-key aspects of all of the above stages should be clearly stated in the final report in order to allow replication and critical appraisal of the analysis
-the process of reporting should be rigorous and explicit
-methodological limitations of original studies and the analysis should be highlighted and any recommendations should be followed by practical and evidence-based advices
-proposal for future research should be included
statistical procedures
-compiling the data for quantitative synthesis and summary effect sizes by using appropriate methods and effects models should be clearly stated
-effect models should explore the sources of variation if variability is present
ex: differences in study quality, participants, treatments, or outcomes
analysis and interpretation
-statistical procedures
-interpretation of results
data evaluation and coding
-study quality assessment
-data identification and qualification
-characteristics of interest
study quality assessment
assessment of the methodology quality and validity of studies
data identification and quantification
-outcome variables are identified and extracted into a coding system
-group's contrasts and effect sizes calculations are performed for meta-analysis
characteristics of interest
-general information regarding study/trial design should be standardized into a coding system for further analysis
-such as treatment and sample characteristics
inclusion criteria
-defined and eligibility criteria are set
-construct definition to distinguish relevant from irrelevant studies
-types of participants: dx, age, time since occurrence, severity level
-types of interventions
-types of outcome measures
grey literature
unpublished studies or those available through alternative sources
data collection
-develop a search strategy: comprehensive, sensitive, and extensive search strategy to compile possible reports
-key words and variables determine sources of potentially relevant reports
-can use grey literature
problem formulation
-a topic is selected followed by a formulation of a research question relevant to the topic of interest
-it includes identification of the problem and formation of the research questions
deductive theory
-use a theory to create a hypothesis then observe to confirm hypothesis
-top down approach
inductive theory
-observe, see a pattern, create a hypothesis and find a theory
-bottom up approach
primary source
account of research by the researcher
secondary source
-account of the research by others
-books, review articles
direct measures
-variables that can be directly observed
-most rely upon surrogate measures rather than the actual attributes of interest
-constructs: assorted traits chosen to represent an otherwise abstract phenomena (ex: intelligence, health, QoL, pain, fatigue, etc.)
example of direct measure
height, ROM
continuous variable
-can be any value along a continuum within a specified range
-measurement can include decimals (fractions of whole numbers)
example of continuous variable
height in cm, time in sec
discrete variable
-limited to whole units (numbers)
-subtype: dichotomous
example of discrete variable
number of children, number of oranges
dichotomous variable
when a variable contains only two discrete levels (0 vs 1)
example of dichotomous variable
gender, Parkinson's diagnosis, yes or no
types of variables
-continuous
ratio, interval
-discrete
ordinal, nominal
what is a surrogate measure?
...
what is an example of surrogate measure?
...
what is a construct measure?
...
what is an example of construct measure?
...
nominal variable
-"named" category
-broadest category of data
-categories of objects or subjects are assigned different labels
-categories are both mutually exclusive and collectively exhaustive
-if numbers are used to label categories, the order is arbitrary and does not reflect the magnitude of the characteristic
example of nominal variable
gender, race, political affiliation, hair color, religion, dominant hand
ordinal variable
-ranked categories based on relative magnitude of the characteristic
-the differences between categories are not consistent (quantifiable) so the values are not meaningful as true quantities
example of ordinal variable
military rank, low-medium-high, manual muscle test
interval variable
-values with known and equal distances (intervals) between units and 'arbitrarily' assigned zero point
-illustrates numerical differences between scores, but not absolute magnitude
-can add and subtract values and calculate means and variance
*cannot compute or compare ratios of scores
example of interval variable
temperature, years on calendar, sea level, elevation
ratio variable
-interval scale with absolute zero
*negative numbers are impossible
-zero indicates the complete absence of the characteristic, but is often a theoretical value that cannot be measured
-can perform all mathematical and statistical processes
example of ratio variable
height, weight, percentiles, t scores, quotients
when can data be translated to a different level?
