3 Concept Correct Sample Size Correct Universe Selection was Random BTQ non-mons & appeals selection Tax has similar Mod 4 for TPS DV Module 4 ensures that the samples selected for nonmonetary determinations and lower authority appeals quality have been randomly selected from the correctly defined universe.
4 ADP (IT) Staff Creates Universe Files ADP (IT) staff creates universe files Non-Mons & Appeals Plain delimited text file Observation numbers and ssn’s Seps / Non-Seps Run at the end of the quarter Same timing as the 9052 / 9054L 3 years if passing, otherwise 1 year Population 5 & 8 must pass Data Validation
5 Sample Size Verify correct sample size 100 or 60 Non-mon count for last calendar year 100,000? Add totals of Sections A & B Intra and Inter-State Obtain values from the State Menu Or use SQL Select sum(c1 + c5 + c97 + c101) From ar9052 Where rptdate between “01/01/2009” and “12/31/2009”
6 Correct Universe Compare number of non-mons reported on the 9052 for the quarter to the number in the universe file. Compare number of appeals reported on the 9054L for the quarter to the number in the universe file.
7 Correct Universe, cont. Determine what was reported Obtain values from the State Menu 9052 for non-mons 9054L for appeals Or use SQL Select sum(c1 + c5 ) seps, sum(c97 + c101) nonseps From ar9052 Where rptdate between “07/01/2009” and “09/30/2009”
8 Correct Universe, Part 2 Did populations 5 & 8 pass Data Validation? Determine the number in the universe file Open the file in a spreadsheet Count the rows / exclude headers The appeals universe file may have some appeals removed. Some ADP (IT) shops remove the appeals that don’t belong in the sample but are counted on the 9054L such as when no testimony was taken. The programmer must then obtain a count for you of those excluded for this reason.
9 Correct Universe, cont. Compare the 2 values Within 2% of reported Formula: Example Universe - reported = difference Difference / reported = Percent Different from reported –14650 = / = =.015%
10 Step 3 Ask ADP (IT) staff how the random selection is made Randomized file Interval
11 Step 3, Randomized A random number is assigned to each transaction. File is sorted by the random number Look for non-random patterns
12 Step 3, Randomized File No systematic review process is possible Observe the file and look for non-random patterns Consecutive numbers All even numbers or all odd numbers Other patterns in the columns Compare to file before randomization
13 Step 3, Interval Determine sample interval Universe size / sample size =N Determine starting number Random number provided by DOL in December Random num * sample interval Round to nearest integer Select every Nth transaction starting with the random starting number
14 Step 3, Interval Sample See the that the correct cases were selected First Case randomly selected Every Nth case selected Match up observation numbers with those in the sample section; prior to importing to the 9056/9057
15 Step 4 Mainly done in Part B For Appeals: Were withdrawals, dismissals and no-shows removed from the universe? Will not match the 9054L Add the number of those excluded
16 Report Results DOL Template: Obtain an MS Word template at:
17 Results Example Results to National Office at
18 Correcting Failures Problems with the universe Reconstruct universe Re-do Mod 4 next year Problems with the random selection Correct and re-do selection prior to BTQ Re-do Mod 4 in three years
19 Review Obtain universe file from ADP Learn how the random sample was selected Compare what was reported on the 9052/9054L to the number in the universe file Verify selection was random Report Results. to the National Office at DOL’s Module 4 website: