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

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

#21
This is a very good paper. However, I think the abstract could be improved. Usually abstracts do not get into technical summaries abruptly. An explanation should be added of what the aims of this investigations were. Ths should be placed between "We conducted an exploratory meta-analysis using 18 samples for which we were able to decompose scores by sociological race and immigrant generation". and "For NH Blacks and NH Whites of the same generation, the first, second, and third+ generation B/W d-values were 0.78, 0.76, and 0.98". An explanation of what the d-value represents should also be added (it may be obvious to you but some readers may ask why you decided to use d values).
Other than this I have no furter comments and I approve publication after abstract will be changed.
 Reply
#22
I fixed the errors noted by Dalliard and Meng Hu; I added to the abstract as requested by Duxide. I corrected some other typos. I redid the tables, except for #2, per Meng Hu's request. Regarding Meng Hu's table format request, I don't think that my screenshot and paste method (see table #2) is out of line with the standards of this journal (or with APA standards). Perhaps Emil could comment. The problem with Meng Hu's method is that it's very difficult to get the spacing right (especially when the columns are long), the tables either end up splitting across pages or taking up too much space, etc., and I end up spending more time on the table format than on the original analysis.

(Also, two reviews, Emil and Dalliard, took issue with my calling this an "exploratory meta-analysis". Emil suggested that it wasn't in fact a meta-analysis because it wasn't systematic. Dalliard noted that the term "exploratory" is ambiguous. I agree with Dalliard and this ambiguity allows me to call it "exploratory". This usage is not just my own idiosyncratic. For example, Jan te Nijenhuis calls these types of analysis "exploratory". He himself told me that I could slap the qualifier "exploratory" onto such meta-analyses to denote that they were not systematic. Per Emil's comment, it's a meta-analysis because I am computing a statistic based on multiple analyses.)


Attached Files
.   U.S. Ethnic-Race Differences in Aptitude by Generation - An Exploratory Meta-analysis (John Fuerst 2014) (07152014) (2). (Size: 807.69 KB / Downloads: 117)
.   U.S. Ethnic-Race Differences in Aptitude by Generation - An Exploratory Meta-analysis (John Fuerst 2014) (07152014) (2). (Size: 165.78 KB / Downloads: 126)
 Reply
#23
(2014-Jul-15, 20:39:44)Chuck Wrote: I fixed the errors noted by Dalliard and Meng Hu; I added to the abstract as requested by Duxide. I corrected some other typos. I redid the tables, except for #2, per Meng Hu's request. Regarding Meng Hu's table format request, I don't think that my screenshot and paste method (see table #2) is out of line with the standards of this journal (or with APA standards). Perhaps Emil could comment. The problem with Meng Hu's method is that it's very difficult to get the spacing right, especially when the columns are long.


I cannot open the file. Can you please reupload a readable file format (PDF or docx)?
 Reply
#24
(2014-Jul-15, 20:54:40)Duxide Wrote:
(2014-Jul-15, 20:39:44)Chuck Wrote: I fixed the errors noted by Dalliard and Meng Hu; I added to the abstract as requested by Duxide. I corrected some other typos. I redid the tables, except for #2, per Meng Hu's request. Regarding Meng Hu's table format request, I don't think that my screenshot and paste method (see table #2) is out of line with the standards of this journal (or with APA standards). Perhaps Emil could comment. The problem with Meng Hu's method is that it's very difficult to get the spacing right, especially when the columns are long.


I cannot open the file. Can you please reupload a readable file format (PDF or docx)?


Fixed.
 Reply
#25
Ok the abstract is much better. I approve publication.
 Reply
#26
I have several requests. In table 16, the row

white 752729

is in the same line as "reference". I think it should be below. (EDIT: forget this comment; after examining it carefully, "reference" was not a part of the text "reading/math d-value". It meant reference groups. But perhaps you can skip a line to help distinguish that)

Concerning table 14, the letters and numbers are smaller than the rest of the text. You should fix it.

Concerning this sentence, page 10 :

Quote:This selectivity could account for some of the
difference. Alternatively, the National IQs of Black majority nations could be underestimated. This issue will require more investigation.

That should be :

Quote:This selectivity could account for some of the difference. Alternatively, the National IQs of Black majority nations could be underestimated. This issue will require more investigation.