-data can always be reduced to a lower scale-at the expense of precision
*cannot be transformed to higher scale
observations
-gather subject characteristics
-measure behavior
-identifies environmental characteristics
guidelines for observations
-remain unbiased
-consider whether some other issue or situation may influence responses
-follow the data gathering protocol
-if you deviate from the guide document the reason for deviation
-answer every item
interviews
-a very common method of gathering information in OT
-types of information gathered:
demographics, health status, medical history, social support, skills and abilities, living arrangement, vocational history, attitudinal and psychosocial data
-can be structured, semi-structured or unstructured
-self reported or proxy reported
standardized tests
-testing and scoring procedure
-a lot of performance tests
-can be norm referenced or criterion referenced
norm-referenced
-linked to a sample
-a score related to the average person in the group
criterion-referenced test
-linked to the outcome
-a fixed standard or acceptable level of performance
record review
-type of data gathering that involves going into documents to find needed information
-primary vs secondary data
what does level of evidence mean?
...
what is level 1 evidence?
...
what is level 2 evidence?
...
what is level 3 evidence?
...
what is level 4 evidence?
...
what is level 5 evidence?
...
descriptive statistics
-summarize sample characteristics
-screen data for inaccuracies prior to analysis
-reduce or summarize data by describing the general distribution of obtained scores
example of descriptive statistic
-average age 80.2 yrs, 91% white, mean FIM 72
-frequency/percentage
-central tendency
-variability
normal distribution
-symmetrical, bell-shaped curve
-shows mean, median, and mode as central and equal
-many population characteristics (biological, psychological and social) follow normal distribution
-ability to draw conclusions from a sample statistics
-mean is between middle 68%
-1 SD from mean = 13.6%
probability
-likelihood that any one event will occur, given all possible outcomes
-chance of event divided by total possibilities
example of probability
coin flip, dice, or normal distribution
p value
-shows the extent of the evidence against H0
-shows the probability that results occurred by chance if H0 is true
-the difference between two estimates occurred by chance, if the estimates being compared are really the same
-significance is shown on a range from 0.0-1.0
-high values indicate that data is compatible with H0
-low values indicate that data is improbable if H0 is true
-if P is less than the alpha threshold, data is considered statistically significant and reject H0
sampling error
-difference between sample mean and true population mean
-is unpredictable because it occurs by chance; resulting in various sampling errors
-can be used interchangeably with chance error
-denotes the relationship between the error and sample
-there will always be a sampling error no matter how perfectly the research is done
standard error of mean
-margin of error
-standard deviation of a theoretical sampling distribution of means
-estimate of the population's standard deviation and population mean
-considers that the sample standard deviation varies from the population as a function of sample size
SEM = SD / square root of N
central limit theory
-for any population with mean and SD, the distribution of sample means will have a mean equal to the population mean and SD equal to the population mean divided by the square root of the sample size
-SD = pop mean / square root of N
sampling distribution
-researcher draws a sample from a larger population
-the assumption is that there are numerous samples of a given size (n = subjects)
-the sample drawn is assumed to be from a broader group of all possible random samples of a particular sample size (n = subjects)
-all possible means
-from the sample a sample mean and deviation can be determined
variability
-range
-percentiles
-variance
-standard deviation
range
difference between highest and lowest scores
percentiles
describes individual score within distribution (100%)
-quartiles: 25, 50, 75% of distribution
-50th percentile is median
-Q3-Q1 = interquartile range (middle 50s%)
variance
-s^2
-measure of variability or dispersion in scores, equal to the squared deviations between the mean and observed score
-has an adverse effect on statistical power-greater the variance, the lower the statistical power (less likely to reject the H0)
standard deviation
-s
-square root of variance
-returns variability to original units
-small SD = small spread with high certainty/precision
-high SD = high spread with high uncertainty
frequency
-number of times that scores occur
-often reported as % total or cumulative %
-typically grouped (range) when reporting on large sample with continuous data (identical scores not regularly observed)
-graphing frequency distributions
central tendency
-point estimate (single value)
-uses mean, median, mode
example of frequency
pie chart, stem & leaf plot, histogram, frequency polygon
inferential statistics
-estimating population characteristics from sample data
-generalizing sample observations to population
example of inferential statistic
older minority patients have significantly longer lengths of stay in rehabilitation following hip fx than older white patients, P<0.05
population
...
sample
...
;