OMB Memo - Sample Boost updated 6-24-13

MEMORANDUM TO: Lynn Murray

Department Clearance Officer

Department of Justice

FROM: William J. Sabol, Ph.D.

Acting Director

Bureau of Justice Statistics

SUBJECT: Submission of burden changes to the National Crime Victimization Survey

Attached please find OMB Form 83-c concerning an increase of 11,545 hours of the current burden for the National Crime Victimization Survey (NCVS). The increase in burden results from a boost of the NCVS sample in 11 states beginning in July, 2013 through December, 2015 to test the feasibility of collecting subnational estimates of victimization.

The 2013 NCVS state sample boost is being conducted to test assumptions about the collection and production of state level estimates, including those pertaining to necessary sample sizes and reliability, expected response rates, victimization rates, state coverage issues, state level sample design and weighting, hiring and training of field representatives, workload and logistic concerns, and cost. The NCVS sample will be increased in 11 states—CA, TX, NY, FL, IL, PA, OH, MI, GA, NC, and NJ—to ensure that the state has representative coverage and that the sample size is sufficient to produce victimization estimates with reasonable reliability and precision.

Initial simulation work by Fay and Li (2012) investigated a boost in all 50 states to determine the proper sample design and reasonable thresholds for considering direct observations as a feasible approach to subnational estimates (http://www.fcsm.gov/12papers/Fay_2012FCSM_I-B.pdf ). The targeted precision is to produce 3-year rolling estimates for violent victimization with a relative standard error equal to or less than 10%. This estimate was determined assuming the national violent victimization estimate and response rate. The rationale for choosing these 11 states was determined based on the current NCVS sample sizes and projections associated with needed sample sizes and costs. Effort needed to boost the sample to meet the established levels of precision are more efficient and economical in the larger states.

The Census Bureau replicated Fay and Li’s simulation and produced estimates for sample sizes, PSUs, field representatives (FRs), and costs associated with boosting sample in 7, 11, and 17 states (7, 11 and 17 represent approximate natural breaks in population size of 10 million or more, 8 million or more, and 6 million or more) (Table 1). Table 2 provides an overview of the estimated sample workload by state by population size that is needed to reach the desired precision under the current design. Please note, these estimates shown in table 2 describe the increase in the specific states or aggregated states and are not for the entire national sample. One constraint with the current sample boost is that the national design could not be changed. In 2016, the NCVS will phase-in a new design based on the 2010 decennial. In addition, the estimates were constructed using the national averages for the victimization rate and response rate. Costs were determined by regional office. These parameters could vary substantially based on the location of new sample. One goal of this state sample boost is to serve as a dress rehearsal for a more comprehensive subnational program and to understand how these factors vary. Obtaining precise cost estimates from Census by task, understanding the complexities and logistics associated with the development of an adaptive sample design that is both responsive to changes in resource availability and stakeholder needs, and developing options for incorporating and validating model-based subnational estimates are critical outcomes of this sample boost. Future decisions to determine the appropriate approach for generating subnational victimization estimates will be informed by what is learned here in terms of controlling and managing survey flexibility, costs, and quality.

The total annual number of respondents will increase by about 21%, from 100,700 to 120,100. Each respondent is interviewed twice per year resulting in an increase of total annual responses from 201,400 to 240,200 (includes respondents and eligible non-respondents). The annual burden will increase by 11,545 hours from 68,905 to 80,450.

If there are any questions concerning this request, please contact Michael G. Planty, Chief, Victimization Statistics, Bureau of Justice Statistics, at (202) 514-9746 or by email at [email protected].

Tabel 1. 17 states in order by population

State/ACS Population

California	36969200
Texas	24774187
New York	19302448
Florida	18688787
Illinois	12790182
Pennsylvania	12660739
Ohio	11525536
Michigan	9920621
Georgia	9600612
North Carolina	9418736
New Jersey	8753064
Virginia	7926192
Washington	6652845
Massachusetts	6512227
Indiana	6454254
Arizona	6337373
Tennessee	6297991

Table 2. Estimated Sample Boost Workloads by State in Population Order (Largest to Smallest):

Pop. Order	State	RO	Total # Current PSUs	Revised Total PSUs	Annual Current Total Sample	Revised Total Sample	Avg Monthly Current Total Sample	Revised Avg Monthly Total Sample	% Increase in Sample	Existing NCVS FRs in Boost PSUs	Expected Non-NCVS Exp. FRs	Expected New FRs	Total FRs Requiring Training	Total FRs Needed for Boost PSUs
1	CA	32	24	26	12727	13576	1061	1131	6.7%	0	0	5	5	5
2	TX	31	46	53	8685	10109	724	842	16.4%	0	3	5	8	8
3	NY	22	29	29	8369	8369	697	697	0.0%	0	0	0	0	0
4	FL	29	24	27	8918	10393	743	866	16.5%	1	0	8	8	9
5	IL	25	21	28	4683	6445	390	537	37.6%	24	1	7	8	32
6	PA	23	20	32	4712	7458	393	622	58.3%	16	0	21	21	37
7	OH	23	36	39	5969	8136	497	678	36.3%	23	0	15	15	38
8	MI	25	18	30	4550	8248	379	687	81.3%	18	0	15	15	33
9	GA	29	32	58	3719	8442	310	704	127.0%	16	13	21	34	50
10	NC	29	16	50	3076	9296	256	775	202.2%	10	12	30	42	52
11	NJ	22	18	19	3421	5408	285	451	58.1%	1	0	17	17	18
12	VA	23	51	66	2496	6460	208	538	158.8%	23	0	29	29	52
13	WA	32	8	15	2743	7496	229	625	173.3%	11	2	18	20	31
14	MA	22	9	14	3351	5954	279	496	77.7%	2	0	15	15	17
15	IN	25	23	47	2453	7691	204	641	213.5%	12	1	22	23	35
16	AZ	31	4	8	3160	7022	263	585	122.2%	4	12	11	23	27
17	TN	23	21	57	2604	8265	217	689	217.4%	29	0	40	40	69
Total			400	598	85636	138768	7136	11564	62.0%	190	44	279	323	513
Top 7 States			200	234	54063	64486	4505	5374	19.3%	64	4	61	65	129
Top 11 States			284	391	68829	95880	5736	7990	39.3%	109	29	144	173	282
Top 17 States			400	598	85636	138768	7136	11564	62.0%	190	44	279	323	513