Back to [Archive] Post-review discussions

1
[OQSPS] Net fiscal contributions of immigrant groups in Denmark and Finland
Admin
Title
Net fiscal contributions of immigrant groups in Denmark and Finland are strongly predictable from country of origin IQ and Muslim%

Abstract
The relationships between national IQs, Muslim% in origin countries and estimates of net fiscal contributions to public finances in Denmark (n=32) and Finland (n=11) were examined. The analyses showed that the fiscal estimates were near-perfectly correlated between countries (r = .89 [.56 to .98], n=9), well-predicted by national IQs (r’s .89 [.49 to .96] and .69 [.45 to .84]), and Muslim% (r’s -.75 [-.93 to -.27] and -.73 [-.86 to -.51]). Furthermore, general socioeconomic factor scores for Denmark were near-perfectly correlated with the fiscal estimates (r = .86 [.74 to .93]), especially when one outlier (Syria) was excluded (.90 [.80 to .95]). Finally, the monetary returns to higher country of origin IQs were estimated to be 917/470 Euros/person-year for a 1 IQ point increase, and -188/-86 for a 1% increase in Muslim%.

Length
14 pages.

Files
https://osf.io/t287j/
Admin
This paper correlates estimates of net fiscal contributions for different immigrant groups with average IQ and percentage Muslim in the origin country. It does so for two small, similar countries in Northern Europe, namely Denmark and Finland. It is a clear and simple analysis, and I have no major points of criticism. I would offer the following suggestions to the author. (Please find grammar/style suggestions in the attached pdf file.)

1. Consider rounding the estimates of net fiscal contribution (and net cost per unit of IQ/% Muslim) to the second significant figure (e.g., –2,238 becomes –2,200). 

2. To give the reader a better idea about the scale of net fiscal contributions, consider providing examples that indicate how large they are as a percentage of Danish/Finish GDP per capita. E.g., –2,238 is roughly 5% of Danish GDP per capita.

3. On p. 2, the author notes "These can broadly be thought of measuring ‘can do’ and ‘will do’ factors (Gottfredson, 1997)". Consider adding another sentence to explain what is meant by this.

4. Add a line of space between the bottom of Figure 1 and the text below. The same goes for Table 1, Table 2 etc.

5. Consider justifying the text, and using a serif font :)
Admin
Noah,

Thanks for reviewing.

I have followed most of your proposed grammar etc. changes.

--


1. Consider rounding the estimates of net fiscal contribution (and net cost per unit of IQ/% Muslim) to the second significant figure (e.g., –2,238 becomes –2,200).


I don't know what you mean. The numbers in Table 3 have no decimals.


--


2. To give the reader a better idea about the scale of net fiscal contributions, consider providing examples that indicate how large they are as a percentage of Danish/Finish GDP per capita. E.g., –2,238 is roughly 5% of Danish GDP per capita.



I would prefer to pave the way forward and leave the economic modeling to actual economists. The values concern only public finances, so comparisons to GDP are not easy.

However, for those curious:

Danish GDP per capita: 47544.82 Euro
Finland GDP per capita: 38327.59 Euro

(From Google, 2015 data.)

Means that the estimates represent:

DK, IQ: 1.9%
DK; Muslim 5%: 2.0%
FI, IQ: 1.2%
FI, Muslim 5%: 1.1%

The values are not so easy to compare due to the scale differences. The range for Muslim% is 0-100, but for IQ only about 70 to 105 (i.e. 35). To set them on the same range, I multiplied the Muslim values by 5, which reduces the range to 20. This makes the values comparable in size.

They are of course not independent, so one cannot just combine them to estimate values for low IQ Muslim countries.

--

3. On p. 2, the author notes "These can broadly be thought of measuring ‘can do’ and ‘will do’ factors (Gottfredson, 1997)". Consider adding another sentence to explain what is meant by this.


This is done in the discussion section, where I write:

The same previous studies have also found Muslim% to be a good predictor. The reason for the validity of Muslim% is less obvious. It is hard to model the predictor jointly with national IQ because the small samples of origin countries make regression model estimates imprecise. A plausible hypothesis is that countries with more Muslims send more Muslim immigrants, and that Muslims immigrants have values that are disharmonious with those of people living in Western countries (see e.g. Koopmans, 2015). The disagreements over preferred policies cause significant outgroup antipathy resulting in crime against the native population and reduced willingness to integrate into the host country’s society and customs. This causal path is more speculative due to a relative dearth of individual-level research on the topic. Unfortunately, it is difficult to find large survey datasets that sample sufficient numbers of Muslims and measure their cognitive ability, religious beliefs, values, cultural practices as well as pertinent socioeconomic outcomes such as crime.



