psychelzh / g-factor Goto Github PK

We could do this by inspecting the loadings.

Originally on #7. Redirect it here:

1. Better estimate of g is achieved by more tasks.
   • Invariance
   • Brain correlation
2. RAPM is moderately correlated with g.
3. g is better correlated with brain network than RAPM; and task-state also have larger prediction.
4. Sex differences.
5. Predictive brain network distribution.

See this documentation, and change this accordingly. Maybe for future reference, and this project will be archived soon.

Current study mainly concerns the selection of tasks/tests to measure intelligence. In fact, Jensen and Humphreys differed with each on this point as follows.

At least Humphreys (1962) emphasized the heterogeneity of tests used to estimate general intelligence. We can test this in certain way.

Note: Humphreys was just emphasizing the importance to include heterogenous formats of materials in building tests.

It is so hard to separate this into branches for further analysis. This object is too large!

https://github.com/psychelzh/model-behav/blob/b7276c4bc09b219da774caf724af9e88b2cd8b08/R/targets_g_stability.R#L92-L108