Hello There, Guest!  
1 2 3 4 5 8 Next   

U.S. Ethnic/Race Differences in Aptitude by Generation

#1
I attached the following paper for review:

U.S. Ethnic/Race Differences in Aptitude by Generation: An Exploratory Meta-analysis

We conducted an exploratory meta-analysis using 18 samples for which we were able to decompose scores by sociological race and immigrant generation. For Blacks and Whites of the same generation, the first, second, and third+ generation B/W d-values were 0.79, 0.75, and 0.97. For Hispanics and NH Whites of the same generation, the first, second, and third+ generation H/W d-values were 0.82, 0.64, and 0.55. For Asians and Whites of the same generation, the first, second, and third+ generation d-values were -0.11, -0.25, and -0.21. Relative to third generation Whites, the average d-values were 0.96, 0.79, and 0.97 for first, second, and third+ generation Black individuals, 1.0, 0.68, and 0.55 for first, second, and third+ generation Hispanic individuals, and 0.03, -0.25, and -0.21 for first, second, and third+ generation Asian individuals.

Included:

PDF
Supplementary excel file


Attached Files
.pdf   RaceGenUSforOpenPsych2.pdf (Size: 1.17 MB / Downloads: 835)
.xlsx   副本RaceAptitudeGeneration3.xlsx (Size: 724.54 KB / Downloads: 583)
 Reply
#2
The survey table is referred to as "table 1" but it has no caption as such. It is also of too low resolution. Similar for the results tables.

There needs to be clearer description of the inclusion criteria and what method was used to find the studies. Otherwise it is a non-systematic review.

One table actually has the caption used in the source paper on it, which makes it confusable with the tables in this study.

I would include a reference to this study somewhere, since it is an important earlier review of Hispanic cognitive ability and discussion of the linguistic bias problem for this population. Dunn, LM. 1988. Bilingual Hispanic children on the US mainline: a review of research on their cognitive, linguistic and scholastic development.
 Reply
#3
(2014-Jun-30, 04:04:58)Emil Wrote: (1)The survey table is referred to as "table 1" but it has no caption as such. It is also of too low resolution. Similar for the results tables.
(2) There needs to be clearer description of the inclusion criteria and what method was used to find the studies. Otherwise it is a non-systematic review. (3) One table actually has the caption used in the source paper on it, which makes it confusable with the tables in this study. (4) I would include a reference to this study somewhere, since it is an important earlier review of Hispanic cognitive ability and discussion of the linguistic bias problem for this population. Dunn, LM. 1988. Bilingual Hispanic children on the US mainline: a review of research on their cognitive, linguistic and scholastic development.


As for (1) and (2), I will make the needed corrections. As for (4), I read over that and didn't see anything interesting. Also, I tried to not get overly involved with the explanations, since the paper was an exploratory meta-analysis and not an exploration of the cause of group differences. If anything I should cut discussions, not add to them. I can add a sentence or so, though -- how would you summarize, as pithily as possible, Dunn's findings with respect to language and H/W group differences? The paper was really difficult to read with the super close spacing and all. As for (3), this was an atypical meta-analysis since it involved me doing original research and meta-analyzing it -- i.e., combining the data. I don't think that a meta-analyses needs to be systematic. And this was not a review because I averaged results. I used this conceptualization:
"To represent this visually, the figure below shows that a meta-analysis may be part of a systematic review. A meta-analysis is also possible without doing a systematic review - you could just find a few studies and calculate a result, with no attempt to be systematic about how the particular studies were chosen"
http://www.cochrane-net.org/openlearning...mod3-2.htm

To cut down on possible bias I used national samples, which were mostly nationally representative. If you want I could call it: "an unsystematic exploratory meta-analysis" -- but I thought that "exploratory" implied that the analysis is not systematic. I could be wrong.

Maybe you can clarify what it is exactly that you would like me to add regarding this point. Maybe you would just like me to change the title. What's another terms for "an analysis in which a bunch of study results are combined together to produce summary results".
 Reply
#4
I updated the paper.

(1) I added table headings
(2) I added a passage which clarifies the selection process:

These surveys in particular were chosen because: (a) they allowed for scores to be decomposed by three generational groups (first, second, and third+) and four racial/ethnic groups (Whites, Blacks, Hispanics, and Asians), (b) they were mostly representative either of the U.S. national population or of the U.S. university population, © the data was publicly available, and (d) the test involved was at least a fair measure of general aptitude. We looked into other surveys but those generally did not meet one of the four criteria mentioned above or, alternatively, were too difficult to analyze. For example, the New Immigrant Survey did not contain third+ generation data and the National Longitudinal Survey of Youth ‘97 did not readily allow for a decomposition by generations. As for published results, we did not conduct a complete review of published studies, thus we classify this as an exploratory not a systematic meta-analysis. The research which we did find and did not include did not meet one of our criteria – typically that that scores were broken down by U.S. sociological race and generation -- or did not present statistics necessary to compute effects sizes.

