Hello There, Guest!  

[ODP] - Parents’ Income is a Poor Predictor of SAT Score

#31
(2014-Jul-02, 17:51:06)nooffensebut Wrote: @menghu1001
My advice is to convince your colleagues to let your journal state up-front that this type of MR research will not be published, and SEM is strongly encouraged. If that’s a good thing, then you will have distinguished your journal from every other one through your unique, principled stance.


This journal is open, as its name says and we do not adopt any principled stance.
I do not know what you mean. Meng Hu is not the editor of this journal. Emil and I are the creators of OpenPsych and Emil and I are the editors of ODP and OBG, respectively. We're not trying to stop you from publishing on this journal. I think Emil is waiting to see if you and Meng Hu come to an agreement before approving publication. I think this paper is fine and approve publication and so does Philbrick, but a third reviewer needs to approve before this paper can be published. I think now the debate between you and Meng Hu has been going on for long enough and someone (if not Meng Hu, another reviewer) is invited to give his opinion on this paper.
 Reply
#32
(2014-Jul-02, 17:51:06)nooffensebut Wrote: I did state the possibility that income can have a big impact in order to find agreement with you, which failed, and I was wrong to do so because income cannot have a big impact


You have not acknowledged it because you have rejected it immediately as "unlikely" because of the absence of multicollinearity. I already said multicollinearity is irrelevant because its definition is that the independent var. are "too highly correlated" and my argumentation only was about the fact that the independent var are correlated. You cannot reject the high correlation between income and grade/education, as I demonstrate above, when using your data in a path analysis, I obtain a high correlation between income and education (about 0.60-0.70). But you never replied to that specific point. I also told you that SEM decomposition is not an interaction effect. I showed you what an interaction effect is. Here's the link again. In light of this, I don't understand why you keep saying "I don’t appreciate that you insist on obfuscating by borrowing the language of multicollinearity (“total effect,” “indirect effect,” and “direct effect”) when you clearly mean interaction effects in order to keep alive the income variable" because you just keep repeating what I have already responded to.

Do you agree with this ?

1. Absence of multicollinearity ("too high correlation") does not imply the independent var. are not correlated.
2. As soon as there is correlation between independent variables (e.g., income-grade at r=0.60-0.70), indirect effect could have been hidden, not seen, in the MR regression coefficient.
3. An interaction effect consists in multiplying variable A by variable B, which creates variable AB, and SEM decomposition is not an interaction effect because it merely multiplies the "path correlations" but not the variables.

Answer these questions, and we will make progress. Ok ?

(2014-Jul-02, 17:51:06)nooffensebut Wrote: If you make an objection that I fail to address (because lack of evidence/proof or unconvincing argumentation)


Suppose a reviewer makes a proposition n°1, and you cannot respond to it (because lack of evidence/proof or unconvincing argumentation). You can make a counter-argument, proposition n°2. And if I fail to respond to it, there is no problem for me if you include my objection and your objection to mine.

Quote:GCTA bivariate correlation for education (my independent variable) and cognitive ability is about perfect.

Remember, your complete sentence was : "I admitted that the bachelor’s degree’s VIF was moderately high, and I offered the omitted variable of “genetic and developmental effects on cognitive ability” as a possible explanation. My prior discussion of GCTA already admitted and supported this." Why this is wrong is because VIF only involves the independent variables, not the dependent.

(2014-Jul-02, 17:51:06)nooffensebut Wrote: You insist on doing so without providing any evidence for your unorthodox view


The proof was already given here, but you never replied to it. You merely said ....

Quote:Okay, so beta coefficients can’t see the total effect, but coefficients of determination can because they can in an SEM model. I assume you are referring to some other research project you have, but if it is not MR, how does that prove that the coefficient of determination in an MR model can see the same thing?

I have replied that "it's common sense ... MR and SEM are both multiple regression." and added the following...

http://humanvarietiesdotorg.files.wordpr...ession.png
http://humanvarietiesdotorg.files.wordpr...os-sem.png

... to which you have never commented. I'm still waiting for my answer. The other proof comes from dominance analysis approach, and I cited several others saying and demonstrating that the beta weights don't add up to the model r². This argues against your views that the MR beta weights provide the total effect.

So, please, stop saying I never provide any proof of my views. You're the only one who need to provide a proof that I am wrong. If not, I have requested you to add a word of cautious about the possibility that income may have large indirect effect, which is not given in your beta weights. For the 3rd (or 4rth) time, I repeat, you don't need to endorse my views, but merely add a word of cautious among the limitations of your paper.

