Post Hoc Tests 0.0 / 5 ? PsychologyStatisticsUniversityAll boards Created by: marmitesarnyCreated on: 16-05-17 16:45 If you are OK with missing a significant effect? Why? Bonferroni, conservative 1 of 13 Not OK with missing a significant effect? LSD, more liberal (higher type 1 error) 2 of 13 Some would argue that LSD... Should not be used when samples have more than 3 groups, as this increases the type 1 error rate even more. 3 of 13 Type 1 Error is what? False Positive 4 of 13 Turkey's should NOT be used when? There are unequal sample sizes. 5 of 13 Tukey's can be used when? A more conservative approach than bonferroni is required: especially when there are lots of IV levels. 6 of 13 Which tests are designed SPECIFICALLY for unequal sample sizes? Turkey's Kramer, Games Howell, Hochberg GT2, Gabriel 7 of 13 Gabriel has a low error rate until... There is a very large degree of variance in sample sizes. 8 of 13 Which test is designed specifically for unequal variances? Games Howell 9 of 13 Which test cannot be used in the case of unequal variances? Dunnett's 10 of 13 All tests take all groups and compare against another apart from... Dunnet's: takes one group and compares it to all other groups individually 11 of 13 Dunnett's is also unique because...? You can select 1-tailed p-values: easier to detect a significant effect 12 of 13 Dunnett's is useful when you require comparisons to be more specific: individual examination of groups. 13 of 13
Comments
No comments have yet been made