SS A1, p1 – where BJS says “race categories” please say “race/ethnicity categories”
Response: Revised.
How does BJS create national estimates from data collected ONLY from the certainty strata? Has it done the research to be able to safely assume zeros for the smaller jurisdictions?
Response: The items collected from the certainty jail jurisdictions (time served, staffing, inmates misconduct) are not intended to generate national estimates. The resulting report will clearly state that these data only represent the 300+ jail jurisdictions. Although these jail jurisdictions account for just 10% of the jail universe, they account for about two-thirds of the jail population.
What is the practical difference between being in a certainty strata and being in a non-certainty strata once sampled since the sample seems to have been held constant since 2006 and is proposed to stay that way for another 3 years?
Response: Jails in certainty strata are self representative and jail jurisdictions in non-certainty strata were randomly selected to represent similar size (based on average daily population) jail jurisdictions. National estimates are generated from the self representing jail populations (certainty) and the weighted non-certainty population counts.
If by “practical” we mean “the way it looks to the respondent,” isn’t the answer “no difference?”
If we look at this from the respondent perspective then there is no difference in the sense that every year since 2006 both certainty and sampled jurisdictions have been surveyed. However the certainty jurisdictions are asked to fill a different questionnaire from the sampled jurisdictions. From the analyst and BJS perspective the two types of jurisdictions are clearly distinct in that analytically they are handled quite differently as we have explained above.
Why has the sample (of non-certainty strata cases) not changed since 2005? The whole point of sampling smaller jurisdictions in most collections is to acknowledge the disproportionate burden on them. By keeping a jail in sample every year for 10 years (by the time you propose to swap the sample) seems excessively burdensome. Is this just BJS/Census not wanting to expend resources to redraw the sample? Were these jurisdictions and/or their associations made aware that they would be in sample for such a long time? How many of the 34 nonresponding jails are in the smallest strata (4, 5, 9 and 10) and what has been their pattern of who is a nonrespondents since 2005?
Response: The panel survey is intended to remain constant until the next sample can be drawn from a Jail Census. Annually the frame is updated to reflect closures and opening in jails jurisdiction included in the ASJ. Ideally the sample refresh would have occurred 5-6 years from 2006 ASJ, but a census type collection (expanded jail DCRP) is planned for yearend 2013 in order to draw a new sample in 2015.
are you saying that the jails were supposed to be in sample until there is a census but that the census was delayed, therefore keeping the original jails in sample an extra several years? also, please respond to the last question in the original item 4.
This is not a rotating panel survey, but a simple panel survey. Panel surveys are the design of choice when the primary goal is monitoring change in some key statistics/measures over time. We did not explicitly address with the respondents about the unanticipated additional years of data collection. However, in the letter to respondents we ask for their time and participation in the current data collection.
The survey response rate was 100% each year between 2005 and 2009 and 99% in 2010. The survey response rate was 96% in 2011 and 94% in 2012. For the most part the non-responders (NR) were unique year-to-year. In 2012, 2 NR were from stratum 4, all responded from stratum 5, 6 NR from stratum 9, and 3 NR from stratum 10.
We apologize for the lack of clarity in our writing and responses. The panel never meant to be an “indefinite” panel. We do have a plan to refresh the sample and in the SS A, page 14 under item 6 we now write the following:
“Because of limited resources and budgetary considerations, the current ASJ panel has been utilized longer than the traditional 5 to 6 years. BJS plans to keep the current panel of jail jurisdictions in the sample until completion of the 2014 Annual Survey Jails. BJS plans to conduct a Census of Jails for yearend 2013. This new Census will be used to design and draw a new sample for ASJ. BJS expects to use the new sample for the 2015 ASJ.”
What is the specific use to which BJS will put the data from the additional California jails if not being used to adjust the national estimates?
Response: The data from the additional California jails will address two major goals. 1. The additional data from California will give us a census of the California jurisdictions. This means we will obtain a better picture of the total number of jail inmates given the recent changes due to the Public Safety Realignment (PSR). 2. Having the data on the non-sampled California jails will allow us to correct the sampling weights so that the similar sized non-California sampled jurisdictions do not represent them and vice-versa. Because of the PSR, the sampled California jurisdictions were weighting up for some non-California jurisdictions (and vice-versa), therefore the total number of jail inmates was over-estimated.