As for this :

Quote:Alternatively, the National IQs of Black majority nations could be underestimated.

Are you referring to Wicherts studies on Ss African countries ? If so, you should cite them, because it's not necessarily clear what you're talking about.

Quote:color IQ correlation for second generation Blacks but not for first generation ones.

But that should be "color-IQ corelation".

Quote:Because there were relatively few second generation Black individuals who reported being mixed race, using an inclusive definition had a little effect on the overall scores

You have probably forgotten a dot somewhere.

In your table 3, when you type "reference". I think you should better type "reference group". It's much clearer this way. And you have enough space to write it. And how did you get the values for black-white biracials (e.g., 0.38 for 2nd generation). Given the data file I have, I don't see it. Is it an averaging of ACT and SAT or something ? Because if so, you should have said it explicitly. Because it's impossible to guess.

At page 13 you present the results for Wordsum in the GSS, but have you presented what the "wordsum" test is ?

Quote:Self-reported scores are probably not the best index of true English ability and they may not be comparable across groups (i.e., measurement invariance might not hold);

It's not a point needed to be focused on specifically, but I want to be clear with that. Lot of people don't know what MI is. And I don't understand why Wicherts says that violation of MI implies incomparability of scores. It's false. When analyzing subtests, when they show no MI, that means either the true ability of one group is underestimated, or overestimated. Or neither, for example, if the subtest biases cancel out at the test level, then there is only subtest bias, but no test bias. If biases are cumulative and thus are mainly one-sided, then there is test bias at the total score as well. Even in this case, the scores can be compared. It's just that the scores are under/over estimated. Nothing more. Either with MGCFA or IRT, you can calculate more or less the amount of IQ that is under/over estimated.

Quote:when second+ generation scores are adjusted for psychometric bias in the form of differential item function, the differences remain large (see, Richwine, 2009, table 2.11).

What Richwine shows is that Piat-math is not biased in hispanic-white comparison. But it's just a math test, whereas in your table 9, the tests are mainly verbal/language tests.

This study should be (a little bit) more relevant to your question.

Schmitt, A. P., & Dorans, N. J. (1990). Differential item functioning for minority examinees on the SAT. Journal of Educational Measurement, 27(1), 67-81.

Quote:Because analyses for the SAT-Mathematical test did not demonstrate much DIF for either Hispanic group, Schmitt focused her analyses on the SAT-Verbal test (Schmitt, 1985, 1988). In the first form analyses, 8 and 12 items out of 85 SAT-Verbal items were flagged for DSTD values greater than .05 in absolute value for Mexican Americans and Puerto Ricans, respectively. In the analyses for the second form, the number of flagged items for Mexican Americans and Puerto Ricans were 14 and 16, respectively. It should be noted that most of the flagged items across both Hispanic groups and across both studies had DSTD values whose absolute magnitudes were smaller than .10. Of the four item types on the SAT (antonyms, analogies, sentence completion, and reading comprehension), analogy items exhibited the greatest number of negative DIF items for both Hispanic groups.

Negative DIF means here that the focus groups (the groups for which you suspect there is bias, e.g., minorities) have underestimated scores. 0.10 is the cutoff for large DIF. Given that they do not give the magnitude of total bias, there is no way for me to draw any conclusion on it. But my impression is that the bias is not trivial. This assumes however that all DIFs are one-sided.

Whatever the case, absence/presence of bias in any given test should not be generalized to other tests. You cite Trundt (2013), who has analyzed the DAS-II. But there is no DAS test in none of the samples you analyze.

Quote:For example, Hansen et al. (2008) report scores for male children of natives

It's Hansen (2010), not (2008).

---

Sometimes, the letter "g" is cut, that is, we don't see the the letter entirely. I don't know why this happened. The zeros in table 2 has the same problem by the way (and particularly for the table for asians).

Concerning my suggestion for making tables, I have proposed another one, when i commented on "Genetic and Environmental Determinants of IQ in Black, White, and Hispanic Americans" and you should have already received that mail already.

Concerning the second file (167 kb) i cannot open it.

EDIT:

When you write this :

Quote:For example, third+ generation English-only speaking Hispanics perform only marginally better than all Hispanics of the same generation. TIMSS (2007) results are shown below in table 11.

I don't understand why you did not even report the d effect size.

