EBP P2

Degree to which two or more raters can obtain the same results when measuring the same variable

A sensitivity of 80% means that for those with a condition a test is accurate 80% of the time

A +PV of 60% means that 60% of those with a positive test had the condition

A dichotomous representation of positive and negative for diagnostic interpretation are called cutoff scores

Does the instrument test what it appears to test?

Like specificity & sensitivity, not affected by prevalence and thus easier to generalize across populations/settings

Simple random sampling, systematic sampling, stratified random sampling, disproportional sampling, and cluster sampling

Chi-square; Spearman rank correlation coefficient are examples of:

Sample of individuals is tested on two or more different occasions; used most when raters are not involved, self report

Variables must be able to be observed or manipulated

Difference in population versus sample averages

Tester/rater, the instrument, and variability of characteristic being measured are all sources of measurement error.

Good for larger population; can involve multistage sampling

Used for nominal/ordinal measurements

Numbers indicate a rank order like a Likert scale (very good, somewhat good, neutral, somewhat bad, very bad)

Categories or labels like gender

Most frequent score

Numbers have equal intervals but no true zero (like SAT scores)

Positive

ROC curves represent the balance b/w sensitivity and specificity

Predictive values predict

Probablilty sampling is nonrandom; Nonprobability is random

Used for ratio/interval measurements

D/t chance and can affect a score in an unpredictable way and in any direction

Helps to avoid negatives in deviations to avoid an average variance of 0/close to 0

Ability to measure an abstract construct; how much does it represent the theoretical components of construct

Pick the incorrect PICO pairing

Is the positive square root of variance

An aggregate of people that meet a set criteria to which the data can be generalized

When searching, using 'AND' means that concepts overlap and would be represented by which area in the venn?

The cutoff score is ultimately based on the consequence of incorrect decision but usually closest to top left corner

Specificity and sensitivity

Probability a condition exists prior to performing a diagnostic test; prevalence of the condition in a specified group

Used when there is homogeneity of variance

A standardized normal curve has 66.2% of scores within 1 SD of the mean.

Used when distribution of scores is not normal

LR measure how likely a condition is; determines degree of certainty about presence of condition

When evaluating resources, you should check _____

Level 1 evidence

Divides sum of squares by number of scores since it is impacted by the number of scores

The research hypothesis assumes a relationship; null assumes no relationship

Ottawa Ankle Rules are an example of diagnostic CPRs to determine who needs an XR following acute ankle injury

Which is not a measure of variability?

Pick the incorrect pairing

Effect or response assumed to be caused by/predicted by independent variable

A -PV of 90% means that of those that had a negative test, 90% did not have the condition

Positive skew has bell curve leaning to the right; negative skew to left.

Value at which there are as many numbers below it as above. Divides rank-ordered distribution into two halves

Level 3 evidence

Total number of population divided by # desired in sample. Every nth person then selected. Works well with lists

Sensitivity, specificity, predictive values, likelihood ratios, and pretest and posttest probability test accuracy

A research question should be all of the following except

The optimal measure of central tendency can be based on scale of measurement or shape of distribution

Indicates how many times more likely a positive test will be seen in those w/the disorder than those w/o the disorder

Indicates how many times more likely a negative test will be seen in those w/the disorder than those w/o the disorder

Must have at least two levels in order to compare

Based on assumptions/conditions about parameters of a population from which a sample was drawn

Used when there is homogeneity of variance criteria is not met

Sensitivity

Extent to which items measure various aspects of the same characteristic and nothing else

Level 5 evidence includes expert opinion w/o expert critical appraisal

High value indicates a good test (>5)

Ideal cutoff scores depends on cost of incorrect outcomes/benefit of correct outcomes/importance of S&S

Disproportional stratified sampling involves pulling from strata to ensure smaller strata are proportionally represented

Specificity

Diagnostic accuracy

A high sensitivity and a + test rules in the diagnosis; high specificity and a negative test rules out the diagnosis

Convenience sampling, quota sampling, and purposive sampling

Subjects hand picked based on eligibility. Often used for qualitative studies

Individuals in the sample overrepresent or underrepresent certain population characteristics

Low value indicates a good test (<.2)

Revised probability a condition exists based on the outcome of a test for the condition

Level 4 evidence

Ratio of SD to mean, expressed as a percentage

Correlation coefficients, linear regression, ANCOVA, multiple or logistic regression, factor analysis are examples of:

Sum of set of scores divided by number of scores

Make fewer assumptions about population data

Every individual has the chance of being selected

Partition members into strata and pull from strata to ensure subsets are represented

CV is independent of original measurement units, SD is not

Do the items that make up an instrument adequately sample universe of content that define the variable being measured?

Degree to which an instrument measures what it is designed to measure

Predictable errors occuring in one direction and consistently over/underestimating the true score

Subjects chosen based on availability

Used when distribution of scores is normal

area under the curve of .76 means the test would be accurate 76% of the time

Causes or predicts an outcome of a dependent variable

Subjects chosen based on availability with added stratification

Degree of consistency with which an instrument or rater measures a variable

Degree to which one rater can obtain the same results on multiple occasions when measuring the same variable

A specificity of 30% means that for those w/o the condition, you get a false positive 30% of the time

Numbers have equal intervals and a true zero like weight

Clinical prediction rules quantify contributions of different variables to diagnosis, prognosis, etc. if complex case

When searching, using 'OR' means that concepts intersent or occur separately & are represented by which area?

Data will often be disclosed in terms of distribution, central tendency, and variability

Level 2 evidence

Negative PV

Can be predictive or concurrent. Outcomes of one measurement can be used as a substitute for an established test

Variance can be defined using sum of squares or mean of squares.