tests of difference (Parametric) Flashcards
what are parametric
parametric tests are more powerful and robust than other tests
If the researcher is able to use a parametric they should do so as these tests may be able to detect significance within some data sets that non - parametric tests cannot
what makes parametric tests unique
Parametric tests require three specific criteria:
- data must be interval level- parametric tests use actual scores rather than ranked data
- the population must show the normal distribution for the variable being measured
If the distribution of the data is skewed a parametric test is not appropriate - There must e homogeneity of variance meaning the set of sores in each condition must have a similar dispersion spread - we can use the standard deviations to check this
when do we use an unrelated t-test
the unrelated t-test is a test looking at the difference between two different groups
(independent group design/unrelated)
The level of measurement required is interval data (we know as we are using a parametric test))
- data must be normally distributed - this usually means a large sample needed
- there should be homogeneity of variance
how do you interpret the results of an unrelated t-test
a value of t is calculated from the statistical test (t)
if the results are significant the calculated values of t is MORE than or EQUAL TO the critical value of t, we can reject the null hypothesis and conclude that there is a difference
how do you find the critical value from the table of values
you must work out the degrees of freedom:
DF=(NA+NB)-2 (number of participants in group A, plus the number of participants in group B - 2)
the significance level you are testing
knowledge whether it is a one-tailed or two-tailed test
how do you report the results of unrelated t-test
REPORTING WHEN RESULTS ARE SIGNIFICANT:
“As the calculated value of t( ) is more than or equal to critical value t( ) (df = p<0.05) the result is significant and must reject the null hypothesis. There is a difference between - and -.”
“As the calculated value of t( ) is less than the critical value t( ) (df = p<0.05) the result is not significant and must accept the null hypothesis. There is a no difference between - and -.”
when do we select a related t-test
the related t-test is a test looking at the difference between two different groups (repeated measures or matched pairs i.e. “related design”)
the level of measurement required is interval data (we know this because we are using a parametric test)
-data must be normally distributed - this means a large sample is needed
-there should be homogeneity of variance - the standard deviation for both group will be similar
how do we interpret the results of related t-test
a value of t is calculated from the statistical test (t)
A SIGNIFICANT RESULT: if the calculated value of t is more than or equal to the critical value of t we reject the null hypothesis and conclude there is a difference
how do we find the critical t - value from a table of values
you need to work out the degrees of freedom
DF=N-1 (number of data sets minus 1)
The significance level you are testing to
knowledge of whether it is a one or a two-tailed test
how do we report the results of the related t-test
“As the calculated value of t( ) is more than or equal to critical value t( ) (df = p<0.05) the result is significant and must reject the null hypothesis. There is a difference between - and -.”
“As the calculated value of t( ) is less than the critical value t( ) (df = p<0.05) the result is not significant and must accept the null hypothesis. There is no difference between - and -.”
