Help
Options
Didn't know it?
click below
click below
Knew it?
click below
click below
Don't Know
Remaining cards (0)
Know
retry
shuffle
restart
0:00
Embed Code - If you would like this activity on your web page, copy the script below and paste it into your web page.
Normal Size Small Size show me how
Normal Size Small Size show me how
ANOVA terms
| Term | Definition |
|---|---|
| What is ANOVA? | - analysis of varianca - parametric test, compare means of 3 or more groups |
| differences between ANOVA + t-test | - multiple t-test = incr chances of type 1 error - ANOVA = allows comp btwn groups while mainting alpha = .05 |
| cumulative probability in mult tests | - over mult t tests, singif lvl accumulates |
| type 1 error | - saying there is a signif diff btwn groups when there isnt - rejecting null hyp when it is true |
| capitalising on chance | conducting so many tests w a 0.05 signif lvl resulting in incr likelihood of t1e |
| between groups variance | - how much group averages differ from each other/grand mean - means of groups are diff = greate deg of variation btwn condishes/lvls - treatment effects + error |
| within groups variance | - diff scores within lvls of IV, how much to individs differ from group mean - error ( |
| error | - individ diffs + random factors |
| treatment effects | - diff levels of IV yield diff results |
| total variance | - overall variation observed in data - btwn groups + within groups variation |
| one way between subjects design | - one way = only 1 IV - between subjects = each partip in only 1 lvl/conditiomn |
| assumptions of OWBS ANOVA | 1. lvls of measurement (DV = continuous) 2. independence (each obs ind of others) 3. normality of residuals (residuals = norm distributed) 4. homogeneity of variance (should be homogen of var for all groups) |
| homogeneity of variance | - homogen of var btwn groups - groups being compd have equal variability iwthin dat - spread of scores around mean = approx same across groups - boxplot, resid vs fitted plot, Levene's test |
| Levene's test | - p value should be larger than .05 - find if homogen of variance |
| interps of OWBS ANOVA | - F value - p value (should be less than .001) - effect size (eta square, partial eta square) |
| partial eta squared ηp2 | - eta sq and partial eta sq always have same value - partial = used report effect size in ANOVA - ηp2 = 0.01 > small effect - ηp2 = 0.06 > med effect - ηp2 = 0.14 > large effect |
| effect size | show how much of var can be attr to IV |
| post hoc test - One way between subjects | - comp every grou against each other, see which groups have signif diffs btwn them -ANOVA = does not say which groups specifically have signif diffs - Tukey's honestly significant difference |
| interp post hoc/tukey | - p must be smaller than .05 |
| one way repeated measures design | - participants take part in all conditions - 1 IV |
| diffs between one-way + repeated measures | - BS= simpler, but large var from pers to pers + needs large dample - RM = more economical, make contrasts within each partip BUT carry over, practice, fatigue effect |
| carryover effect | exposure to treatment at one time influences responses in another time |
| assumptions of OWRM ANOVA | 1. levels of measurement 2. Normality of residuals 3. assumption of sphericity |
| sphericity assumption | - variance of the differences between all pairs of conditions are approximately equal - variance of difference btwn 2 lvls of IV should be same as diff btwn any other 2 IVs - if spher violated = spher corrections (GG or HF) |
| Mauchly's test | if p greater than .05 = sphericity assumption is satisfied |
| Greenhouse-geisser correction | - if less than .75, use GG -if greater than .75 use HF |
| Huynh-Feldt correction | - use if GGis greater than .75 |
| Epsilon (sphericity estimate) | - measures how far data is from ideal spher - from 0-1, 1 = no violation, variation of diffs btwn all combs of related groups = equal |
| generlalised effect size | general eta squared - general effect size |
| post hoc test - one way repeated measures | - bonferroni - p < .05 = stat signif diff btwn conditions |
| factorial anova | - examine effects of 2+ IVs (factors) on DV - each IV = 2+ lvls - tests interaction effects (shows whether combined effects differs from individ effects) |
| complete factorial design | all lvls of each IV are paired w all lvls of every other IV lvl |
| incomplete factorial design | - all lvls are not paired w all lvls of every other IV |
| factorial notation | 2 x 2 = 2 IVs, each with 2 lvls 4 x 3 x 3 = 3 IVs, one with 4 lvls, one with 3 and another with 3 |
| condition | - cell - level of IV |
| two way between subjects anova | 2 x 2 factorial design |
| features of factorial designs | - main effects - interaction effects |
| main effect | - effect on signle IV on the DV, irrespectve of any other DV - IV1 affects DV without taking into cons either levels of IV2 - IV1 affects DV across both lvls of IV2 |
| interaction effect | - when effect of 1 variable depends on another variable/lvls pf other IV - eg: effect of caffeience depends on the time of day - no interaction = can talk abt each IVS effect on DV on its own |
| assumptions of 2 way between subjects ANOVA | 1. lvls of measurement (DV = cont) 2. independence of obs 3. normality of resids 4. homogeneity of variance (4 conditions being compd have same variance within dat points), use Levene's test (p > .05 = assumpsh satisfied) |
| interps of two way btwn subjects anova | 3 outputs, all must be below .05 to have significance 1. = main effect 1st IV 2. = main effect 2nd IV 3. interaction effect + partial eta squared |
| post hoc tests 2 way btwn subjects anova | - NOT REQUIARED for 2x2 factorial anova - if more than 2, post hoc can be conducted if signif main effect |
Created by:
melissa.sjolin