A study titled "Immigration and American Jobs", written by economist Madeline Zavodny, was released by the American Enterprise Institute and the Partnership For A New American Economy in December of 2011. On page 4 of this study, the second main finding states that "[a]dding 100 H-1B workers results in an additional 183 jobs among US natives." This is expanded upon on page 11 which states:

**
The estimates show that a 10 percent increase in H-1B workers, relative to total employment, is associated with a 0.11 percent increase in the native employment rate. During the sample period of 2001–2010, this translates into each additional 100 approved H-1B workers being associated with an additional 183 jobs among US natives.
**

The data used to reach this finding is explained on page 16 as follows:

**
The Department of Labor publishes data on applications for temporary foreign workers through the H-1B, H-2A, and H-2B programs. Those data are used here for the years they are available: 2001–2010 for the H-1B program, 2006–2010 for the H-2A program, and 2000–2010 for the H-2B program.[29] The measure of temporary foreign workers used here is the number of approved foreign workers in a given state and year. These counts of approved workers proxy for the ultimate number of new temporary foreign workers in each state, since data on actual temporary foreign worker inflows by geographic area are not available.[30]
**

Hence, the study is using the number of approved workers as proxy for the ultimate number of actual workers. In addition, this explanation refers to footnote 29 on page 23 which start as follows:

**
The data are from Foreign Labor Certification Data Center, www.flcdatacenter.com (accessed November 12, 2011), and are for fiscal years. The public-use H-1B data for 2007 contain erroneous codes for the work state, so the analysis here does not include that year.
**

For that reason, 2007 is likewise excluded for this replication. The following table shows the study's key H-1B data by year.

TABLE 1: H1B AND NATIVES: NUMBER EMPLOYED AND LEVELS, BY YEAR year emp_h1b emp_native emp_total pop_native h1b_level nat_level ln_h1b_level ln_nat_level ---- --------- ----------- ----------- ---------- --------- --------- ------------ ------------ 2001 1,432,702 102,771,951 119,489,933 155,359,951 1.1990148 66.15086 0.1815002 4.191938 2002 644,581 101,993,592 118,798,501 157,039,250 0.5425834 64.94783 -0.6114134 4.173584 2003 512,697 101,737,445 119,047,895 159,092,173 0.4306645 63.94874 -0.8424260 4.158082 2004 627,326 102,441,401 120,166,278 160,589,716 0.5220483 63.79076 -0.6499952 4.155608 2005 682,013 104,114,426 122,535,961 162,271,513 0.5565819 64.16063 -0.5859409 4.161390 2006 628,951 105,253,440 124,517,864 163,469,261 0.5051090 64.38730 -0.6829809 4.164916 2008 658,604 105,480,298 124,964,666 165,699,739 0.5270322 63.65749 -0.6404937 4.153517 2009 517,196 101,483,644 120,106,722 167,119,661 0.4306137 60.72514 -0.8425439 4.106358 2010 498,434 100,668,333 119,534,797 167,863,135 0.4169782 59.97048 -0.8747214 4.093852Columns 2 and 3 are the number of employed h-1b and natives and columns 4 and 5 are the total employed and total native population. Columns 6 is the "h-1b level" and equals 100 times the employed h-1b (column 2) divided by the total employed (column 4). Column 7 is the "native level" and equals 100 times the employed natives (column 3) divided by the total native population (column 5). Columns 8 and 9 are just the logs of the levels in columns 6 and 7, respectively. These are the key numbers in the regressions.