(3) I didn't add anything about Dunn's paper because his discussion was not directly relevant as he did not discuss linguistic bias by generation.

(4) The graphic resolution is a problem for some tables. Before I try to address it, I will wait for other criticisms and see if I need to fully redo the relevant tables due to statistical errors, etc.
 Reply
#5
It's clearly not ready for publication, and remember the last time I saw that, I pointed it out many inconsistencies, notably the Add health. You should have waited until I finish with the add health and NLSY97.

I also said that I wanted to re-do your analysis for the NAAL 2002, but I don't find the variables for prose and math. We can discuss that in email.

1. PAPER :

For table 6, GSS, you must have precised the years, e.g., 1972 to 2010 or 1972 to 2012 ?. you must also precise which sampling weight you use, see below, the original configuration (COMPWT).
http://sda.berkeley.edu/cgi-bin/hsda?harcsda+gss12

For table 7, it needs to be made explicit that NLSF (public) data has not provided sampling weight.

In page 13, it should be "(see, Richwine, 2009, table 2.11)"

Concerning table 5, you should tell us how these numbers are computed because they are not the same it appears in Capps et al. (2012) table 1.

Page 15, "Assuming that there is a robust 2nd to third+ generation narrowing, some of this could be due to cultural and genetic assimilation." you should better type "generation cognitive gap narrowing". Same thing for this passage in page 7, "Alternatively, there might have been a genuine first to second generational cognitive narrowing in unbiased aptitude".

There is also a lot of missing references, such as Capps et al. (2012).

In the paper, table 1 is wide, you can probably rotate it. But the worse thing is that all pictures are blurred. You must do something with that. Copy pasting your pictures from your blog posts does not improve readability. The presentation is that of a complete "amateur" and if you want this journal to be taken very seriously by scientists in general, the presentation is very important. In my opinion, i don't necessarily care so much about presentation, but here, that's a peer-reviewed journal, so you should improve this.

The quoted passage under table 8 is in grey, not in black.

Also, you need to add a page number in the paper, next time. And also needed, since this journal wants you to share the data (that is not available for some of the data set you use), is a description and pictures of the statistical procedure you use, see e.g., here :
http://www.openpsych.net/forum/showthrea...524#pid524

"If anything I should cut discussions, not add to them"

The text is useful, otherwise the article is boring. You should keep them.

2. DATA FILE :

The tables in the excel data file are also very hard to read. Each cell needs to be much wider. For instance, the column width of L is just 1.80. Just see how L351 is difficult to read. Furthermore, you should put the data tables in sheet 1, and the syntax in sheet 2.

I don't see what the cell W5 refers to.

In cell F11, I prefer you write the entire word instead of something like "Approx."

Cell A169, you have :

"Variables: Student race/ethnicity (collapsed) (U.S. only) (2007, 2011), Gen \ born in [country], Gen \ [stmo or fem guard] born in [country]
Years: 2011, 2007"

But I prefer one variable per rows. here you should have 4 or 5 rows, for example.

The table in analysis 6 "eat" the space for analysis 7 and make some of your numbers (apparently, the black scores) disappeared. As i said before, it needs to be corrected.

in analysis 9, you should avoid "semi-sentence" like "used 3rd gen White SD, did not pool". It's better to make a complete sentence, and begin with "I have" or "we have" etc.

In general, for the data file, you should avoid copy-pasting tables and picture from others' work, and type the numbers yourself.

in analysis 13/14, the question "Where you born in the US?" is odd and i believe it should have said "were".

In cell Q1308, you have typed "Composite SAT and ACT Scores by Rac/ Generation" and it must be corrected.

....
....

In general, I need to replicate all of the data in your file. That will take some time. (We can do the exchange in mail)

I also think everyone should name the attached data file according to the title of the paper, just to make sure there is no ambiguity. (by the same token, the title of your pdf "RaceGenUSforOpenPsych2" should be renamed)

By the way, speaking about the title of the paper, i would prefer "U.S. Ethnic/Race Differences in Ability by Generation". I'm not sure what "aptitude" refers to and I believe "ability" is a clearer word.
 Reply
#6
(2014-Jun-30, 21:42:14)menghu1001 Wrote: It's clearly not ready for publication, and remember the last time I saw that, I pointed it out many inconsistencies, notably the Add health. You should have waited until I finish with the add health and NLSY97.


I fixed the problem with Add health. Check the code. I checked NLSY97. I didn't use it because for student US born there were a lot of "I don't know" probably meaning "Illegal, but I'm not going to say". Look at variable: R1201300, CV_CITIZENSHIP 1997. I'll send you my file. As for NAAL, I'm confident with the results. If I redo them, I will not use the AM software.

