Using the information that users have shared publicly on trackitt.com, I made some graphs and did some basic hypothesis testing. In order to get the data, I simply scraped all the webpages using R, and cleaned the data by removing incomplete cases.

Percentage of applicants in each category, and the correspondent percentages of accepted, pending, and denied cases

##     types percentage accepted pending denied
## 5    EB1A      3.358    71.84  17.352 10.350
## 6    EB1B      3.900    92.14   5.505  2.359
## 1    EB1C      7.498    82.41  15.951  1.431
## 2     EB2     58.137    91.59   7.104  1.090
## 3 EB2-NIW      6.328    83.84  13.651  1.939
## 4     EB3     20.778    89.37   7.577  2.706

Countries leading in the number of applicants

## EB2-NIW
## 
##       India       China South Korea        Iran      Taiwan    Pakistan 
##      22.536      16.721       9.855       7.431       4.281       2.342
## All categories
## 
##       India       China South Korea    Pakistan Philippines      Canada 
##      63.356       6.226       2.730       2.152       1.983       1.769

How about checking if there is any significant difference between countries, in NIW category?

##         types percentage accepted pending denied
## 1 South Korea     15.601    87.70  12.295  0.000
## 2       India     35.678    82.08  15.054  2.509
## 3       China     26.471    89.37   8.696  1.932
## 4        Iran     11.765    79.35  18.478  2.174
## 5      Taiwan      6.777    86.79  11.321  1.887
## 6    Pakistan      3.708    79.31  13.793  6.897

Seems like there is a difference between Pakistan and South Korea (pvalue = 0.0383)! But if we correct for multiple-hypothesis testing, the difference is not significant.

What about for all categories?

##         types percentage accepted pending denied
## 1       India     81.002    89.05   8.770 1.9524
## 2 South Korea      3.490    90.64   8.989 0.1873
## 3    Pakistan      2.751    85.99   8.789 3.8005
## 4 Philippines      2.536    82.99  14.948 2.0619
## 5       China      7.960    90.48   8.128 1.3957
## 6      Canada      2.261    91.33   7.225 1.1561

And there is no difference between any other countries.

Focusing on the approved case, I was curious to see what’s the difference between the total time it takes to be approved between different categories. Categories are further divided by their processing speed.

plot of chunk unnamed-chunk-14

Obviously, premium processing is way faster than regular processing, but alas, it’s not available for EB2-NIW!

Focusing on EB2-NIW, we see that the processing speed has a clear periodic pattern! plot of chunk unnamed-chunk-15

Does total time it takes to be accepted depend on nationality? plot of chunk unnamed-chunk-16 It seems like Chinese, Indians and Pakistanis have a longer wait time than Iranians, South Koreans and Taiwanese. Not sure if it’s because of the priority date, or what.

Again in EB2-NIW, you can either apply for I-140 concurrently with I-485, or not. But would that change the processing time, or the likelihood to be approved?

## Concurrent application
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##      10     124     190     235     327    1040      55
## Non-concurrent application
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##       5     145     218     257     374     877     117
##            types percentage accepted pending denied
## 1     concurrent      40.63    87.48   10.54  1.392
## 2 non-concurrent      59.37    81.36   15.78  2.313

Applying concurrently decreases the process time for NIW I-140 (pvalue = 0.022), for almost 4 weeks. But how does it affect the chance of being rejected? It seems like the rate of rejection goes up to 2.3%, from 1.4%, if you apply non-concurrently (but that difference is not significant).

So how do pending cases accumulate over time? plot of chunk unnamed-chunk-18 It seems like the number of pending cases in 2014 is significantly higher than the last 6 years. Isn’t this the underlying reason we are here now?

How about RFE? How does receiving one changes the chance of getting approved?

##   types percentage accepted pending denied
## 1    no      79.48    84.65   14.33 0.5081
## 2   yes      20.52    80.71   11.02 7.4803

If you don’t get RFE, chances to be rejected is 0.5%. If you get a RFE, rejection rate goes up to 7%! And the difference is significant (pvalue = 0.0228). About the application center, there is no significant difference between Nebraska and Texas centers, in terms of the outcome.