As can be seen, employed h-1b (column 2) for 2001 is more than double its value for 2002 and later years. I looked at this several years ago and wrote the following at this link:

**
It should be noted that many applications did request just under 1,000 worker positions. I would assume that requests of 999 and slightly less often served as placeholders when large numbers of positions were desired but the exact number was unknown. The relatively large number of such requests in 2001 caused the total number of work positions to be over 1.4 million.
**

Because of this, the number of workers approved in 2001 may be far greater than the final actual number. In fact, the second table at this link shows that only 331,206 H-1B petitions were approved in 2001. Hence, the Labor Condition Applications (LCAs) may be a very poor proxy for the actual H-1B worker inflow per state since their is no guarantee that states will have similar ratios of certified to actual workers. Finally, it should be noted that this is only a proxy (likely a poor one) for the new H-1B workers. It reveals nothing about the cumulative number of such workers.

In any case, following table shows the results of regressions on the study's H-1B data:

[1] "TABLE 2: RESULTS OF REGRESSIONS ON STUDY'S H-1B DATA " [1] " " [1] " JOBS CORREL " [1] " N INTERCEPT SLOPE CREATED COEF DESCRIPTION " [1] "-- --------- -------- ------- ------- -----------------------------------" [1] "2001-2010, ALL DATA" [1] " 1) -0.3193 0.0219 326.4 0.2989 ln_nat_level ~ ln_imm_level" [1] " 2) -0.3227 0.0170 253.6 0.2989 ln_nat_level ~ ln_imm_level + fyear" [1] " 3) -0.4718 -0.0008 -11.5 0.2989 ln_nat_level ~ ln_imm_level + fyear + floc" [1] " 4) -0.3956 0.0110 182.5 0.2989 ln_nat_level ~ ln_imm_level + fyear + floc, weights=weight" [1] "USING STUDY'S FORMULA" [1] " 5) -0.3956 0.0115 171.6 0.2989 2001-2010, study's data with corrected job count" [1] " 6) -0.4265 0.0081 140.3 0.2439 2002-2010, skip bad data for 2001" [1] " 7) -0.4912 -0.0023 -37.6 0.2177 2002-2008, skip worse years of job loss" [1] " 8) -0.5230 -0.0064 -108.3 0.2142 2003-2006, skip all years of job loss" [1] " 9) -0.5716 -0.0132 -223.4 0.2450 2003-2005, longest span of growing h1b level"Regressions 1 to 4 show the results of the regression after adding each additional variable. The variable fyear is a dummy variable for year, floc is a dummy variable for the state, and weight is a weight used by the study. The weight is based on the native population of each state in each year. Regression 4 is the one on which the study's 183 number is based. However, it appears that Zavodny used the total number of H-1B workers from 2001 to 2010 but the total number of native workers from

Regressions 5 to 9 show the results using the study's formula for different spans of years. The reason for looking at different spans of years is that the study claims to be looking at jobs being created due to an increase in the level of H-1B workers. It therefore makes sense to look at spans of years when the level of H-1B workers is rising. Certainly, one could claim that decreasing the level of H-1B workers has a negative effect on the native employment level. However, this occurred most severely after the 2001 tech crash and the 2008 financial crisis. The loss of jobs for both H-1B and natives was likely due to these economic events. In these events, I don't know that anyone would claim that the drop in H-1B workers was an independent event and that caused the drop in native employment. At the very least, there was likely a very different dynamic at work during these periods and they should be measured separately. In any event, the table shows that, when focusing on those years for which there were job gains for H-1B workers, those gains, on average, were correlated with drops in native employment, at least when using the study's own methodology.

The following table shows the slopes obtained for all spans of two or more years using the same formula that the study used to obtain the 183 figure. The figures in red are negative and indicate a negative correlation between H-1B and native employment level over the given time spans when using the study's methodology.

