kokmk Flashcards
what is a t-test
Compares the actual difference between two mean in relation the variance in the data (standard deviation of the difference between means)
used when 2 groups or sets with normal distribution
if samples come from the same population, we expect their means to be roughly equal (Ho = no effect)
when will the difference between groups be considered significant?
If the probability of getting that difference by random chance is very small
p < 0.05 (a)
What do we want in a T test in terms of hypothesis
we try to reject Ho to make Ha more likely to be true
Ho - no difference
Ha - with difference
What are we interested in a t test where in we are testing 2 groups
Sample mean difference
where standard error is the deviation of the sample mean differences from the population mean difference
What are the types of t-test
Independent t-test - data is from two completely different groups ; independent participatory groups
Dependent t-test - Data is from the same group, and is collected from two groups that are related to each other; data collected is paired
each design has its own pros/cons
What is independent t-tests
measures the significant difference in the means of two groups/categories/levels
E.x - is there a difference between the satisfaction of pre-paid and post-paid
What are assumptions for independent variable t tests
IV must be nominal and dichotomous (categorical at 2 levels)
dependent variable must be interval or ratio
normality
test variable must be homogenous - data is drawn from a single population
no outliers
What is another assumption for independent t tests
Homoscedasticity
known as homogeneity of variance
means that all the variances across the group are the same (since there are no outliers)
use levene’s test - measures the equality of variance (significant output - statistically different / non significant output - statistically the same)
we want levene’s test to not be significant
What if the test variable is abnormal or heterogenous
switch to a non-parametric test to compare independent groups
non parametric tests are alternative statistical tests that we use when data does not meet the assumptions
how does hypothesis testing go for independent samples test
- State the hypothesis
Ho - There is no significant difference
Ha - There is a significant difference - Set level of significance
Alpha level of 0.05 - Solve for test statistic
2 formulas - Make a decision
What are the 2 formulas for independent T test
- Standard error of the difference between means
- t-test/t value
how do make a decision for independent t test
if t obt is greater than our t crit, we reject the null (Ho) and conclude that there is a significant difference (Ha)
using table c and df (degrees of freedom)
what is a degrees of freedom (df)
Number of scores in a sample that are independent and free to vary
how to obtain t critical
use table C
need the df to determine the critical value
if df is not explicitly stated, look for the closest LOWER value
What are the uses for one tailed t test
rarely used now
used for more directional power coated to two tailed test
type 1 error is the same but type 2 is reduced
divide p value in half for one tailed test
What are t test for dependent samples
Aka paired samples t test
there are two groups/samples that are related to one another
dependent samples - researcher pair each score in one sample with a particular score in the other sample
generate related samples to have more equivalence (more comparable)
what are the usual samples in a dependent samples t test
paired situations - before and after an event/paired samples
Time usually becomes the grouping/independent variable
measure of variability will change since paired data counts as one score
What are the types of t test for dependent samples
Matched samples design - 2 separate samples that are not connected to each other (*are men more likely to suppress feelings compared to women)
repeated measures design - same subjects in treatment condition (*happiness scores from the start and end of sem for students)
what are advantages for t test dependent samples
requires fewer subjects
able to study changes over time
reduces influence of individual characteristics and differences between subjects
less variability in scores
what are the disadvantages of t test for dependent samples
factors beside the test may cause participants score to change during the time between
participation in the first test may included the results in the second test
counterbalancing is needed to control time-related problems
what are the characteristics for a dependent samples t test
IV is dichotomous - same subjects are present in both groups during the two occasions
DV is continuous - interval/ratio
Normality
observation for one group should have a corresponding pair in other group
data should be approximately normally distributed
how does hypothesis testing work for dependent samples t test
- State the hypothesis
Ho - no difference bet groups
Ha - Significant difference bet groups - Set the level of signif (Alpha - 0.05)
- Compute - similar to Ind T test but based on the difference of scores rather than the means
What is D in the formula of dependent samples t test
D is the difference between X1 scores and X2 scores of each pair
what is degree of freedom computation for dependent samples t test
describes the number of scores in a sample that are free to vary
subtract 1 because the participant count in one group should be the same in the other
count paired scores as one observation
how do we make decisions for dependent samples t test
find t critical using table C
if t obt > t cri then reject Ho
t value falls critical region, reject Ho and conclude that samples are significantly different
t value falls within area of acceptance, fail to reject and samples are not significantly different
