ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
Question/comparisonValuesStatistical test and n numbersstat results
2
Tuning boutsMcNemar
3
Are more units tuned during locomotion?total N = 1583
4
StatGood: 407 (25.71%)
RunGood: 606 (38.28%)
BothGood:344 (21.73%)
(+stat+run, +stat-run
-stat+run, -stat-run)
5
344 63
262 914
6
7
8
Tuning start timesQuantilesLME
9
Stat: 0.0600 0.1700 0.4400
f = 'startTime ~ state + (1|unit) + (1|sesh)';
p = 3.3555E-11
10
Run: 0.0500 0.1000 0.2100n = 688 tuned in both
11
12
Tuning finish timesQuantilesLME
13
Stat: 0.8250 1.0700 1.1650
f = 'FinishTime ~ state + (1|unit) + (1|sesh)';
p = 1.645e-13
14
Run: 1.0500 1.1500 1.3400n = 688 tuned in both
15
16
Tuning durationsQuantilesLME
17
Stat: 0.1800 0.4650 0.9000
f = Duration ~ state + (1|unit) + (1|sesh)';
p =6.4217e-46
18
Run: 0.5350 0.9550 1.1300n = 688 tuned in both
19
20
21
Dynamic rangeMean (SEM)LME
22
Mean value over stimulus periodStat: 2.1741 (0.1093) f = 'vals ~ state + (1|unit) + (1|sesh)';p = 1.4188e-76
23
Run: 4.4138 (0.1917)n=1,583 good units
24
25
26
27
28
29
30
31
LDA - indy neurons by sessionRM-ANOVA
[p,tbl,stats,terms] = anovan(decVec,{timeVec,stateVec,subjVec},'model',2,'random',3,'varnames',{'Time','State','Subj'});
32
----------------------------------------------------------
33
Source Sum Sq. d.f. Mean Sq. F Prob>F
----------------------------------------------------------
Time 47.1951 190 0.2484 67.1 0
State 12.4225 1 12.4225 183.15 0.0002
Subj 4.8857 4 1.2214 17.64 0.0072
Time*State 3.6898 190 0.0194 8.52 0
Time*Subj 2.8132 760 0.0037 1.62 0
State*Subj 0.2713 4 0.0678 29.77 0
Error 1.7318 760 0.0023
Total 73.0094 1909
n = 5 subjects