stats

stats

1. one anova

   • compares within and between variance
     • f-ratio
   • to report
     • f(b,w)=...MSE=.... P<.05
       • use smallest p value possible
   • compares 2 or more means
   • to compute
     • factor=IV
       • stats >Descriptive and homogeneity
         • explore histograms

2. Post-Hoc

   • BONFERRONI
     • 0.05/n
       • Family wise error rate 1-(1-0.05)xn
   • Tukeys
     • it tests all conditions
       • click post hoc
   • priori
     • contrast
       • for one way anova
         • eg: 2 -1 -1 to opposite signs to compare
           • or 0. 1 -1
             • not interested in 0

3. REPEATED MEASURES ANOVA

   • add, define
     • if significant then paired T tests using Bonferroni adjustments
       • if it is one way
         • 1 Dv ,within participants
   • if it is two way
     • Analyse >GLM>
       • This is to get interaction plus main effect
         • Repeated measures again choose one factor > define > choose that you are interested on for the other level
           • FOR SIMPLE MAIN EFFECT
         • Check sphericity if Maunch is significant use Hyung feldt

4. Two way ANOVA

   • interaction
     • The effect one iv varies according to the other iv
       • compares all cell meas
   • simple main
     • one iv has an effect on the dv in one particular conditin of the dv
       • compares all cell means in one condition
         • to compute
           • analyze
             • GLM
               • univairate
                 • option descriptives = Marginal means
                   • paste syntax
             • for simple main effect
               • add compare to syntax
                 • /EE: eg compare (sex)
                   • look at univariate test
   • main effect
     • an IV has some effect of the Dv
       • compares marginal means for each condition

5. t test

   • uses do compare two means
     • independent
       • compares two unrelated samples
         • between participants
     • paired sample
       • related
         • within participants
     • one sample
       • compare one sample to pop mean
   • assumptions
     • normally distribution
       • Homogeneity
         • if Levene's is significant = not homogenous..
   • t (DF)=2.56 , p , . 05
     • if not significant use actual figure
   • To compute
     • Analyze
       • Descriptives
         • explore
     • compare means
       • paired sample meand
         • within
     • compare means
       • independent t test
         • Between
           • test variable = DV
     • if you want one level only eg specidic role
       • split file
         • organize by group > role
       • select cases
         • if role =1

6. 3 WAY INTERACTION

   • GLM> univariate add 3 factors
     • if significant look for interaction effects
   • L-matrix
     • remember the order
     • test results
     • look at custom Hypothesis test

7. MIXED MODEL

   • Analyze-GLM>repeated measures
     • Within factors add numbers of levels
       • Define
         • add Between
           • see test of within and test of between ? look at each separate error
   • Follow up
     • test interaction between 2 at a level of third factor
     • simple main effect
       • when you want to compare two levels from diff iv eg male patients select cases if gender = 1 if patients = 1
         • if I want to compare eg male patients and female patients
           • split file compare groups and select if role = 1