Statistical Testing Flashcards
What does standard deviation show?
How much variation there is between the data and the mean value
How consistent the data is, unlike range which only shows biggest and smallest
What does each part of the standard deviation equation stand for?
s is standard deviation
x refers to each measured value in data set
x¯ (line is normally over the x) refers to arithmetic mean
n is number of values measured
∑ = the sum of, means add together the values indicated to right of symbol
What is an unpaired t-test?
Compares the means from different individuals (eg treated and untreated)
Looks to see if there is a significant difference between two sets of data, comparing means to see if different.
Tells if difference is random, due to chance or not
Gives probability
Example of null hypothesis compared to hypothesis
Hypothesis: chickens fed maize supplemented by lipid produce more male offspring than those fed maize alone
Null Hypothesis: There is no difference between the number of male and female offspring of chickens fed maize supplemented by lipid and those fed maize alone
What does each part of the unpaired test equation mean
x1¯ (line is normally over the x1) is the mean of sample 1
s1 is the standard deviation of sample 1
n1 is number of individuals in sample 1
x2¯ (line is normally over the x2) is the mean of sample 2
s2 is the standard deviation of sample 2
n2 is the number of individuals in sample 2
When do you accept or reject the null hypothsis
when t>= critical value, reject the null hypothesis as significant different present
when t < critical value, accept the null hypothesis as not a significant difference
What is a critical value?
The value corresponding to a given significant level
Cut off value determines boundary, leads to decision not to reject null hypothesis
What is a degree of freedom, and how is it calculated?
The number of independent pieces of information used to calculate a statistic
(n1+n2) - 2
What is probability level?
Probability of rejecting null hypothesis when it is true
Normal standard is P=0.05 (5% uncertainty/risk), as still 95% confident that choice is right
In certain trials, P=0.01, as 1% risk of being wrong but 99% certain of correct result
Lower P in trials such as drugs/medical trials