nhaliday + graphical-models   30

Fitting a Structural Equation Model
seems rather unrigorous: nonlinear optimization, possibility of nonconvergence, doesn't even mention local vs. global optimality...
pdf  slides  lectures  acm  stats  hypothesis-testing  graphs  graphical-models  latent-variables  model-class  optimization  nonlinearity  gotchas  nibble  ML-MAP-E  iteration-recursion  convergence 
november 2017 by nhaliday
Does Learning to Read Improve Intelligence? A Longitudinal Multivariate Analysis in Identical Twins From Age 7 to 16
Stuart Richie, Bates, Plomin

SEM: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4354297/figure/fig03/

The variance explained by each path in the diagrams included here can be calculated by squaring its path weight. To take one example, reading differences at age 12 in the model shown in Figure​Figure33 explain 7% of intelligence differences at age 16 (.262). However, since our measures are of differences, they are likely to include substantial amounts of noise: Measurement error may produce spurious differences. To remove this error variance, we can take an estimate of the reliability of the measures (generally high, since our measures are normed, standardized tests), which indicates the variance expected purely by the reliability of the measure, and subtract it from the observed variance between twins in our sample. Correcting for reliability in this way, the effect size estimates are somewhat larger; to take the above example, the reliability-corrected effect size of age 12 reading differences on age 16 intelligence differences is around 13% of the “signal” variance. It should be noted that the age 12 reading differences themselves are influenced by many previous paths from both reading and intelligence, as illustrated in Figure​Figure33.

...

The present study provided compelling evidence that improvements in reading ability, themselves caused purely by the nonshared environment, may result in improvements in both verbal and nonverbal cognitive ability, and may thus be a factor increasing cognitive diversity within families (Plomin, 2011). These associations are present at least as early as age 7, and are not—to the extent we were able to test this possibility—driven by differences in reading exposure. Since reading is a potentially remediable ability, these findings have implications for reading instruction: Early remediation of reading problems might not only aid in the growth of literacy, but may also improve more general cognitive abilities that are of critical importance across the life span.

Does Reading Cause Later Intelligence? Accounting for Stability in Models of Change: http://sci-hub.tw/10.1111/cdev.12669
Results from a state–trait model suggest that reported effects of reading ability on later intelligence may be artifacts of previously uncontrolled factors, both environmental in origin and stable during this developmental period, influencing both constructs throughout development.
study  albion  scitariat  spearhead  psychology  cog-psych  psychometrics  iq  intelligence  eden  language  psych-architecture  longitudinal  twin-study  developmental  environmental-effects  studying  🌞  retrofit  signal-noise  intervention  causation  graphs  graphical-models  flexibility  britain  neuro-nitgrit  effect-size  variance-components  measurement  multi  sequential  time  composition-decomposition  biodet  behavioral-gen  direct-indirect  systematic-ad-hoc  debate  hmm  pdf  piracy  flux-stasis 
september 2017 by nhaliday
PsycARTICLES - Is education associated with improvements in general cognitive ability, or in specific skills?
Results indicated that the association of education with improved cognitive test scores is not mediated by g, but consists of direct effects on specific cognitive skills. These results suggest a decoupling of educational gains from increases in general intellectual capacity.

look at Model C for the coefficients

How much does education improve intelligence? A meta-analysis: https://psyarxiv.com/kymhp
Intelligence test scores and educational duration are positively correlated. This correlation can be interpreted in two ways: students with greater propensity for intelligence go on to complete more education, or a longer education increases intelligence. We meta-analysed three categories of quasi-experimental studies of educational effects on intelligence: those estimating education-intelligence associations after controlling for earlier intelligence, those using compulsory schooling policy changes as instrumental variables, and those using regression-discontinuity designs on school-entry age cutoffs. Across 142 effect sizes from 42 datasets involving over 600,000 participants, we found consistent evidence for beneficial effects of education on cognitive abilities, of approximately 1 to 5 IQ points for an additional year of education. Moderator analyses indicated that the effects persisted across the lifespan, and were present on all broad categories of cognitive ability studied. Education appears to be the most consistent, robust, and durable method yet to be identified for raising intelligence.

three study designs: control for prior IQ, exogenous policy change, and school age cutoff regression discontinuity

https://westhunt.wordpress.com/2017/11/07/skoptsys/#comment-97601
It’s surprising that there isn’t much of a fadeout (p11) – half of the effect size is still there by age 70 (?!). That wasn’t what I expected. Maybe they’re being pulled upwards by smaller outlier studies – most of the bigger ones tend towards the lower end.

https://twitter.com/gwern/status/928308706370052098
https://archive.is/v98bd
These gains are hollow, as they acknowledge in the discussion. Examples:
albion  spearhead  scitariat  study  psychology  cog-psych  iq  large-factor  education  intervention  null-result  longitudinal  britain  anglo  psychometrics  psych-architecture  graphs  graphical-models  causation  neuro-nitgrit  effect-size  stylized-facts  direct-indirect  flexibility  input-output  evidence-based  preprint  multi  optimism  meta-analysis  west-hunter  poast  commentary  aging  marginal  europe  nordic  shift  twitter  social  backup  ratty  gwern  links  flynn  environmental-effects  debate  roots 
march 2017 by nhaliday
The genetics of politics: discovery, challenges, and progress
Figure 1. Summary of relative genetic and environmental influences on political traits.