[1] "TABLE 3: SLOPE BETWEEN GIVEN YEARS (using same regression as was used to obtain 183 job finding)" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "year 2003 2004 2005 2006 2007^ 2008 2009 2010 year" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "2001 0.0071 0.0048 0.0014 0.0049 NA 0.0043 0.0104 0.0115 2001" [1] "2002 -0.0030 -0.0095 -0.0035 NA -0.0023 0.0069 0.0081 2002" [1] "2003 -0.0132 -0.0064 NA -0.0041 0.0061 0.0070 2003" [1] "2004 -0.0104 NA -0.0047 0.0084 0.0079 2004" [1] "2005 NA -0.0018 0.0097 0.0066 2005" [1] "2006 0.0010 0.0073 0.0009 2006" [1] "2007 -0.0192 -0.0152 2007" [1] "2008 -0.0152 2008"The following table shows the corresponding jobs gained or lost using the same formula that the study used to obtain the 183 figure. As previously mentioned, however, it appears that the study incorrectly uses 2000-2010 for the native employment instead of 2001-2010 as with H-1B employment. Also, the study appears to have truncated the above slopes to 3 decimal places as shown in Table 4 of the study. As before, the figures in red are negative and indicate a drop in native employment levels when using the study's methodology.

[1] "TABLE 4: JOBS GAINED/LOST BETWEEN GIVEN YEARS (with incorrect native employment in 1st row and truncation errors in all rows)" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "year 2003 2004 2005 2006 2007^ 2008 2009 2010 year" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "2001 110.7 63.7 15.8 63.7 NA 63.8 162.8 182.5# 2001" [1] "2001 82.8 50.8 13.2 54.6 NA 55.8 144.7 164.2* 2001" [1] "2002 -68.6 -166.3 -66.6 NA -49.6 101.5 138.1 2002" [1] "2003 -236.9 -118.1 NA -83.5 102.7 104.9 2003" [1] "2004 -177.0 NA -80.3 133.3 120.0 2004" [1] "2005 NA -32.0 150.7 103.9 2005" [1] "2006 0.0 121.1 0.0 2006" [1] "2007 -352.0 -294.0 2007" [1] "2008 -294.0 2008" # value is 182.5 if 2000-2010 is incorrectly used for total native employment and the slope is truncated. * value is 164.2 if 2001-2010 is correctly used for total native employment but the slope is truncated. ^ the study excludes 2007 stating "[t]he public-use H-1B data for 2007 contain erroneous codes for the work state, so the analysis here does not include that year."The following table shows the corresponding jobs gained or lost using the same formula that the study used to obtain the 183 figure but with no truncation or other errors. As before, the figures in red are negative and indicate a drop in native employment levels when using the study's methodology.

[1] "TABLE 5: NATIVE JOBS GAINED/LOST PER EACH 100 NEW H-1B WORKERS BETWEEN GIVEN YEARS (using the study's methodology with no errors)" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "year 2003 2004 2005 2006 2007^ 2008 2009 2010 year" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "2001 83.5 61.2 18.9 66.2 NA 59.4 150.6 171.6* 2001" [1] "2002 -51.5 -157.3 -59.0 NA -37.6 115.9 140.3 2002" [1] "2003 -223.4 -108.3 NA -68.6 104.2 122.2 2003" [1] "2004 -168.1 NA -76.2 139.1 135.7 2004" [1] "2005 NA -29.4 162.7 114.9 2005" [1] "2006 15.6 126.0 17.0 2006" [1] "2007 -338.8 -278.6 2007" [1] "2008 -278.6 2008" * value is 171.6 if 2001-2010 is correctly used for total native employment and the slope is not truncated. ^ the study excludes 2007 stating "[t]he public-use H-1B data for 2007 contain erroneous codes for the work state, so the analysis here does not include that year."The p-value is the standard method that statisticians use to measure the "significance" of their analyses. Wikipedia defines it as follows:

**
In statistics, the p-value is a function of the observed sample results (a statistic) that is used for testing a statistical hypothesis. Before performing the test a threshold value is chosen, called the significance level of the test, traditionally 5% or 1% [1] and denoted as alpha. If the p-value is equal or smaller than the significance level (alpha), it suggests that the observed data are inconsistent with the assumption that the null hypothesis is true, and thus that hypothesis must be rejected and the alternative hypothesis is accepted as true. When the p-value is calculated correctly, such a test is guaranteed to control the Type I error rate to be no greater than alpha.
**

