Research Flashcards
Internal Consistency
how well a measure measures something i.e. outcome measures consistently measure what is being asked
Cronbach Alpha test
close to 0 = dont correlate
close to 1= correlate
> 90 suggest repetition between items
Test retest reliability
how well an item measures something (performance)
- take berg balance and score 5, next trial you should also score a 5
Reliability measured by
Cohens Kappa (K)
.75 excellent
Removes chance from something happening
Also ICC measure
.75 good
Consistency, Reliability, Validity
Consistency
Reliability
Validity
Validity
How well a test measures what it is intended to measure
- Face- a measure appears what is intended to measure
- less formal than content validity - Content- a measure that covers the broad range of areas within the concept of study
- ex. theatre degree not only covers acting, but stagecrew, lighting etc - Construct- ensures measure is actually measuring what it is intended to measure
- Criterion- GOLD STANDARD
- compares to an estabilished mesaure of the same characteristic or construct
- how does the measure compare to other more estabilished measures
P Value
% that something happens not due to chance
can also show if a finding is significant. usually set at P < .05 for significance. If so, you reject the null hypthesis which states that there will be no chance, so youre saying it is significant or there will in fact be a change.
Confidence interval
Measures reliability- a little better than P value
Ex. 95% CI from 1-5 means 95% chance the value falls within 1-5. Narrower the CI the better
“ estimates of a car- would go to a guy who says costs will be 500-550$ vs 500-1000$”
Effect size
- tells us if a study is important, if it has clinical signifciance not statistical significance
- can measure % of experimental group scores > control group average ( higher the effect size, higher the # >)
Positive likelihood ratio
Sensitivity/1-specificity
> 10 good
5-10 moderate
2-5 mild
1-2 small
Negative likelihood ratio
1- sensativity/specificity
Number needed to treat
number of patients needed for a treatment to be effective
- smaller the number the better
- ex. pill needs to be taken by 2 patients to get a postive result vs 1000 to see a positive result
Iontophoresis indication
heel pain
TENS indication
knee OA
US indication
Calcific tendonitis, RA
Fibromyalgia
11/18 pain points (Bilateral)- both axial and appendicular pain > 3 months
(2) - occiput
(2) - lower cervical c5/7 region
(2) - upper traps
(2) - supraspinatus
(2) 2nd rib (anterior)
(2) lateral epicondyle
(2) Gluteal/ SI joints
(2) greater Troch
(2) knees