| Characteristic | N = 655 |
|---|---|
| schoolstart | NA |
| No risk | 314 (48%) |
| Undecided | 211 (32%) |
| Risk | 127 (19%) |
| Missing | 3 (0.5%) |
| ggrisk | NA |
| No risk | 246 (38%) |
| Risk | 289 (44%) |
| Missing | 120 (18%) |
| march | NA |
| No risk | 255 (39%) |
| Undecided | 173 (26%) |
| Risk | 93 (14%) |
| Missing | 134 (20%) |
| screeningapril | NA |
| No risk | 422 (64%) |
| Risk | 204 (31%) |
| Missing | 29 (4.4%) |
Assessment of at-risk across three occassions
Participants and variables
We have 655 participants. These are nested withing teachers, but the nesting was lost during data collection, unfortunately. So inferential statistics is invalid mostly. Nevertheless we can look at the data from a descriptive perspective
Our focus variables have the following characteristics
Visualization
Addendum PH request june 6th 25 With only complete data and complete+collapsed
A total of 186 students had missing at at least one occassion. We remove these and redo the graph:
Transitions pairwise
Let us look at all pairs of measurement occasions.
schoolstart -> ggrisk
category
time No risk Undecided Risk Missing
schoolstart 314 211 127 3
ggrisk 246 0 289 120# A tibble: 10 × 4
schoolstart ggrisk n prop
<fct> <fct> <int> <dbl>
1 No risk No risk 173 0.551
2 No risk Risk 82 0.261
3 No risk Missing 59 0.188
4 Undecided No risk 61 0.289
5 Undecided Risk 121 0.573
6 Undecided Missing 29 0.137
7 Risk No risk 12 0.0945
8 Risk Risk 86 0.677
9 Risk Missing 29 0.228
10 Missing Missing 3 1 
schoolstart -> march
category
time No risk Undecided Risk Missing
schoolstart 314 211 127 3
march 255 173 93 134# A tibble: 13 × 4
schoolstart march n prop
<fct> <fct> <int> <dbl>
1 No risk No risk 195 0.621
2 No risk Undecided 50 0.159
3 No risk Risk 6 0.0191
4 No risk Missing 63 0.201
5 Undecided No risk 56 0.265
6 Undecided Undecided 91 0.431
7 Undecided Risk 23 0.109
8 Undecided Missing 41 0.194
9 Risk No risk 4 0.0315
10 Risk Undecided 32 0.252
11 Risk Risk 64 0.504
12 Risk Missing 27 0.213
13 Missing Missing 3 1 
ggrisk -> march
category
time No risk Undecided Risk Missing
ggrisk 246 0 289 120
march 255 173 93 134# A tibble: 12 × 4
ggrisk march n prop
<fct> <fct> <int> <dbl>
1 No risk No risk 175 0.711
2 No risk Undecided 44 0.179
3 No risk Risk 5 0.0203
4 No risk Missing 22 0.0894
5 Risk No risk 63 0.218
6 Risk Undecided 115 0.398
7 Risk Risk 74 0.256
8 Risk Missing 37 0.128
9 Missing No risk 17 0.142
10 Missing Undecided 14 0.117
11 Missing Risk 14 0.117
12 Missing Missing 75 0.625 
schoolstart -> screeningapril
category
time No risk Undecided Risk Missing
schoolstart 314 211 127 3
screeningapril 422 0 204 29# A tibble: 10 × 4
schoolstart screeningapril n prop
<fct> <fct> <int> <dbl>
1 No risk No risk 263 0.838
2 No risk Risk 47 0.150
3 No risk Missing 4 0.0127
4 Undecided No risk 118 0.559
5 Undecided Risk 86 0.408
6 Undecided Missing 7 0.0332
7 Risk No risk 41 0.323
8 Risk Risk 71 0.559
9 Risk Missing 15 0.118
10 Missing Missing 3 1 
ggrisk -> screeningapril
category
time No risk Risk Missing
ggrisk 246 289 120
screeningapril 422 204 29# A tibble: 9 × 4
ggrisk screeningapril n prop
<fct> <fct> <int> <dbl>
1 No risk No risk 202 0.821
2 No risk Risk 37 0.150
3 No risk Missing 7 0.0285
4 Risk No risk 154 0.533
5 Risk Risk 127 0.439
6 Risk Missing 8 0.0277
7 Missing No risk 66 0.55
8 Missing Risk 40 0.333
9 Missing Missing 14 0.117 
march -> screeningapril
category
time No risk Undecided Risk Missing
march 255 173 93 134
screeningapril 422 0 204 29# A tibble: 12 × 4
march screeningapril n prop
<fct> <fct> <int> <dbl>
1 No risk No risk 227 0.890
2 No risk Risk 25 0.0980
3 No risk Missing 3 0.0118
4 Undecided No risk 111 0.642
5 Undecided Risk 60 0.347
6 Undecided Missing 2 0.0116
7 Risk No risk 25 0.269
8 Risk Risk 62 0.667
9 Risk Missing 6 0.0645
10 Missing No risk 59 0.440
11 Missing Risk 57 0.425
12 Missing Missing 18 0.134 
Regression KP risk as outcome
First, regress on the september data alone
| Characteristic | log(OR) | 95% CI | p-value |
|---|---|---|---|
| schoolstart | |||
| No risk | — | — | |
| Undecided | 1.4 | 1.0, 1.8 | <0.001 |
| Risk | 2.4 | 1.9, 2.9 | <0.001 |
| Missing | 16 | -50, |
>0.9 |
| Abbreviations: CI = Confidence Interval, OR = Odds Ratio | |||
It is clear that the septemberteacher predictions can predict the national screening test results.
Let us see whether Graphogame adds anything over and beyond by including it as well
| Characteristic | log(OR) | 95% CI | p-value |
|---|---|---|---|
| schoolstart | |||
| No risk | — | — | |
| Undecided | 1.3 | 0.84, 1.7 | <0.001 |
| Risk | 2.1 | 1.6, 2.6 | <0.001 |
| Missing | 16 | -51, |
>0.9 |
| ggrisk | |||
| No risk | — | — | |
| Risk | 0.76 | 0.32, 1.2 | <0.001 |
| Missing | 0.95 | 0.42, 1.5 | <0.001 |
| Abbreviations: CI = Confidence Interval, OR = Odds Ratio | |||
Finally, including the march predictions
| Characteristic | log(OR) | 95% CI | p-value |
|---|---|---|---|
| schoolstart | |||
| No risk | — | — | |
| Undecided | 1.1 | 0.64, 1.6 | <0.001 |
| Risk | 1.4 | 0.86, 2.0 | <0.001 |
| Missing | 15 | -51, |
>0.9 |
| ggrisk | |||
| No risk | — | — | |
| Risk | 0.29 | -0.22, 0.80 | 0.3 |
| Missing | -0.05 | -0.67, 0.57 | 0.9 |
| march | |||
| No risk | — | — | |
| Undecided | 0.85 | 0.29, 1.4 | 0.003 |
| Risk | 2.1 | 1.3, 2.8 | <0.001 |
| Missing | 2.1 | 1.5, 2.7 | <0.001 |
| Abbreviations: CI = Confidence Interval, OR = Odds Ratio | |||