r - How to do something like repeated measures but with multiple variables tested - Cross Validated


we did experiment , want @ results, have lot of problems finding right statistical test.

to line out experiment: our subjects scanned twice on , off medication. take performance test on both days, , on both days measure concentration of molecule in 3 different areas. because effect can depend on property of area (say density) have 1 density measure per area per subject.

we suspect 2 things: concentration of molecule - in area 1, , potentially area 2, not in area 3 - have effect on performance. medication affect molecule concentration in area 1 (and maybe 2), , hence alter performance.

and further know: density in region can have effect our molecule measure, it's nuisance variable have take account.

now want test whether see effect of molecule concentration on performance, specific region. , whether medication can affect concentration (and hence, performance).

my question is: how correctly statistically answer this.

my problem using repeated measures (e.g. in spss) let account 3 regions , medication (3measuresx2days anova) impossible incorporate either performance on both days (1measurex2days), or density (3measuresx1day).

i've tried setup mixed models in r, gives me error not have sufficient datapoints, since have 1 measure , not several trials per day.

using linear model can solve day effect , can separately test: relation between performance , concentration:

lm(meanconcentration~meanperformance + differcenceperformance + differenceconcentration ) 

effect of medication:

lm(differenceconcentration ~differcenceperformance +meanperformance + meanconcentration) 

however, cannot test whether effect specific 1 of regions.

anyone have idea test use (and how!)?


Comments

Popular posts from this blog

PHP and MySQL WP -

android - InAppBilling registering BroadcastReceiver in AndroidManifest -

go - golang pprof for c library code -