After 40 testees, together with second norm report, I'll
publish detailed statistical information about quality of each test.
First of all, I give some notes and explain my procedure.
S=number of points achieved by top 90% scorers on particular
item
R=the worst ranking among testees from previous step
For each solved item I'll calculate the ratio S/R.
This ratio would ideally be equal to 1, and otherwise less
than 1.
If this ratio will be lower than 0.6, item will be qualified
as average.
If this ratio will be lower than 0.4, item will be qualified
as suspicious.
If this ratio will be lower than 0.25, item will be qualified
as bad.
Quality will be calculated as average ratio.
| IQ test |
good items |
average
items |
suspicious
items |
bad items |
unsolved
items |
quality |
| FREE Fall | ||||||
| LSHR | 9 |
10 |
4 |
5 |
2 |
51,9% |
| LSHR Light | 16 |
3 |
4 |
5 |
2 |
62,4% |
| LSHR Classic | 14 |
12 |
7 |
3 |
0 |
55,4% |
| Numerus | 10 |
4 |
11 |
5 |
0 |
46,4% |
| Numerus Light | 23 |
5 |
1 |
1 |
0 |
79,7% |
| Numerus Classic | 18 |
7 |
11 |
0 |
0 |
59,5% |
| Elementary | ||||||
| Triplex | ||||||
| Einplex | ||||||
| HI-Qlass |
These data will be updated after each 40 testees, up to 200 of them.
CONCLUSION:
Strong rules are presented here. In my opinion, tests with
quality higher than 60% are what we are looking for in IQ world,
and tests with quality lower than 35% should be avoided.