=(483-478)/((73+67)/2)=0.07

If you do this :

=(533-483)/((67+73)/2)
=(533-478)/((67+67)/2)

the d difference relative to whites who always speak english is a little bit larger, something like 0.11 SD.

Quote:It has been suggested that the African migrant IQ might be on par with that of Whites; if so, the first and second generation B/third+ generation W gaps

Personally, I prefer Black and White instead of B and W, because the "B/third+" is somewhat confusing.

In table 5, the data is from Capps et al. (2012), it is not mentioned at the bottom of the table but in the text instead. Personally, I prefer to see the reference "attached" to the table, it's easier to detect it.

For tables 8 & 15, you must say explicitly, e.g., at the bottom of the table, what's the reference for the numbers in these columns come from (eg. Heoffel et al). No one can guess that. I can say the same thing for table 12. For your table 16, now that I've looked at it once again, I think the IQ for nepalese should be 78.0 and not 78.8. And for your figure 1, if it is based on data given in table 16, you should say it explicitly.

Quote:Globally, Vietnamese, Asian Indians, and Filipinos are estimated to have national IQs, respectively, 0.40, 1.19, and 0.93 SD below the White mean and yet the Californian CAT differences between American Whites and American individuals of these nationalities is, respectively, - 0.13, -0.11, and 0.13.

You must precise what the negative sign would mean. If you have computed white minus asian, and it's negative, it mean those asian groups score higher. But there is no possibility to guess it unless it is stated explicitly.

Quote:The computations are presented in the excel file.

It's just my opinion, but I prefer "presented in the supplementary file".

For table 17, it is difficult for me to follow. Perhaps you can improve it. For example, you do not need to put "White" and "Black" in the line below "non-Hispanic". You have enough space to write them in the same row.

If possible, I would like to see the newer version with modified text in color or in bold. That would help me a lot. And probably others too.
 Reply
#27
(2014-Jul-15, 20:39:44)Chuck Wrote: I fixed the errors noted by Dalliard and Meng Hu; I added to the abstract as requested by Duxide. I corrected some other typos. I redid the tables, except for #2, per Meng Hu's request. Regarding Meng Hu's table format request, I don't think that my screenshot and paste method (see table #2) is out of line with the standards of this journal (or with APA standards). Perhaps Emil could comment. The problem with Meng Hu's method is that it's very difficult to get the spacing right (especially when the columns are long), the tables either end up splitting across pages or taking up too much space, etc., and I end up spending more time on the table format than on the original analysis.

(Also, two reviews, Emil and Dalliard, took issue with my calling this an "exploratory meta-analysis". Emil suggested that it wasn't in fact a meta-analysis because it wasn't systematic. Dalliard noted that the term "exploratory" is ambiguous. I agree with Dalliard and this ambiguity allows me to call it "exploratory". This usage is not just my own idiosyncratic. For example, Jan te Nijenhuis calls these types of analysis "exploratory". He himself told me that I could slap the qualifier "exploratory" onto such meta-analyses to denote that they were not systematic. Per Emil's comment, it's a meta-analysis because I am computing a statistic based on multiple analyses.)


I never meant to say that it wasn't a meta-analysis. Of course it is. I wrote that it isn't a systematic meta-analysis. If studies are heterogeneous, then it is possible to select subsets of studies such as to bias the effect size upwards or downwards. For this reason it is best to be systematic. All it means to be systematic is that there was some objective literature search and clear inclusion and exclusion criteria. The masters of meta-analysis, The Cochrane Collaboration, recommends this practice.

We might call a non-systematic meta-analysis for an exploratory meta-analysis, that seems in line with current practice and is fine with me.
 Reply
#28
In the PDF file, for a reason unknownst to me, some of the table border lines don't appear. I'm not going to attempt to fix this. If that's a problem, Meng Hu can diagnose and fix the issue -- since I redid the tables using his format. I am not going to keep monkeying around with table formats.

I made all of the corrections MH requested except:

(1) "Are you referring to Wicherts studies on Ss African countries ? If so, you should cite them, because it's not necessarily clear what you're talking about."

I wasn't referring to Wicherts. I was simply offering an explanation for why migrants from region x don't perform as their region x National IQs would predict.

(2) "Whatever the case, absence/presence of bias in any given test should not be generalized to other tests. You cite Trundt (2013), who has analyzed the DAS-II. But there is no DAS test in none of the samples you analyze."