Table 2 of the study showed that the regression on which the 2.6 job claim is made had a p-value such that 0.05 < p-value < 0.1. Following is one rough description of how to interpret a p-value:

0.10 < P No evidence against the null hypothesis. The data appear to be consistent with the null hypothesis. 0.05 < P < 0.10 Weak evidence against the null hypothesis in favor of the alternative. 0.01 < P < 0.05 Moderate evidence against the null hypothesis in favor of the alternative. 0.001 < P < 0.01 Strong evidence against the null hypothesis in favor of the alternative. P < 0.001 Very strong evidence against the null hypothesis in favor of the alternative.The following table shows the p-values that correspond to the regression slopes in the prior tables:

[1] "TABLE 6: P-VALUE OF REGRESSION SLOPE BETWEEN GIVEN YEARS (using study's methodology)" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "year 2003 2004 2005 2006 2007 2008 2009 2010 year" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "2001 9.5e-15 4.2e-19 6.2e-21 3.0e-16 3.0e-16 8.3e-17 9.5e-30 2.2e-39 2001" [1] "2002In the above table, p-values showing strong evidence (P < 0.01) are colored red, p-values showing moderate evidence (0.01 < P < 0.05) are colored orange, and p-values showing weak evidence (0.05 < P < 0.10) are colored green. In addition, p-values corresponding to negative slopes are in bold lettering. Looking at the groups of two or more strong (red) p-values, it appears that there is a strong positive association from 2001 to 2010 and from 2002 through 2006 to 2009 and 2010. On the other hand, there is a strong negative association from 2004 and 2005 to 2008 and from 2007 and 2008 to 2010. These conflicting results initially seem difficult to resolve. However, the fact that all of the time spans starting with 2001 have a positive association and those spans from 2002 through 2008 have negative associations would suggest that the former positive association is chiefly due to the steep job losses for native and foreign workers in 2001, the start of the tech crash. Similary, the fact that the time spans from 2001 through 2006 to 2009 and 2010 turn positive in 2009 is likely due to the steep job losses for native and foreign workers in 2009, the start of the financial crisis. On the other hand, the strong negative association from 2004 and 2005 to 2008 occurred during a period of general growth in jobs for foreign workers. Therefore, if the goal is to study the effect of growth in the immigrant level, 2004 to 2008 would seem a better period to since it does not contain a period of steep job losses. During this period, growth in the immigrant level was associated with a LOSS in native employment.0.45491 0.15246 0.43191 0.43191 0.071742.3e-08 2.9e-14 2002" [1] "20030.07563 0.32983 0.32983 0.049663.2e-08 1.7e-13 2003" [1] "20040.01110 0.01110 0.000856.0e-13 1.1e-19 2004" [1] "20050.03515 0.002205.5e-13 1.2e-19 2005" [1] "2006 0.21520 9.7e-09 2.1e-14 2006" [1] "20074.0e-12 2.7e-172007" [1] "20082.7e-172008"

Before looking at the regression that led to the 183 number, it helps to look at the distribution of workers among the states in the following plot:

The numbers to the right of each point give the last digit of the year. As can be seen, many of the states have an extremely large number for H-1B workers for 2001. It's unclear what the exact cause of this was. In any event, the following plot omits 2001 so that the other data for other years is more visible:

The following plot shows the native employment level verus the H-1B Certification level. Note that the native employment level averages around 65 percent. Hence, this is not the native employment rate but a measure of the native employment divided by the native population.

The following two graph show plots of the actual data that is being fit with a regression. The second of the plots show the weighted and unweighted regression of the data. The black line shows the weighted regression that was used by the study to calculate the 183 figure.

The following two graphs show the same data for 2003 to 2006, a time span when the H-1B level was generally increasing.

