Statistical Testing
Statistical Testing = Provides a way of determining whether hypothesis should be accepted or rejected; tell us whether differences or relationships have
happened by chance – A significant result means the researcher can reject the null hypothesis
Probability Levels Sign Test:
0.05/5% = If hypothesis was accepted, there is less Need to convert to nominal data Critical Value
than 5% probability results occurred by chance; = Calculated must be EQUAL TO 1. Significance level
researcher is pretty certain difference was due to OR LESS THAN the critical value to 2. Number of pps
manipulation of IV be significant 3. One-Tailed (Directional) or
Best balances type 1 and 2 errors Two-Tailed (Non-directional)
0.01/1% = Stricter level if researchers need to be Test of Difference Test of
more confident that results were not due to chance Correlation
Human cost e.g. drug trials OR one off trials Unrelated: Related:
Independent Repeated - Chi-Squared:
Groups Measures/Matched = Calculated must be EQUAL TO
Pairs OR MORE THAN the critical value
Nominal Data Chi- Squared Sign Test Chi-Squared to be significant
Mann Whitney and Wilcoxon: = Data in form
= Calculated must be EQUAL TO of categories
OR LESS THAN the critical value to Ordinal Data Mann Whitney Wilcoxon Spearman’s Rho Spearman’s Rho and Pearson’s R:
be significant = Data ordered = Calculated must be EQUAL TO
in some way OR MORE THAN the critical value
Interval Data Unrelated t test Related t test Pearson’s r to be significant
= Numerical
scales
Errors
Type 1 (False Positive) = When null is Unrelated and Related T Tests:
rejected but in reality, it is true and should = Calculated must be EQUAL TO
have been accepted OR MORE THAN the critical value
TIP –
to be significant
If the test name has an R in it, the calculated value
Type 2 (False Negative) = When null is must be equal to or MORE than the critical
accepted but is reality the alternative is true
and should have been accepted If the test name does not have an R in it, calculated
More likely if significance level is 1% value must be equal to or LESS than the critical
Statistical Testing = Provides a way of determining whether hypothesis should be accepted or rejected; tell us whether differences or relationships have
happened by chance – A significant result means the researcher can reject the null hypothesis
Probability Levels Sign Test:
0.05/5% = If hypothesis was accepted, there is less Need to convert to nominal data Critical Value
than 5% probability results occurred by chance; = Calculated must be EQUAL TO 1. Significance level
researcher is pretty certain difference was due to OR LESS THAN the critical value to 2. Number of pps
manipulation of IV be significant 3. One-Tailed (Directional) or
Best balances type 1 and 2 errors Two-Tailed (Non-directional)
0.01/1% = Stricter level if researchers need to be Test of Difference Test of
more confident that results were not due to chance Correlation
Human cost e.g. drug trials OR one off trials Unrelated: Related:
Independent Repeated - Chi-Squared:
Groups Measures/Matched = Calculated must be EQUAL TO
Pairs OR MORE THAN the critical value
Nominal Data Chi- Squared Sign Test Chi-Squared to be significant
Mann Whitney and Wilcoxon: = Data in form
= Calculated must be EQUAL TO of categories
OR LESS THAN the critical value to Ordinal Data Mann Whitney Wilcoxon Spearman’s Rho Spearman’s Rho and Pearson’s R:
be significant = Data ordered = Calculated must be EQUAL TO
in some way OR MORE THAN the critical value
Interval Data Unrelated t test Related t test Pearson’s r to be significant
= Numerical
scales
Errors
Type 1 (False Positive) = When null is Unrelated and Related T Tests:
rejected but in reality, it is true and should = Calculated must be EQUAL TO
have been accepted OR MORE THAN the critical value
TIP –
to be significant
If the test name has an R in it, the calculated value
Type 2 (False Negative) = When null is must be equal to or MORE than the critical
accepted but is reality the alternative is true
and should have been accepted If the test name does not have an R in it, calculated
More likely if significance level is 1% value must be equal to or LESS than the critical