There is an implied inductive argument here: (a) There are large unbiased gaps in some tests (PIAT and DAS); (b) there are likely large unbiased gaps in general; © thus, the found gaps based on the tests used are probably not substantially due to bias. You might deem that my evidence for (a) is weak. But I was unable to find better sources; this is what I have. If you can show me other sources -- that show the magnitude of adjusted or MI scores -- I will add them.

(3) "Sometimes, the letter "g" is cut, that is, we don't see the the letter entirely. I don't know why this happened. The zeros in table 2 has the same problem by the way (and particularly for the table for asians)."

I have no idea what you are referring to.

(4) "I don't understand why you did not even report the d effect size...
the d difference relative to whites who always speak english is a little bit larger, something like 0.11 SD."

I didn't feel like it.

(5) "[b]If possible, I would like to see the newer version with modified text in color or in bold."

No.


Attached Files
.   U.S. Ethnic-Race Differences in Aptitude by Generation - An Exploratory Meta-analysis (John Fuerst 2014) (07162014) (2). (Size: 943.7 KB / Downloads: 102)
.   U.S. Ethnic-Race Differences in Aptitude by Generation - An Exploratory Meta-analysis (John Fuerst 2014) (07162014) (2). (Size: 94.56 KB / Downloads: 119)
.xl   U.S. Ethnic-Race Differences in Aptitude by Generation - An Exploratory Meta-analysis (John Fuerst 20147162014) (ODP).xl (Size: 347.23 KB / Downloads: 98)
.odt   U.S. Ethnic-Race Differences in Aptitude by Generation - An Exploratory Meta-analysis (Appendix) (1).odt (Size: 731.41 KB / Downloads: 379)
 Reply
#29
I read the most recent PDF version. Overall good. Soon ready for publication. I have some comments.

Perhaps add "national IQs" to the key words.

In the list of studies on pp. 2-3, the words "publicly" has a grey box. Is that on purpose? What is the purpose?

"Generally, the research which we did find and did not include did not meet one of our inclusion criteria."

p. 3

This seems to imply that you found some research and didn't include it although it met the inclusion criteria. Clarify please.

Is there a list of samples you considered but did not include?

p. 4

"We were unable to compute sample sizes for a number of the studies, as many were analyzed with online statistical tools and as these tools did not provide the necessary statistical options to generate sample sizes; as such, we did not report them in table 2 and we did not weight the survey d-values when computing meta-analytic averages; even if sample sizes were available for all surveys, doing otherwise arguably would have been undesirable given the heterogeneity of the samples, which varied in birth year, age, test type, representativity, and sample size."

Generally people use the median when calculating a central tendency for heterogeneous results (e.g. from many different kinds of methods and samples). Using the median instead means that outliers have no effect on the result which they do on the mean. If outliers are skewed in a certain direction for whatever reason, the mean will be a biased estimate.

However, the median does not work well when K is small. It seems like you used the mean. Did you consider using the median? If you feel like it, you could have a look and see if using the median changes things. My hunch is that it won't change much.

As an example of a study using the median. The IPCC used the median (50th centile) result in literature review of greenhouse gases from different energy sources.

Moomaw, W., P. Burgherr, G. Heath, M. Lenzen, J. Nyboer, A. Verbruggen, 2011: Annex II: Methodology. In IPCC: Special Report on Renewable Energy Sources and Climate Change Mitigation

"are listed in table 1"

Normally one capitalizes the word table because it is a proper name in this context that refers to a specific table. The rest of the paper uses the same practice of not capitalizing references to tables. I prefer capitalization, but it's a stylistic disagreement.

p. 5:

The table does not seem to be a real table as the text does not follow vertical lines exactly. Did you make it look like a table using spaces instead of tabs?

"We reported results for other studies such as TIMSS 1995, TIMSS 1999, TIMSS 2003, and PIRLS 2001 in the supplementary file. We did not include these results in the meta-analysis because we desired a balanced sample of surveys."

How would including them change results?

In a meta-analysis, including many studies of the same type biases the main result in the direction, if any, that the methodology of that type of study biases results in, so wanting a balanced sample is not irrational.

"When sample sizes were too small to generate reliable results, scores were left blank in the chart and were not factored into the meta-analytic averages."