Duxide Wrote:I think now the debate between you and Meng Hu has been going on for long enough


Does that mean we have no right to continue ? That's a bad idea, I believe. Personally, anyway, if he refuses to respond to my 3 questions without going around anywhere, I will stop the conversation. That's my last attempt. I can see he repeats his old arguments I have responded to already. I don't want to continue like that. If he does not reply to all the 3 questions, I stop. Simply.

EDIT. I forgot something. I also want him to reply to a fourth question. The two pictures png I linked to, above. I request an answer to this.
 Reply
#33
@menghu1001

Calling interaction effects “indirect effects” or “SEM decomposition” doesn’t appear to be working. Why not call them something more creative, like Puff-the-Magic-Dragon vectors? Here’s another idea: point me to a person—any person—who says “SEM decomposition” effects in a multiple regression model need not be specified to influence R^2—<i>anyone other than you</i>. Show where a statistician says “multicollinearity is ambiguous.” Introduce me to someone who believes what you believe, like I did. There’s no need to link to yourself because I already completely understand that you agree with yourself.
 Reply
#34
So, everyone...

nooffensebut appears to focuse on interaction effect. I have dealt with that this way :

Quote:An interaction effect consists in multiplying variable A by variable B, which creates variable AB, and SEM decomposition is not an interaction effect because it merely multiplies the "path correlations" but not the variables.

He is not answering to that, and I'm sorry but I'm still waiting for an answer to my 4 questions. He is also saying that :

nooffensebut Wrote:point me to a person ... who says “SEM decomposition” effects in a multiple regression model need not be specified to influence R^2


Even though my pictures show the proof that the r² model, beta coefficients, standardized or not, t-stats, p-values, correspond. That SEM is a multiple regression as well, and both have the same basic assumptions and requirements for correct inferences, etc.

I didn't want to get to this point, after spending so much time, hoping it would have been published. It is no good news to say this, but until I get my answer to my list of questions, I will not give my approval for the publication of the paper, since the author refuses to address my points.
 Reply
#35
It seems that there can be no agreement between Meng Hu and nooffensebut. My position is that it is a fine paper and I was merely waiting to see if agreement could be reached.

Recall that the function of peer review is not to ensure consensus about statistical methods and interpretation among reviewers and authors. It is to catch any obvious blunders in analysis and interpretation. I don't see any such.

I approve of the paper, raising the reviewer count to 3 as needed. Can nooffensebut submit a final version ready for publication? Remember to fill in the date where it is to be published.
 Reply
#36
Here is the final version with the same supplements originally posted:

Author: nooffensebut

Abstract

Parents’ annual income lacks statistical significance as a predictor of state SAT scores when additional variables are well controlled. Spearman rank correlation coefficients reveal parents’ income to be a weaker predictor of average SAT scores for each income bracket within each state than parents’ education level as a predictor of average SAT scores for each education level within each state. Multiple linear regression of state SAT scores with covariates for sample size, state participation, year, and each possible combination of ordinal variables for parents’ income, parents’ education, and race shows income to lack statistical significance in 49% of the iterations with greater frequency of insignificance among iterations with higher explained variance. Cohen’s d comparisons of the yearly individual SAT advantage of having educated parents show a fairly consistently increasing positive relationship over time, whereas similar analysis of the yearly individual SAT advantage of having high-income parents shows variability somewhat coinciding with the business cycle.

Key words: SAT; socioeconomic status; income; education; race

(Please disregard the first file)


Attached Files
.pdf   ParentsIncomeisaPoorPredictorofSATScores7-3-2014Final.pdf (Size: 1.34 MB / Downloads: 427)
.xlsx   supplement 1 - education data.xlsx (Size: 242.46 KB / Downloads: 418)
.xlsx   supplement 2 - income data.xlsx (Size: 303.62 KB / Downloads: 411)
.xlsx   supplement 3 - states continuous variables data.xlsx (Size: 146.54 KB / Downloads: 411)
.pdf   ParentsIncomeisaPoorPredictorofSATScore7-3-2014Final.pdf (Size: 1.34 MB / Downloads: 523)
 Reply
#37
Thank you. The paper is published here: http://openpsych.net/ODP/2014/07/parents...sat-score/

If you see any errors, please let me know. If none, let me know as well, so I can move this to the post-publication forum.
 Reply
#38
Thank you.

There are no errors.
 Reply
#39
I have updated it with the new name, which sounds sort of silly when using the lastname first convention.
 Reply
 
Forum Jump:

Users browsing this thread: 1 Guest(s)