Note that both the weighted and unweighted lines show a negative slope, counter to the study's finding. It should also be noted that the lines do not seem appear to be the best fit of the data. This is caused by the fact that there are dummy variables in addition to the main independent and dependent variable. The cloud-like appearance of the data in these and the prior two graphs suggest that the data is not strongly correlated. However, it is difficult to judge the exact level of correlation since there are more than two variables.

All of the above arguments apply to the 262 number from the study as well. The following table shows the study's key data for this specific class of foreign STEM worker:

TABLE 7: FOREIGN-BORN STEM WORKERS WITH ADVANCED U.S. DEGREE: NUMBER EMPLOYED AND LEVELS, BY YEAR year emp_imm emp_native emp_total pop_native imm_level nat_level ln_imm_level ln_nat_level ---- -------- ----------- ----------- ----------- --------- --------- ------------ ------------ 2000 148,984 103,082,363 119,167,108 154,182,309 0.1250216 66.85745 -2.079268 4.202563 2001 141,657 102,771,951 119,489,933 155,359,951 0.1185521 66.15086 -2.132403 4.191938 2002 121,521 101,993,592 118,798,501 157,039,250 0.1022919 64.94783 -2.279925 4.173584 2003 151,761 101,737,445 119,047,895 159,092,173 0.1274797 63.94874 -2.059798 4.158082 2004 156,425 102,441,401 120,166,278 160,589,716 0.1301738 63.79076 -2.038885 4.155608 2005 176,080 104,113,290 122,534,825 162,270,376 0.1436980 64.16038 -1.940042 4.161386 2006 175,012 105,253,440 124,517,864 163,469,261 0.1405517 64.38730 -1.962180 4.164916 2007 186,874 105,904,192 125,808,429 164,786,302 0.1485389 64.26759 -1.906909 4.163056 2008 185,667 105,480,298 124,964,666 165,699,739 0.1485759 63.65749 -1.906659 4.153517 2009 203,877 101,483,644 120,106,722 167,119,661 0.1697470 60.72514 -1.773446 4.106358 2010 177,485 100,668,333 119,534,797 167,863,135 0.1484805 59.97048 -1.907302 4.093852 2011 160,208 101,328,629 120,445,461 168,384,994 0.1330134 60.17676 -2.017305 4.097286 2012 199,521 102,432,985 122,263,732 168,723,889 0.1631893 60.71042 -1.812844 4.106115 2013 207,956 103,388,318 123,574,162 169,356,101 0.1682847 61.04788 -1.782098 4.111659The following table shows the results of regressions on the study's data:

[1] "TABLE 8: RESULTS OF REGRESSIONS REGARDING STUDY'S 262 JOB FINDING " [1] " " [1] " JOBS CORREL " [1] " N INTERCEPT SLOPE CREATED COEF DESCRIPTION " [1] "-- --------- -------- ------- ------- -----------------------------------" [1] "2000-2007, ALL DATA" [1] " 1) -0.4284 -0.0073 -481.1 -0.0640 ln_nat_level ~ ln_imm_level + ln_imm_level2" [1] " 2) -0.3998 -0.0054 -356.0 -0.0640 ln_nat_level ~ ln_imm_level + ln_imm_level2 + fyear" [1] " 3) -0.4316 0.0021 137.3 -0.0640 ln_nat_level ~ ln_imm_level + ln_imm_level2 + fyear + floc" [1] " 4) -0.4167 0.0040 263.0 -0.0640 ln_nat_level ~ ln_imm_level + ln_imm_level2 + fyear + floc, weights=weight" [1] "USING STUDY'S FORMULA" [1] " 5) -0.4167 0.0045 293.4 -0.0640 2000-2007, study's data with corrected job count" [1] " 6) -0.5193 -0.0005 -32.2 -0.0299 2002-2005, during first span of increasing immigrant level" [1] " 7) -0.4772 -0.0036 -198.2 -0.1662 2006-2009, during second span of increasing immigrant level" [1] " 8) -0.4862 -0.0020 -121.1 -0.1111 2002-2009, during increasing immigrant level (except 2005-06)"Regressions 1 to 4 show the results of the regression after adding each additional variable. The variable fyear is a dummy variable for year, floc is a dummy variable for the state, and weight is a weight used by the study. The weight is based on the native population of each state in each year. Regression 4 is the one on which the study's 262 number is based. The calculated value was actually just under 263 and the study appears to truncate it to 262. This value, in turn, appears to be based on a truncated value for the slope of 0.004. When using the exact slope, shown to be about 0.0045 above, I came up with a calculated job gain of 293. However, regressions 6 through 9 show that when looking at periods of growth in the level of the foreign stem worker being studied, the level of native workers dropped according to the study's formula.

