Standard deviation & T-test Flashcards
Purpose of student t-test?
- continuous data
- investigating a diff between 2 means
- judges the significance in the diff between 2 sample means
Standard deviation?
a measure of how spread out all your repats r from your mean
- the spread of data ab the mean
Advantages of using Standard deviation instead of range?
- SD shows spread of data around mean whereas - range is only the diff between highest & lowest value
- SD reduces effect of anomalies whereas - range includes anomalies
- SD be used to indicate whether a diff between results is significant
Why is knowing standard deviation more useful for conclusion?
If: when u + or - SD from means there’s no overlap - ts indicates diff is likely to be sig
*sd isn’t a full statistic so - doesn’t tell u u can be 95% confident (can’t be w/o use of statistic)
if: when u + or - SD from means there is an overlap - ts indicates diff is likely not to be sig between those 2 means
Conclusions using SD?
- & - SD from mean (to see highest & lowest variation compared mean)
- look for overlaps of SDs between higher & lower ends of Group A & B (if higher end>lower)
- whether ts indicates likely to be/not be a sig diff between… of A & B
Null hypothesis?
- aka H0
- there is not a statistically significant difference between the two means and any difference is due to chance
Alternative hypothesis?
- aka H1
- there is a statistically significant difference between the two means, the difference is not due to chance
P value?
- tells u the probability that result is due to chance
- in bio sig level of 0.05 (5%) required - means we want atl 95% confidence that diff between 2 means is statistically significant/only 5% probability that difference could be due to chance
- aka the p value - p=0.05
Meaning of t equation?
S = standard deviation
X = mean
N = number of samples
Paired v unpaired T test?
Paired:
- Df (degrees of freedom) = n-1
- if 2 sets of data r from same individual so both groups hv same sample size
Unpaired:
- Df = n1 + n2 - 2
Critical value table?
- once hv Df - read across and compare it to p value of 0.05
- does calculated T value exceed or not critical value
- if exceed - p value < 0.05 so - prob that diff between 2 means is due to chance less than 5% so - reject null hypothesis aka is statistically sig diff
T-test conclusion?
- accept or reject null hypothesis?
- is there more or less than 5% probability that diff is due to chance