--

4. Add a line of space between the bottom of Figure 1 and the text below. The same goes for Table 1, Table 2 etc.
5. Consider justifying the text, and using a serif font :)


No. :)

--

Files updated.
Admin
I don't know what you mean. The numbers in Table 3 have no decimals.


It isn't very important, but the unrounded numbers seem to suggest a greater level of precision than actually obtains. Significant figures are different from figures after the decimal point.

This is done in the discussion section


I feel a sentence could be added to the introduction, or at least a comment saying something like, "I expand on this in the discussion section".

No.


Surely you agree about the line of space?! :)

Otherwise, I approve the paper for publication.
Admin
Noah,

I prefer not to round number unnecessarily (e.g. for meta-analysis purposes).

--

I have changed the early sentence to:

These can broadly be thought of as measuring ‘can do’ and ‘will do’ factors, and will be discussed further below.

In the discussion, I have added:

The two predictors are broad ‘can do’ and ‘will do’ factors (Gottfredson, 1997). The first concerns what people are able to do; a person with below average cognitive ability will never become a computer scientist even if he wants to. The second concerns what people want to do; a person with a very strong desire to become a computer scientist will never become one unless he possesses sufficient cognitive ability. What people – and groups of people – actually do in society is thus a function of what they can do and want to do.

--

I don't like justified text or extra spacing. :)
Just 2 comments on the paper:

1. In the heading of Figure 6, this should be Finland, not Denmark.

2. You report only zero-order correlations because you think the sample sizes are too small. For the Danish samples, number of countries should be sufficient to predict fiscal contributions in a regression model with IQ and Muslim%. There seems to be much collinearity between these two predictors (does your statistics software calculate a VIF?), but at least such a regression would give an indication of whether Muslim% makes an incremental prediction above and beyond IQ. Otherwise it could be argued that the use of Muslim% is perfunctory if it is just a stand-in for IQ.


Gerhard
Admin
Gerhard,

Thanks for taking a look.

1. In the heading of Figure 6, this should be Finland, not Denmark.


Fixed.

2. You report only zero-order correlations because you think the sample sizes are too small. For the Danish samples, number of countries should be sufficient to predict fiscal contributions in a regression model with IQ and Muslim%. There seems to be much collinearity between these two predictors (does your statistics software calculate a VIF?), but at least such a regression would give an indication of whether Muslim% makes an incremental prediction above and beyond IQ. Otherwise it could be argued that the use of Muslim% is perfunctory if it is just a stand-in for IQ.


I try to live under the principle that one shouldn't look at the results of under-powered models (that's how you get a replication crisis). I believe n=32 is under-powered for 2 predictors, and so I did not fit the model. However, because you requested it, the model results are:

##     ---- Model summary ----    
## Model coefficients
##                 Beta        SE     CI_lower  CI_upper
## IQ          458.7514 223.96243    0.6968428 916.80605
## Muslim_pct -124.9885  43.19323 -213.3285737 -36.64842
##
##
## Model meta-data
##     outcome  N df        R2   R2-adj. R2-cv
## 1 dk_fiscal 32 29 0.5912745 0.5630865    NA
##
##
## Etas from analysis of variance
##                  Eta Eta_partial
## IQ         0.2431751   0.3555178
## Muslim_pct 0.3435352   0.4733390


Thus, IQ is just 'barely significant' per traditional p<.05 standards (p=.04967). Do note that this model is mostly redundant with the analyses already carried out on a more complete dataset with n=70 countries, see:

https://openpsych.net/paper/21

For the current dataset, Muslim% is the superior predictor. This was also true for the prior model fit with n=70. The variable inflation factor is 2. Muslim% and IQ are correlated at -.44 using worldwide data, but -.71 in this sample.

I do not conduct these models because I'm waiting on having access to large multi-national dataset so that I can conduct more powerful modeling. Thus, I'm working towards this dataset by publishing 'meta-analysis fodder' for it.

I added a new para to the discussion:

During review, it was suggested that the author fit a model with both predictors. However, in the author’s judgment, this model would be severely underpowered and the results thus uninformative. Readers may consult the prior study of a larger Danish immigrant dataset which included multiple regression models (Kirkegaard & Fuerst, 2014).
1