The following tables show the slopes obtained for all time spans of 3 or more years between 2000 and 2013 and the corresponding native gained or lost:

[1] "TABLE 9: SLOPE BETWEEN GIVEN YEARS (using same regression as was used to obtain 262 job finding) " [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 year" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "2000 0.0034 0.0043 0.0040 0.0038 0.0045 0.0025 0.0015 0.0018 0.0025 0.0036 0.0038 2000" [1] "2001 0.0044 0.0038 0.0036 0.0043 0.0018 0.0008 0.0012 0.0019 0.0031 0.0033 2001" [1] "2002 -0.0005 -0.0009 0.0005 -0.0018 -0.0020 -0.0011 -0.0002 0.0012 0.0017 2002" [1] "2003 -0.0006 0.0001 -0.0027 -0.0027 -0.0019 -0.0008 0.0009 0.0014 2003" [1] "2004 -0.0012 -0.0035 -0.0030 -0.0022 -0.0010 0.0009 0.0013 2004" [1] "2005 -0.0058 -0.0044 -0.0034 -0.0020 0.0001 0.0007 2005" [1] "2006 -0.0036 -0.0030 -0.0020 0.0002 0.0009 2006" [1] "2007 0.0002 0.0006 0.0025 0.0024 2007" [1] "2008 -0.0005 0.0015 0.0013 2008" [1] "2009 0.0019 0.0016 2009" [1] "2010 0.0026 2010" [1] "" [1] "TABLE 10: JOBS GAINED/LOST BETWEEN GIVEN YEARS (using same regression as was used to obtain 262 job finding but with truncation errors)" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 year" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "2000 217.9 284.3 206.2 202.0 263.0# 129.2 62.8 62.2 124.5 183.8 180.8 2000" [1] "2001 286.3 205.9 201.1 261.1 64.1 0.0 61.6 61.7 182.1 179.0 2001" [1] "2002 -67.7 -66.0 0.0 -126.1 -122.1 -121.1 -60.8 59.8 58.8 2002" [1] "2003 -62.7 0.0 -181.7 -176.4 -117.1 -59.0 0.0 57.3 2003" [1] "2004 -120.3 -237.8 -230.5 -172.5 -58.1 0.0 56.4 2004" [1] "2005 -348.9 -281.5 -225.5 -171.7 0.0 0.0 2005" [1] "2006 -222.6 -223.4 -170.8 0.0 0.0 2006" [1] "2007 0.0 0.0 110.9 109.1 2007" [1] "2008 -56.2 55.2 54.2 2008" [1] "2009 54.8 53.7 2009" [1] "2010 109.5 2010" [1] "" [1] "TABLE 11: NATIVE JOBS GAINED/LOST PER EACH 100 STEM WORKERS WITH ADVANCED US DEGREES BETWEEN GIVEN YEARS (using study's methodology with no errors)" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 year" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "2000 245.5 306.9 274.8 257.1 293.4* 158.3 96.3 112.4 155.0 221.8 226.0 2000" [1] "2001 317.8 262.7 242.4 280.2 114.4 46.9 73.2 117.8 188.0 194.5 2001" [1] "2002 -32.2 -59.1 30.4 -110.5 -121.1 -65.4 -12.2 74.4 98.8 2002" [1] "2003 -39.2 3.2 -162.8 -161.6 -108.7 -47.0 50.5 78.5 2003" [1] "2004 -70.6 -206.9 -175.5 -125.5 -57.9 52.5 72.7 2004" [1] "2005 -338.6 -249.7 -189.4 -114.9 5.0 41.3 2005" [1] "2006 -198.2 -168.7 -115.4 8.9 50.6 2006" [1] "2007 13.3 33.9 136.9 132.9 2007" [1] "2008 -27.4 82.8 73.1 2008" [1] "2009 101.5 83.9 2009" [1] "2010 142.8 2010" # value is 263.0 if the slope is truncated. * value is 293.4 if the slope is not truncated.As with the H-1B data, the table above shows that when one looks at a time span for which the foreign worker being studied is increasing, a loss of native jobs is indicated. For example, 2002 to 2009 shows an steady increase in such workers and their share of the employment pool was generally increasing. The table, however, shows a negative slope of -0.0020 during this period, indicating a loss of native jobs.

