My response variable is accuracy (0 or 1) with 10 trials per subject. I thought about treating the response variable as a percentage and logit-transforming it to deal with its boundedness so that the predicted values were sensible, but this doesn't deal with the non-normality of the errors due to a large number of zeros in the data (i.e., About 25 of 83 subjects have only zeros (i.e., their accuracy rate was 0%). So now I am thinking about using a mixed logistic regression (i.e., generalized linear mixed model via glmer() in R with family = binomial("logit") and accounting for dependence among observations by modeling intercepts for subjects.
Is GLMM an adequate way to deal with the zeros? Is there another more appropriate model I could use that isn't overly complicated?
Also, any suggestions for how to approach this in R?
Edit: Here are my data (which if I were to run a GLMM I would convert back to long form):
zz <- "id successes ageMonths
2 2 10 83.1
3 3 0 82.0
4 4 0 80.2
5 6 0 63.5
6 7 3 85.2
7 8 0 84.0
9 10 0 83.2
10 11 0 74.8
11 12 0 65.6
12 14 0 81.8
13 15 0 83.0
14 16 2 70.2
15 17 0 76.2
16 18 10 64.4
17 19 0 79.9
18 20 1 76.9
19 21 0 71.3
20 22 0 79.8
21 23 0 82.1
22 24 1 79.8
23 25 0 77.1
24 26 10 63.5
25 27 0 80.3
26 28 0 81.8
27 29 10 68.8
28 30 0 81.5
29 31 0 74.8
30 32 0 60.4
31 33 0 77.8
32 35 10 69.1
34 37 0 70.4
35 38 0 83.3
36 39 3 69.2
37 40 0 77.4
38 41 0 77.2
39 42 9 64.3
40 43 9 69.2
41 44 0 68.4
42 45 1 60.3
43 46 3 73.1
44 47 10 67.4
45 48 0 67.2
46 49 8 68.0
47 50 0 60.8
48 51 0 63.5
49 52 0 66.4
50 53 3 71.7
51 54 0 67.4
52 55 0 60.2
53 56 10 70.8
54 57 3 78.1
55 58 10 61.4
56 59 1 61.3
57 60 0 60.9
58 61 0 67.3
59 62 0 73.2
60 63 0 70.8
61 64 0 64.8
62 65 0 62.4
63 66 6 73.3
64 67 0 62.3
65 68 10 63.3
66 69 10 60.1
67 70 0 72.6
68 71 10 63.6
69 72 7 73.0
70 73 0 61.3
71 74 6 83.7
72 75 10 82.2
73 76 7 82.7
74 77 10 61.1
75 78 0 71.8
77 80 3 70.3
78 81 0 62.5
79 82 4 72.3
80 83 5 73.1
81 84 0 70.4
82 85 1 75.4
83 86 0 77.1
84 87 0 75.6
85 88 0 73.4
86 89 0 73.8
87 90 0 72.3"
Data <- read.table(text=zz, header = TRUE); Data
Edit 2: clarity