- heritability increases discontinuously on leaving home
- pretty big range of heritability for different particular traits (party identification is lowest w/ largest shared environment by far)
- overall ideology quite highly heritable
- social trust is surprisingly highly compared other measurements I've seen...
- ethnocentrism quite low (sample-dependent?)
- authoritarianism and traditionalism quite high
- voter turnout quite high

Genes, psychological traits and civic engagement: http://rstb.royalsocietypublishing.org/content/370/1683/20150015
We show an underlying genetic contribution to an index of civic engagement (0.41), as well as for the individual acts of engagement of volunteering for community or public service activities (0.33), regularly contributing to charitable causes (0.28) and voting in elections (0.27). There are closer genetic relationships between donating and the other two activities; volunteering and voting are not genetically correlated. Further, we show that most of the correlation between civic engagement and both positive emotionality and verbal IQ can be attributed to genes that affect both traits.

Are Political Orientations Genetically Transmitted?: http://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=1006&context=poliscifacpub
TABLE 1. Genetic and Environmental Influences on Political Attitudes: The 28 Individual Wilson–Patterson Items

The origins of party identification and its relationship to political orientations: http://sci-hub.tw/http://www.sciencedirect.com/science/article/pii/S0191886915002470

All models showed a good overall fit (see Table 3). The data indicate that party identification is substantially heritable, with about 50% of the variation in PID attributable to additive genetic effects. Moreover, the results indicate that the non-genetic influences on party identification stem primarily from unique environmental factors rather than shared ones such as growing up in the same family. This too is not consistent with the Michigan model.

Table 3 also indicates that genetic influences explained about 50% of the variance in liberalism–conservatism. This estimate is similar to previous behavior genetic findings on political attitudes (e.g., Alford et al., 2005; Bouchard, 2004; Hatemi et al., 2014; Kandler, Bleidorn, & Riemann, 2012). The remaining variance was again due primarily to nonshared environmental influences. The latter finding indicates that the Michigan hypothesis that partisan social influences affect political orientations may have some merit, although the substantial level of heritability for this variable suggests that genetic effects also play an important role.

...

As Table 4 reveals, the best fitting model indicates that 100% of the genetic variance in PID is held in common with liberalism–conservatism ([aC2]/[aC2 + aPID2] = 1.00). Similarly, 73% of the environmental variation in PID is shared with liberalism–conservatism ([eC2]/[eC2 + ePID2] = .73). All told, only 13% of the total variance in PID cannot be explained by variation in liberalism–conservatism (1 [aC2 + eC2] = .13), as illustrated in Fig. 3. Since only a small proportion of the variance in PID cannot be explained by liberalism– conservatism, the findings are consistent with the hypothesis that genetic and environmental factors influence liberalism–conservatism, which in turn affects party identification. However, as discussed below, other causal scenarios cannot be ruled out.

Table 4 and Fig. 3 also show that 55% of the total variance in liberalism–conservatism cannot be accounted for by variance in PID

Fig. 3. Venn diagram mapping the common and specific variance in party
identification and liberalism–conservatism.

intuition for how you can figure out overlap of variance: look at how corr(PID, liberal-conservative) differs between MZ and DZ twin pairs, etc., fit structural equational model

p_k,i,j = r_A a_k,i,j,p + r_C c_k,i,p + r_E e_k,i,j,p (k=MZ or DZ, i=1..n_k, j=1,2, p=PID or LC value)

c_k,i,j,p = r_{C,p} c'_k,i,p + r_{C,common} c'_k,i,common (ditto)
e_k,i,j,p = r_{E,p} e'_k,i,j,p + r_{E,common} e'_k,i,j,common (ditto)

MZ twins:
a_MZ,i,j,p = r_{A,p} a'_MZ,i,p + r_{A,common} a'_MZ,i,common (i=1..n_k, j=1,2 p=PID or LC value)

DZ twins:
a_DZ,i,j,p = r_{A,p} (1/2 a'_DZ,i,p + 1/2 a'_DZ,i,j,p) + r_{A,common} (1/2 a'_DZ,i,common + 1/2 a'_DZ,i,j,common) (i=1..n_k, j=1,2 p=PID or LC value)

Gaussian distribution for the underlying a', c' and e' variables, maximum likelihood, etc.

see page 9 here: https://pinboard.in/u:nhaliday/b:70f8b5b559a9

basically:
1. calculate population means μ from data (so just numbers)
2. calculate covariance matrix Σ in terms of latent parameters r_A, r_C, etc. (so variable correlations)
3. assume observed values are Gaussian with those parameters μ, Σ
4. maximum likelihood to figure out the parameters r_A, r_C, etc.