The following table shows the p-values that correspond to the slopes in the prior tables:

[1] "TABLE 12: P-VALUE OF REGRESSION SLOPE BETWEEN GIVEN YEARS (using study's methodology)" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 year" [1] "---- ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ------ ----" [1] "2000 7.6e-06 2.5e-11 9.6e-16 1.8e-20 2.9e-26 6.3e-31 3.8e-38 1.0e-38 2.4e-35 1.2e-40 1.6e-46 2000" [1] "2001 5.8e-11 9.8e-15 3.7e-19 1.4e-24 3.3e-29 7.1e-36 4.0e-36 8.8e-33 7.7e-38 1.1e-43 2001" [1] "2002In the above table, all of the p-values show strong evidence (P < 0.01) and are therefore colored red. Still, the table does show that the p-value for 2002-2009 is 8.6e-33, a bit stronger than the p-value of 2.9e-26 for 2000-2007, the time span used in the study.1.3e-11 1.6e-168.1e-227.3e-27 8.6e-33 2.0e-32 8.2e-291.8e-33 2.4e-39 2002" [1] "20031.7e-164.9e-221.9e-26 7.9e-31 1.2e-29 8.9e-266.0e-30 1.1e-35 2003" [1] "20042.9e-20 2.4e-25 2.0e-29 3.3e-28 1.6e-249.8e-29 1.0e-34 2004" [1] "20051.4e-22 4.4e-26 7.6e-25 2.3e-211.2e-25 5.1e-32 2005" [1] "20061.6e-23 9.4e-24 4.4e-216.1e-26 8.0e-33 2006" [1] "2007 3.1e-17 1.1e-14 2.1e-19 5.9e-27 2007" [1] "20081.2e-111.4e-16 9.1e-25 2008" [1] "2009 2.5e-15 2.6e-24 2009" [1] "2010 9.2e-20 2010"

**Source code for amstem183.R (processes study's data file to analyze 183 claim)****Source code for amstem262.R (processes CPS MORG data from 2000-2013 to analyze 262 claim)****Source code for amjobsg.R (various functions)****Source code for morg13.R (extracts CPS MORG data from 2000-2013)**

Analysis of "Immigration and American Jobs"

Analysis of the claim that each H-1B worker creates 1.83 jobs

Analysis of the claim that each STEM worker with an advanced U.S. degrees creates 2.62 jobs

References to Claims that Foreign-born Workers Create Jobs

Analysis of "Foreign STEM Workers and Native Wages and Employment in U.S. Cities"

Analysis of "STEM Workers, H-1B Visas, and Productivity in US Cities"

A Look At Mariel Using R

Commentary on the Skills Gap

Composition of STEM Workers in Selected Locations: 2013

Computer Workforce by Age

H-1B Labor Condition Applications: 2001-2013

Information on H-1B Visas

Go to Budget Home Page