SS p6 talks about some stakeholder meetings. If BJS plans to use a structured interview/focus group protocol approach, it should request clearance (which we are happy to provide) for such interactions.
Response: BJS is referencing the use of stakeholders in the previous ASJ/SJIC OMB clearance cycle. There are no plans in the current cycle for structured interviews/focus groups.
There seem to be some difference between SS A and B about item nonresponse rates. From SS A, p7 for example, it looks like there may be at least one item with nonresponse rates around 35% -- I expected to see some discussion of assessing bias in this/these estimates but Part B claims that no items has non response rates above 10%.
Response: In SS A, p7, the lowest reference response rate is 55% and is connected to the certainty only items. I revised the section in SS B to address the statement on non response rates above 10%. The 10% excludes items about the distribution of time served, staffing, and inmate misconduct.
Rather than excluding “distribution of time served, staffing, and inmate misconduct” from your discussion of item nonresponse, please discuss what is done to assess nonresponse bias for these items.
Currently, the resulting report excludes findings from the expanded information collected on time served, staffing, and inmate misconduct. We are in the process of analyzing the differences between the jurisdictions that responded to these questions versus those that did not in the 2010, 2011 and 2012 surveys. These preliminary analyses will be used to develop a methodology to address these issues.
Again we apologize for the lack of clarity and confusion. The SS A on page 19 states that we have plans to publish a special report on staff safety and security in 2014. We have also revised further this section to elaborate our analysis plan. Due to constraints on time and resources, a 2013 publication is not expected. However, the expanded data has been already used by researchers to describe staff safety and security. The researchers utilized the 2011 ASJ jail misconduct items to draw a relationship between misconduct and facility characteristics. http://johnjayresearch.org/rec/files/2012/11/asc2012cubellis.pdf.
The SS we now reads: “ Included in Among the indicators suggested as measures of safety and security are those related to safety of inmates and staff and inmate misconduct. These comprise the core measures in ASJ. Over time, as the number of years of data collected on safety and security increase, BJS plans to report annual statistics on safety and security. These statistics include counts, rates, and distributional measures.
So far BJS has created the panel for the 2010 and 2011 dataset for analysis and will add the 2012 data in the next several months. Once this step is complete, BJS will evaluate the response rates for all three years and address data quality issues. The preliminary analyses will be used to develop a methodology to address these issues. Next steps will include work on the model specification and estimation methods with the goal to publish a special report on staff safety and security in 2014.”
Please provide information in SS A3 about the percentage of respondents who use web and how those changes are trending.
Since 2000, BJS has offered a web reporting option for respondents and will continue to do so. Since the inception, the use of the web option has increased steadily, from 12% in 2000 and 27% in 2006, to about 55% in 2011 and 2012.
Please provide sample screen shots of the web version of the new instrument or at least of the previously used version.
See attachments.
To clarify on confidentiality – does BJS make jail-level data publicly available since it claims that the data are in the public domain and since the only confidentiality is pledged to individuals? I’d like to be clear on what is protected versus what is just a practice.
Response: Yes, BJS makes reporting unit (which are jails, but is the central data reporter) publicly available for statistical and research purposes only. Based on our pledge, which I updated in confidentiality statement, respondent name and contact information is excluded.
Assurance of Confidentiality BJS’ pledge of confidentiality is based on its governing statute Title 42 USC, Section 3735 (see attachment 10), which establish the allowable use of data collected by BJS. Under this section, data collected by BJS shall be used only for statistical or research purposes and shall be gathered in a manner that precludes their use for law enforcement or any purpose relating to a particular individual other than statistical or research purposes. BJS staff, other federal employees, and BJS data collection agents shall not use or reveal any research or statistical information identifiable to any specific private person for any purpose other than the research and statistical purposes for which it was obtained.
Was the 2005 census conducted in 2006? SS A6 says the 2006 census was basis for sampling, but SS B1 seems to say it was conducted in 2005 and I’m trying to reconcile them.
Response: Two jail census’s were conducted between 2005 (Census of Jail Inmates) and 2006 (Census of Jail Facilities). The ASJ sample, as indicated in SS B, was drawn in 2006 and is based on the 2005 Census of Jails inmates. The reference in SS A is only noting that the last census was conducted in 2006 (Census of Jail Facilities).
| File Type | application/vnd.openxmlformats-officedocument.wordprocessingml.document |
| Author | adamsd |
| File Modified | 0000-00-00 |
| File Created | 2021-01-29 |