What was the threshold for "too small"?

p. 9:

"Relative to third+ generation Whites, the average d-values were 0.98, 0.80, and 0.98 for first, second, and third+ generation Black individuals, 1.02, 0.68, and 0.56 for first, second, and third+ generation Hispanic individuals, 0.10, -0.21, and -0.19 for first, second, and third+ generation Asian individuals, and 0.20 and 0.04 for first and second generation White individuals. For Blacks and Whites of the same generation, the first, second, and third+ generation B/W dvalues were 0.78, 0.76, and 0.98. For Hispanics and Whites of the same generation, the first, second, and third+ generation H/W d-values were 0.78, 0.65, and 0.56. For Asians and Whites 10 of the same generation, the first, second, and third+ generation d-values were -0.10, -0.21, and -0.19."

Is this paragraph necessary? It is just a complete repetition of the results in the tables just presented.

p. 13:

"Table 5. Percent of Black Immigrants to the U.S. by Region of Origin, 1980 to 2008*"

In this table and others the author refers to the numbers as percentages, but they are not multiplied by 100 and are merely parts of 1. It can throw the reader off.

p. 19:

Is there some reason why there are missing values in SD and IQ columns? Presumably IQs are calculated by converting the Score column values. Clarify?

p. 20:

What are theta scores? Is that from IRT? http://en.wikipedia.org/wiki/Item_response_theory

p 24:

I don't understand how col G works. For Chinese, the prediction based on LV IQ is -.39, while the actual performance is -.46, a difference of |.07|. Very small. Col G says it is .86. Compare with the Japanese below. Predicted -.28, actual -.40, delta |.12|, also small. G says -.01.

What about the three missing values? Presumably the one in F is because the composition of "other Asian" is unknown while "All Asians" uses the estimated proportions from Table 15 to get to .4 ((100-94)/15=.4).

p. 26:

There is a black dot on the right of the regression plot.
 Reply
#30
(2014-Jul-16, 23:43:08)Chuck Wrote: I wasn't referring to Wicherts. I was simply offering an explanation for why migrants from region x don't perform as their region x National IQs would predict.


Ok, but it's not easy to understand.

(2014-Jul-16, 23:43:08)Chuck Wrote: There is an implied inductive argument here: (a) There are large unbiased gaps in some tests (PIAT and DAS); (b) there are likely large unbiased gaps in general; © thus, the found gaps based on the tests used are probably not substantially due to bias. You might deem that my evidence for (a) is weak. But I was unable to find better sources; this is what I have. If you can show me other sources -- that show the magnitude of adjusted or MI scores -- I will add them.


The argument is wrong here. Bias has different sources, and reasons. Difference in speededness, attitude, differential interpretation regarding words, knowledge, etc. Blacks, for example, can show DIF for easy and hard items, but those have likely different explanation, such as easy DIF item owing to difference in interpretation (given that the words are widely known and heard) and hard items due to rarity. Speeded tests can induce people to guess, and more so for members for which the mean score is lower than the other group. If tests differ in properties, they can differ in the amount of bias and in its direction. You cannot imply (a) to say in general there is no bias. You cannot, for example, predict from PIAT and DAS that the ASVAB is not biased. And there is proof that ASVAB is biased, although the author has not made clear the direction of bias.

Gibson, S. G. (1998). Gender and ethnicity-based differential item functioning on the Armed Services Vocational Aptitude Battery (Doctoral dissertation, Virginia Polytechnic Institute and State University).
http://scholar.lib.vt.edu/theses/availab...Sggps2.pdf

---

I cannot open your 2nd and 3rd files.

------

Emil :

Quote:What are theta scores? Is that from IRT?

Concerning HSLS2009, it's not the IRT scores. The syntax look something like :

WEIGHT BY w1student.
MEANS TABLES=x1txmth BY x1race BY gen
/CELLS MEAN COUNT STDDEV.

x1txmth is for X1 Mathematics theta score

whereas

x1txmscr is for X1 Mathematics IRT-estimated number right score (of 72 base year items)

See here (pp 13-14)
http://nces.ed.gov/pubs2014/2014361_AppendixI.pdf
 Reply
 Previous 1 2 3 4 5 8 Next   
 
 
Forum Jump:

Users browsing this thread: 1 Guest(s)