A Genetic Basis of Economic Egalitarianism: http://sci-hub.tw/10.1007/s11211-017-0297-y
Our results show that the large portion of the variance in a four-item economic egalitarianism scale can be attributed to genetic factor. At the same time, shared environment, as a socializing factor, has no significant effect. The effect of environment seems to be fully reserved for unique personal experience. Our findings further problematize a long-standing view that social justice attitudes are dominantly determined by socialization.

published in the journal "Social Justice Research" by some Hungarians, lol

various political science findings, w/ a few behavioral genetic, focus on Trump, right-wing populism/authoritarianism, and polarization: http://www.nationalaffairs.com/blog/detail/findings-a-daily-roundup/a-bridge-too-far
pdf  study  org:nat  biodet  politics  values  psychology  social-psych  genetics  variance-components  survey  meta-analysis  environmental-effects  🌞  parenting  replication  candidate-gene  GWAS  anthropology  society  trust  hive-mind  tribalism  authoritarianism  things  sociology  expression-survival  civic  shift  ethnocentrism  spearhead  garett-jones  broad-econ  political-econ  behavioral-gen  biophysical-econ  polisci  stylized-facts  neuro-nitgrit  phalanges  identity-politics  tradition  microfoundations  ideology  multi  genetic-correlation  data  database  twin-study  objektbuch  gender  capitalism  peace-violence  military  labor  communism  migration  civil-liberty  exit-voice  censorship  sex  sexuality  assortative-mating  usa  anglo  comparison  knowledge  coalitions  piracy  correlation  intersection  latent-variables  methodology  stats  models  ML-MAP-E  nibble  explanation  bioinformatics  graphical-models  hypothesis-testing  intersection-connectedness  poll  egalitarianism-hierarchy  envy  inequality  justice  westminster  publishing 
february 2017 by nhaliday
What is the difference between inference and learning? - Quora
- basically boils down to latent variables vs. (hyper-)parameters
- so computing p(x_h|x_v,θ) vs. computing p(θ|X_v)
- from a completely Bayesian perspective, no real difference
- described in more detail in [Kevin Murphy, 10.4]
q-n-a  qra  jargon  machine-learning  stats  acm  bayesian  graphical-models  latent-variables  confusion  comparison  nibble 
january 2017 by nhaliday
CS 731 Advanced Artificial Intelligence - Spring 2011
- statistical machine learning
- sparsity in regression
- graphical models
- exponential families
- variational methods
- MCMC
- dimensionality reduction, eg, PCA
- Bayesian nonparametrics
- compressive sensing, matrix completion, and Johnson-Lindenstrauss
course  lecture-notes  yoga  acm  stats  machine-learning  graphical-models  graphs  model-class  bayesian  learning-theory  sparsity  embeddings  markov  monte-carlo  norms  unit  nonparametric  compressed-sensing  matrix-factorization  features 
january 2017 by nhaliday
D-separation
collider C = A->C<-B
A, B d-connected (resp. conditioned on Z) iff path A~>B or B~>A w/o colliders (resp. path excluding vertices in Z)
A,B d-separated conditioned on Z iff not d-connected conditioned on Z

http://bayes.cs.ucla.edu/BOOK-2K/d-sep.html
concept  explanation  causation  bayesian  graphical-models  cmu  org:edu  stats  methodology  tutorial  jargon  graphs  hypothesis-testing  confounding  🔬  direct-indirect  philosophy  definition  volo-avolo  multi  org:junk 
january 2017 by nhaliday
A Variant on “Statistically Controlling for Confounding Constructs is Harder than you Think”
It’s taken me some time to master this formalism, but I now find it quite easy to reason about these kinds of issues thanks to the brevity of graphical models as a notational technique. I’d love to see this approach become more popular in psychology, given that it has already become quite widespread in other fields. Of course, Westfall and Yarkoni are already advocating for something very similar by advocating for the use of SEM’s, but the graphical approach is strictly more general than SEM’s and, in my personal opinion, strictly simpler to reason about.
bayesian  stats  thinking  visualization  study  science  gelman  hmm  methodology  causation  acmtariat  meta:science  graphs  commentary  techtariat  hypothesis-testing  org:bleg  nibble  scitariat  confounding  🔬  info-dynamics  direct-indirect  volo-avolo  endo-exo  endogenous-exogenous  control  graphical-models 
may 2016 by nhaliday

