Stats Flashcards
Effect size
How much better is one intervention/independent variable (US) on effecting the dependent variable (pain)
0.8+ = large
0.5-0.79 = moderate
0.2-0.49 = small
Less than 0.2 = trivial
“2,5,8 is it great”
Reliability
How likely it is to get the same results
Cohen’s kappa coefficient (κ)
0 = no better than chance
<0.4 = poor
0.4-0.6 = fair
0.6-0.75 = good
>0.75 = excellent
“4,6,7.5. Is it the same for you and I”
Likelihood Ratios
Positive = how much to increase your suspicion of a condition based on a + result
Negative = how much to decrease your suspicion of a condition based on a - result
+LR:
>10 = large shift in probability
5-10 = moderate
<5 = small
1 = no change
-LR:
0.1 = large
0.1-0.2 = moderate
>0.2 = small
Type I error
“Backing the loser”
Think there was an effect BUT there wasn’t
Type II error
“Missing the winner”
Thinking there wasn’t an effect, BUT there was
Placebo effect
No therapeutic value, but there is a BELIEF that they will improve
To be effective: needs to look like the real tx, best with outcomes mediated by the brain (pain, fear etc)
Nocebo effect
Negative outcome occurs due to a belief that the intervention will cause harm
Exaggerated if pts have neg tx in the past
Use exclusion criteria or randomization to control
Hawthorn effect
Subjects that know that they are being observed, as part of a research study, tend to work harder than they would have otherwise (I.e. adherence to HEP)
Still generally accepted
Observer effect
People work harder when they’re being watched - or in healthcare, where they respond with more improvement in response to more attention
Control: all subjects get same amount of attention (30 mins of tx)
John Henry effect
A control group that believes they’re at a disadvantage, so they work harder to overcome it (more self tx or seeks other tx)
Control: blind them from control vs experimental group
Pygmalion effect
Expectations of those in authority shape the outcome of their subjects
Aka Rosenthal effect
Blinding clinicians administering tx and assessment
Double blind studies
Both clinicians and participants are blind to study
