T-test Flashcards
what is the alpha level
the cut off point of the t-test (usually 0.05 or 0.01)
what is the difference between z-test and t-test
if we know the S.D of the population (highly unlikely) = z-test
if we don’t know the S.D of the population = t-test
What are the three different experiment designs
One-sample design
use the mean of a single group and compare it with the public data
> no comparison between groups or between conditions
> sometimes may be hard to have the public data
Between group (independent design) design
use the mean of one group and compare with the mean of the other group
> have comparison
> scores are from each individual = have variance = independent
> external factors may affect the scores given by individuals (e.g., education level)
> cannot study the difference between time
Within group (dependent) design
use the mean of the single group but measures in different timepoints
> external factors minimised
> can compare between timepoints
> no variance
List the formulas for one-sample design and how to reject H0
(Mean of the group - mean of the public) / Standard error
Standard error = S.D/square root of n
S.D = Square root of (Square of sum/df)
then, use the t(empirical) and the df to look for the t(critical)
if the t(empirical) > t(critical) = can reject the null hypothesis
Explain the difference between non-directional and directional hypotheses, what is the important not regarding the t(critical)
Non-directional
Only examining whether there is a difference between the variables, and the direction (i.e., higher/lower) does not matter
(e.g.,: there would be a significant difference between the MEP signals when rotating than at baseline)
the alpha level area applies for both sides (contributing to 2.5% each side)
Directional
Also examining the direction of the variables (i.e., higher/lower)
(e.g., there would be a significantly higher MEP level when rotating the, comparing to baseline)
Check the remaining formulas
what is a flaw of t-test
highly dependent to the number of participants
(i.e., the larger the sample size, the easier to obtain a higher empirical t)
the variance size also affects the t-test result
What are the two effect size interpretations
cohen’s d and percentage of variance explained
formula for cohen’s d
independent: M1-M2/square root of SS1+SS2/df1+df2
repeated: Md/square root of S.D.M
effect size
0.2 small
0.5 medium
0.8 large
formula for rˆ2
rˆ2 = tˆ/tˆ2+df
effect size
0.01 small
0.09 medium
0.25 large
What are the two effect size interpretations
cohen’s d and percentage of variance explained
formula for cohen’s d
independent: M1-M2/square root of SS1+SS2/df1+df2
repeated: Md/square root of S.D.M
effect size
0.2 small
0.5 medium
0.8 large
formula for rˆ2
rˆ2 = tˆ/tˆ2+df
effect size
0.01 small
0.09 medium
0.25 large
What are the three things that have to bear in mind before running the t-test
the observations have to be independent (i.e., no bias, no interference from others)
the sample distribution has to be normal (however, can be overridden if the sample size is large enough)
the sample must have the same variance
how to report the statistics
(M, SD), t(df) = t-test score, p-value, d/rˆ2, one tailed (not mentioned = two-tailed)
