Dear fellow R users:
I am a novice to using R, and especially nlme. I am writing to get some feedback about the accuracy of my analysis. Here is the experimental design:
My experiment investigates spatial perception as a function of the levelness of the ground surface. Participants are asked to observe a target object either on horizontal ground or when placed on a small hillside (5 or 10 degrees of steepness), and estimate the distance using two response methods (operationalized with the Experiment variable in the attached datafile). The basic dependent variable is perceived distance of target objects (Pdist column in the attached datafile). Each participant will observe 5 distances, repeated 3 times.
Here is how I treat my variables:
Subject: nominal, not interested in it, random effect
Trial: ordinal, not interested in it, random effect
Repetition: ordinal, not interested in it, random effect
Experiment: fixed effect, categorical variable
Dist: continuous variable, five values, fixed effect
Angle: the inclination of the ground surface (0, 5 or 10 degrees), continuous variable with three values, fixed effect
Pdist: perceived distance, continuous, this is the dependent variable
Here is a sample of the datafile:
Subject Experiment Angle Trial repetition Dist Pdist
1 0 0 1 1 3.048 3.8735
1 0 0 2 1 3.6576 3.6322
1 0 0 3 1 1.8288 1.91135
1 0 0 4 1 4.2672 5.8166
1 0 0 5 1 2.4384 2.5908
1 0 0 6 2 1.8288 1.49225
1 0 0 7 2 2.4384 2.6416
1 0 0 8 2 3.048 3.61315
1 0 0 9 2 3.6576 4.12115
1 0 0 10 2 4.2672 5.08
1 0 0 11 3 1.8288 2.19075
1 0 0 12 3 4.2672 4.7498
1 0 0 13 3 3.6576 3.7338
1 0 0 14 3 2.4384 2.6924
1 0 0 15 3 3.048 3.2258
2 1 5 1 1 3.6576 3.9243
2 1 5 2 1 1.8288 1.75895
2 1 5 3 1 2.4384 2.667
2 1 5 4 1 3.048 3.5814
2 1 5 5 1 4.2672 6.6802
My model is the following:
m.base01<-lme(Pdist ~ dist*Experiment*Angle+Trial, data=data, random=~ (Trial)|Subject, method='ML')
Here is the output:
Fixed effects: Pdist ~ dist * Experiment * Angle + Trial
Value Std.Error DF t-value p-value
(Intercept) 0.2142267 0.2597276 779 0.824813 0.4097
dist 1.0651591 0.0447145 779 23.821333 0.0000
Experiment -0.0880098 0.3650675 52 -0.241078 0.8104
Angle 0.0062341 0.0399418 52 0.156079 0.8766
Trial -0.0061085 0.0070924 779 -0.861274 0.3894
dist:Experiment -0.0061113 0.0645122 779 -0.094731 0.9246
dist:Angle 0.0001253 0.0070482 779 0.017772 0.9858
Experiment:Angle -0.0617941 0.0572538 52 -1.079300 0.2854
dist:Experiment:Angle 0.0442439 0.0101107 779 4.375942 0.0000
I notice that I have significant effects like dist, and a dist x Exp x Angle interaction. How do you interpret the value of the effects? What does negative and positive mean, and is the value expressed in units of measurement, such as meters for Pdist?
Am I doing anything wrong here? I know, for example, when I compare this output to a plain ANOVA output, I get slightly different significant main effects and interactions.
Any help would be much appreciated!
Alen
----------
Alen Hajnal, PhD.
Department of Psychology
The University of Southern Mississippi
118 College Drive #5025
Hattiesburg, MS 39406
USA
Tel. +1 (601) 266-4617
alen.hajnal @ usm.edu
http://ocean.otr.usm.edu/~w785427/lab.html