I am running my data using Polynomial Regression. The variables are all measured on 6-point Likert-type scales. I run Polynomial Regression in SPSS and and I center the predictors at 3.5 (which is the midpoint of 6-point Likert-type scales). The mean values of all variables are ranged from 3.8-4.2. However, my constant value is 5.78 and it seems to be too high as from my understanding constant value helps to predict the mean value of the outcome variable when the predictors are 0. The constant value is 4.2 when I didn't add the control variables.

Need help in Polynomial Regression : constant value and control variables

My questions are:

1. From the graph above, the highest point to plot on Z-axis is 11.67. Since my variables are all measured on 6-point Likert-type scale, is this abnormal?

2. What does high constant value imply? I did double check my SPSS syntax and every step of the progress, yet still unable to get any insight

3. Is there any special way to deal with control variables in Polynomial Regression? The way I do it is just adding them with independent variables

I urgently need answers for these. Please help. ]]>

I am currently running an experiment where we measure escape latency and seeing if the presence of a drug has an effect on this. Each rat has 100 trials, which we organized into blocks of 10. Rather than throwing out the data on individual variances by averaging the data together, I was wondering how a multilevel model would be organized for this in an R script since I haven't done one before.

Right now I have the data in long format (columns being Drug, Subject, Block, Trial, Latency). The DV's and IV's seem pretty straightforward but I'm not quite sure how to format this using lmer.

So DV = latency, IV's = Drug (0 or 1), Subject (1-18, 9 in each group), Block (1-10), and Trial (100 per rat). I think my confusion lies in how to organize this where 10 trials would be nested under 1 block, and 9 subjects would be nested under each drug group, but all subjects run the same amount of trials.

Right now my code is

Code:

`m3 <- lmer(Latency ~ Block + Group + Block:Group + (1|Subject) + (Block/Trial), data=Block)`

I'm also having trouble figuring this out conceptually, even after looking through a handful of articles on it. I understand that subject would have their own intercept, but I'm kind of confused on the random slope part of it. ]]>

2 IVs: Exposure Therapy (2 levels: control, or virtual reality) and Medication (2 levels: Propranolol, or placebo) (Giving 4 treatment groups/combinations)

2 DVs: The CAPS-5 and Skin conductance. (To measure PTSD symptom levels in two forms)

I get I do a two-way MANOVA, and then can do 2 ANOVAs each looking at one DV (with Bonferonni correction).

But to find out which treatment group is having significant effects, assuming MANOVA interaction was sig, and both ANOVAs were sig, will I be doing simple effects tests, scheffe test. This is where I am confused. (This is a theoretical project, so no data is collected, but I must describe what analysis would be used).

Please, any help greatly appreciated!! ]]>