Module 10 Flashcards
how small does a p-value have to be before researchers feel results are not chance
p value of 0.05 or less is the cut-off point for researchers to have confidence in studies
What is a P-Value
P-value assesses the probability that the results of the a study are due to chance or random error
for ex; a p-value of 0.05 tells us there is a 5% possibility that the results are due to random error or chance.
Statistically signilficant
when the p value is less than 0.05
this means there are less than 5% probability 5/100 that the observed results were due to chance
What three things influence the P-value
Magnitude of the effect (how much change in outcome actually occurred) Sample size (how many people were in the study) Variability (how different were the people in the study from one another)
Magnitude of Effect
As the magnitude gets of effect gets larger, we become more confident the results are not due to chance alone (p-value gets smaller)
Those with the greatest change… have the lowest p-value
true
P-value and Sample Size
So as sample size increases….
If you flip a coin 10 x and it lands on heads 7 times are you confident?
The larger the sample size, the more confident you can be in the study’s results.
p- value decreases
P-value and variability
as the variability gets smaller we become more confident that the results area not due to chance.. p value gets smaller
Can you trust a study based on p-values?
No… must still consider bias, relevance of study, critically appraise the study methods
and if study is clinically important.
what is power?
in studies about treatment, power is the ability of a study to detect a statistically significant result if one exists.
- Power depends on two things:
2. Where can I find info about power in a research article?
- Sample size: were there enough subjects
Magnitude of effect: was the size of the result large enough. large magnitude, more statistical power - Methods section
How does p-value and power work together?
Power=the abilitly to FIND the significant result
p-value = the CONFIDENCE to know the result is not due to chance
p-value is found where?
power is found where?
Results
methods
confidence interval = range of effect
overall effect
range within true effect really lies
range of effect
confidence interval
cyber bullying example
true effect 10% - 70% reduction
95% confidence level
low end of range might not make it worth it
if confidence level crosses the 1 line?
-90% +10%
0% = the line of no difference
90% crosses the line, no better on either side of the line
The study does not suggest an effect.. We don’t have enough information or evidence. We would need a more narrow confidence interval(which would require greater statistical power)
confidence intervals replace p-value
true
clinical significance
pushes us to assess whether or not size of effectis big enough to justify the investment
95% confidence level
creation of bike paths
led people to increase acti to 30 min /wk
confidence interval true effect 10 - 50 minutes 95/100
bike path example
upper limit of CI 50 min increase
lower limit of CI 10 min increase . not impressive
assessing clinical significance…
A CI of -90 to +10%
It is a 95% CI meaning we are 95 % confident that the difference falls w/in that range