(2014-Jun-30, 21:42:14)menghu1001 Wrote: For table 6, GSS, you must have precised the years, e.g., 1972 to 2010 or 1972 to 2012 ?. you must also precise which sampling weight you use, see below, the original configuration (COMPWT).
http://sda.berkeley.edu/cgi-bin/hsda?harcsda+gss12


I originally had that but then removed it when Emil told me to redo the tables. I can put a copy of the analysis in my excel, though. As for the NLSF, I put the syntax at the bottom on my excel. The NLSF wasn't intended to be a nationally representative sample and I noted that it wasn't so I don't see why I should note that there were no sample weights -- that would only matter if it was implied that the sample was representative. Maybe you want me to exclude this sample instead. I was thinking of that myself. But I didn't want an odd number i.e., 17.

(2014-Jun-30, 21:42:14)menghu1001 Wrote: There is also a lot of missing references, such as Capps et al. (2012)."


I don't think that it's necessary to cite every last paper referenced somewhere in a paper. Many people don't do this. I could give you examples. What's the exact policy here? Emil?

(2014-Jun-30, 21:42:14)menghu1001 Wrote: In the paper, table 1 is wide, you can probably rotate it. But the worse thing is that all pictures are blurred. You must do something with that. Copy pasting your pictures from your blog posts does not improve readability. The presentation is that of a complete "amateur" and if you want this journal to be taken very seriously by scientists in general, the presentation is very important. In my opinion, i don't necessarily care so much about presentation, but here, that's a peer-reviewed journal, so you should improve this.]


As I noted to Emil, I will address those issues when I have confirmation about the data in the figures and their overall presentation. I appreciate that reviewers here might be annoyed that I purposely submitted a semi-finished paper, but I'm simply not going to redo the figures a million times according to every suggestion and I knew that there would be numerous objections when it came to the data. There are basic issues such as how to average the data, whether to compute some variance statistic, whether to try to weight the d-values e.g., by overall sample size. I wanted to wait to see if these were a concern first before I address Mickey Mouse (i.e., trivial) concerns.

(2014-Jun-30, 21:42:14)menghu1001 Wrote: In general, I need to replicate all of the data in your file. That will take some time. (We can do the exchange in mail)


I figured as much. I'm not in a hurry. I think Emil et al. are rushing these publications as it is. I'm fine with a couple of months, actually. Slow, slow hippo here. I just want to get the process started.
 Reply
#7
Quote:I didn't use it because for student US born there were a lot of "I don't know" probably meaning "Illegal, but I'm not going to say".

I don't know the slightest thing about the U.S. policies, but what's the relationship between the two ideas, here ?
 Reply
#8
(2014-Jul-01, 03:40:18)menghu1001 Wrote:
Quote:I didn't use it because for student US born there were a lot of "I don't know" probably meaning "Illegal, but I'm not going to say".
I don't know the slightest thing about the U.S. policies, but what's the relationship between the two ideas, here ?


In the '90s we weren't letting everyone who popped up in the U.S. stay. So people were reluctant to admit if they were illegal. That is if (a) they were not US born and (b) if they (or their parents) did not do the paper work. So in surveys people would often say "I don't know" or refuse to answer. "I don't know" mostly equals "No". But you can't say for sure.
 Reply
#9
This is an excellent paper and should be published.
 Reply
#10
Chuck Wrote:I don't think that it's necessary to cite every last paper referenced somewhere in a paper. Many people don't do this. I could give you examples. What's the exact policy here? Emil?


When I submitted my first paper to Mankind Quarterly a year ago or so, I had the same idea that one might just skip the full citations for some well-known papers. In my case I skipped the full reference for "Jensen, 1998" since everybody knows what that refers to.

Reviewers then told me I had to add everything. I think that's the standard policy for papers. I think that when you find papers where the reference is missing in the references section this is due to an error on the author's part, not an intended policy.

I prefer that all sources are cited in full in the reference section. However, I'm willing to take a vote among reviewers to decide the journal policy.

Chuck Wrote:As I noted to Emil, I will address those issues when I have confirmation about the data in the figures and their overall presentation. I appreciate that reviewers here might be annoyed that I purposely submitted a semi-finished paper, but I'm simply not going to redo the figures a million times according to every suggestion and I knew that there would be numerous objections when it came to the data. There are basic issues such as how to average the data, whether to compute some variance statistic, whether to try to weight the d-values e.g., by overall sample size. I wanted to wait to see if these were a concern first before I address Mickey Mouse (i.e., trivial) concerns.


That's another policy question regarding how ready papers should be before submission. We can discuss that in the meta forum if anyone desires.
 Reply
1 2 3 4 5 8 Next   
 
 
Forum Jump:

Users browsing this thread: 1 Guest(s)