bundles : academeacmframe

related tags

2016-election  academia  acm  acmtariat  age-generation  aging  akrasia  albion  algorithmic-econ  analysis  anglo  anthropology  apollonian-dionysian  applications  approximation  arms  arrows  article  assortative-mating  atoms  authoritarianism  backup  bayesian  behavioral-gen  big-picture  bio  biodet  bioinformatics  biophysical-econ  blowhards  books  britain  broad-econ  c:***  canada  candidate-gene  capitalism  causation  censorship  chart  checklists  civic  civil-liberty  class  clever-rats  cmu  coalitions  coding-theory  cog-psych  columbia  commentary  communism  comparison  composition-decomposition  compressed-sensing  concept  conceptual-vocab  confidence  confluence  confounding  confusion  control  convergence  convexity-curvature  correlation  course  curvature  data  data-science  database  debate  deep-learning  definition  demographics  developmental  differential  dimensionality  direct-indirect  discussion  distribution  DP  draft  dysgenics  economics  eden  education  effect-size  egalitarianism-hierarchy  embeddings  empirical  endo-exo  endogenous-exogenous  ends-means  entropy-like  environmental-effects  envy  essay  estimate  ethnocentrism  europe  evidence-based  examples  exit-voice  experiment  expert  expert-experience  explanation  exposition  expression-survival  features  fertility  finiteness  flexibility  flux-stasis  flynn  fourier  game-theory  garett-jones  gelman  gender  generative  genetic-correlation  genetics  genomics  gotchas  gradient-descent  graph-theory  graphical-models  graphs  ground-up  GWAS  gwern  GxE  hari-seldon  hi-order-bits  hive-mind  hmm  homepage  homogeneity  human-capital  hypothesis-testing  ideas  identity-politics  ideology  idk  impact  inequality  info-dynamics  information-theory  init  input-output  intelligence  interdisciplinary  internet  intersection  intersection-connectedness  intervention  iq  ising  iteration-recursion  jargon  justice  knowledge  labor  language  large-factor  latent-variables  learning-theory  lecture-notes  lectures  levers  libraries  linear-algebra  linear-models  linearity  links  list  longitudinal  machine-learning  manifolds  marginal  markov  martingale  math  math.CA  math.DS  math.GN  matrix-factorization  measure  measurement  mechanism-design  meta-analysis  meta:math  meta:science  metabuch  metameta  methodology  microfoundations  migration  military  ML-MAP-E  model-class  models  monte-carlo  motivation  multi  n-factor  nationalism-globalism  neuro-nitgrit  news  nibble  nitty-gritty  nlp  nonlinearity  nonparametric  nordic  norms  nostalgia  null-result  objektbuch  off-convex  optimism  optimization  org:bleg  org:edu  org:junk  org:mag  org:mat  org:nat  oss  overflow  p:*  p:***  p:someday  PAC  papers  parenting  paste  pdf  peace-violence  pennsylvania  people  personality  pessimism  phalanges  phase-transition  philosophy  phys-energy  physics  pigeonhole-markov  piracy  poast  polarization  polisci  political-econ  politics  poll  pop-structure  population-genetics  populism  postmortem  ppl  pragmatic  pre-2013  preprint  princeton  prioritizing  probability  problem-solving  productivity  prof  programming  psych-architecture  psychology  psychometrics  publishing  python  q-n-a  qra  QTL  quantified-self  quixotic  race  random  ratty  reading  recent-selection  recommendations  reference  reflection  regression  reinforcement  religion  replication  repo  responsibility  retrofit  roadmap  roots  rot  s:***  sample-complexity  sampling  sanjeev-arora  sapiens  scholar-pack  science  scitariat  sequential  series  sex  sexuality  shift  sib-study  signal-noise  simulation  skeleton  slides  smoothness  social  social-psych  society  sociology  software  solid-study  sparsity  spearhead  spectral  spock  stanford  stat-mech  stat-power  state-of-art  stats  status  stochastic-processes  students  study  studying  stylized-facts  summary  survey  synthesis  systematic-ad-hoc  talks  tcs  techtariat  telos-atelos  the-bones  theos  thermo  things  thinking  time  toolkit  tools  top-n  topology  track-record  tradition  trends  tribalism  tricki  trump  trust  tutorial  twin-study  twitter  unit  usa  values  variance-components  virginia-DC  visualization  volo-avolo  wealth  west-hunter  westminster  wiki  woah  working-stiff  yoga  🌞  🎓  👳  🔬 

Copy this bookmark:



description:


tags: