STANDARDS
FOR CONDUCTING THE RECENT GRADUATES
EMPLOYMENT AND EARNINGS
SURVEY
November 2015
STANDARDS FOR CONDUCTING
THE RECENT GRADUATES
EMPLOYMENT AND EARNINGS SURVEY
The Recent Graduates Employment and Earnings Survey (RGEES) can be used in an earnings appeal under the Gainful Employment Regulations as an alternative to the Social Security Administration (SSA) earnings data. The alternative earnings survey may be administered to graduates in a specified cohort of students who completed the program and who received Title IV assistance to attend a gainful employment program.
The Recent Graduates Employment and Earnings Survey must be administered to the finalized list of graduates that the U.S. Department of Education (ED) uses to calculate the debt-to-earnings ratios under the Gainful Employment Regulations, as determined under 34 CFR 668.405(c)(4). However, the survey may also be administered to graduates who were excluded by ED due to activities that were assumed to depress their incomes (e.g., enrolled student, military service) (CFR 668.406(b)(3)(i)).
This kind of study, in which every individual on the list is asked to participate, is called a universe survey. For comparability, the Recent Graduates Employment and Earnings Survey at a minimum must be administered to every one of the graduates (i.e., the universe) identified on the finalized list of Title IV-assisted program completers that was used by the Department of Education for matching with the SSA records.
To facilitate data collection for this appeals process, the U.S. Department of Education will make a data collection and processing platform available with the RGEES survey. The RGEES Platform, when loaded onto an institution’s server, can be used for Internet-based administration of the RGEES. The RGEES Platform will assist the survey administrator in monitoring response rates during data collection. Once the data collection is completed, the RGEES platform will do the following:
compile the data;
apply data edits (e.g., top-coding edits on extremely high values, coding negative values to zero);
provide a count of the number of edited records;
compute total earnings for each respondent;
compute the response rate;
compute the mean and median earnings for the responding population; and
compute the nonresponse bias analysis if the response rate is 50% or above.
NOTE: In the standards that follow the requirements that are starred are met with information produced using the RGEES Platform and those using the platform need only address the unstarred requirements. For those electing to not use the platform, the survey administrator must adhere to each of the requirements.
Standard 1:
The survey design components
required to conduct the Recent Graduates Employment and Earnings
Survey include a plan that addresses the objectives of the survey,
the survey design, the data collection plan, the confidentiality
pledge, a data security plan, and the human and fiscal resources and
time needed to achieve high data quality. To meet this standard, the
survey design plan must include the following:
1.1: *A detailed discussion of the goals and objectives of the survey or survey system, including the information needs that will be met, content areas included, the list of program completers to be surveyed, and analytic goals (see Standard 7)
1.2: *A listing of all survey data items. It is preferred that only RGEES items appear in the survey used to collect data for the Gainful Employment Regulations earnings appeal process.
If the survey provided by ED is administered in conjunction with another survey of program completers, the RGEES questions must be used intact. That is, no alteration of the wording of the survey questions is permitted, and the order of individual items must be preserved.
Any additional items must be separate and not applicable to the Gainful Employment issue. These additional items can be added only at the end of the survey to avoid affecting the responses to the questions provided in the Recent Graduates Employment and Earnings Survey.
Only data from the RGEES questions may be used in an appeal.
1.3: The anticipated data collection procedures must include the elements that follow.
Timing of data collection;
Primary mode of collection (e-mail, mail survey, telephone, or in-person interview) (See RGEES Best Practices Guide; for information based on OMB “Guidance on Agency Survey and Statistical Information Collections,” 1/20/2006);
Data collection protocol to be used by data collection staff;
Training of survey collection staff and persons coding and editing the data;
Anticipated response rate for the data collection (see Standard 5.1); and
Methods used to achieve acceptable response rates (see Standard 2).
1.4: *A nondisclosure pledge (see Standard 3.1).
1.5: A security plan for preserving the confidentiality of the data during collection, processing, and analysis (see Standard 3.2 for details).
1.6: If data may be used now or in the future for any purpose beyond the appeal, provide a disclosure analysis plan that describes how disclosure risk will be controlled (see Standard 3.3 for details).
1.7: Outline the quality assurance plan for each phase of the survey process that will permit monitoring and assessing the performance during implementation.
1.8: Outline the general parameters for monitoring nonresponse and evaluating survey procedures and results (see Standards 5 and 6 for details).
1.9: Identify and monitor key milestones of the survey and the time relationships among them.
1.10: An estimate of the target time period needed for the full survey cycle, including the estimated times for the items that follow.
Planning and development;
Data collection;
*Processing and data editing;
Disclosure avoidance plan and analysis, if data will be used for any purpose other than the aggregated data from the RGEES;
File construction;
Survey documentation; and
*Completion and review of the reporting template and any additional internal reports.
Standard 2:
Administer data collection
instruments and methods in a manner that achieves the best balance
between maximizing data quality and controlling measurement error
while minimizing respondent burden using best practices established
in the field of survey methodology.
2.1: Encourage respondents to participate to maximize response rates and improve data quality. (See the RGEES Best Practices Guide for data collection strategies that can be used to achieve high response rates).
2.2: Design and administer the data collection to achieve the best data quality possible. Doing so requires the steps that follow.
Identify the appropriate methods / modes (e.g., e-mail, mail, telephone, Internet) of data collection for recent program completers, the objectives of the data collection, the resources available, and time constraints;
When using the RGEES platform, or any other electronic data collection system, establish protocols for computer systems to ensure that Internet data collection functions correctly (see the RGEES Platform User Guide on how to install/use the platform);
Use more than one mode of data collection to improve response rates and / or if needed to meet response rate requirements (e.g., follow up nonrespondents in a mail or Internet survey with telephone calls). (See Standard 5.2 and the Data Collection section in the RGEES Best Practices Guide for additional guidance on response modes); and
Establish the data collection protocol to be followed by the staff involved in the data collection (see the RGEES Best Practices Guide).
2.3: Monitor data collection activities.
Use internal reporting systems that provide timely reporting of response rates and the reasons for nonresponse throughout the data collection (e.g., paradata, such as refusals, hard to locate cases). These systems should be flexible enough to identify important subgroups with low response rates for more intensive follow-ups. The RGEES Platform has a participation rate reporting function that supports this monitoring.
Specify procedures to use to identify and correct problems. (See the RGEES Best Practices Guide for details.)
Standard 3:
Federal law requires that the confidentiality of data that
contain information about individuals (i.e., personally identifiable
information or PII) must be protected (20 U.S.C. 1232g). Anyone who
will have access to the information must understand the importance of
protecting the confidentiality of the survey respondents’
information, be cognizant of the requirements of the law, and monitor
the confidentiality of PII in their daily activities and in the
release of information to the public. Steps must be taken throughout
the data collection, processing, and reporting activities to ensure
that data are handled in such a way as to avoid disclosure of PII.
LEGAL
REQUIREMENTS:
The Family Educational Rights and Privacy
Act (FERPA) (20 U.S.C. 1232g) and any relevant state confidentiality
or privacy laws apply to personally identifiable information used or
collected in the survey. Data that include personally identifiable
information from students’ education records are protected
under FERPA and may only be used for the evaluation of federally
supported education programs or for conducting studies for, or on
behalf of, educational agencies or institutions to improve
instruction.
3.1: *The Recent Graduates Employment and Earnings Survey uses the graduation/completion status of program participants, student background information, and their contact information from the students’ education records for graduates of a specific Gainful Employment program as part of the evaluation of Title IV-supported gainful employment education programs. This information is protected by FERPA. The respondents must be informed of these protections, and must be assured that their individual data, including their employment and earnings data, will be protected from unauthorized disclosure, will be combined with those of other respondents from their program, and will only be reported in the aggregate when there are a sufficient number of graduates to protect each respondent’s data. Note that the reporting rule in the Gainful Employment Regulations requires 30 cases to report data for a program (Section 668.402).
3.2: Establish procedures and mechanisms to ensure the information is protected during the production, use, storage, transmittal, and disposition of the survey data in any format (paper survey forms, electronic files, printouts). Ensure that personally identifiable survey data are protected from physical and electronic intrusion (see RGEES Best Practices Guide for a template for your Security Plan).
3.3: The reporting rule in the Gainful Employment Regulations requires 30 cases to report data for a program (Section 668.402). If any additional cross-tabulations are released for external use (e.g., using characteristics of the graduates), any table with a cell with 1 or 2 cases must be recategorized to ensure that each cell in the table has at least 3 cases. This rule applies to counts and to the numerator of any computed rates or percentages, and to the number of cases used to compute an average.
STANDARD 4: Data editing will be used to correct inconsistent data in the RGEES data.
4.1: *The edits that follow must be used. No additional edits may be used.
Negative values for any of the income questions will be replaced with zeroes (i.e., survey items 2a, 2b, 3a, and 4a).
When the total income (i.e., sum of survey item responses for 2a, 2b, 3a, and 4a) is calculated, values over $999,999 will be replaced with $999,999.
4.2:* Code the data set to flag each action taken during editing; also retain the unedited data along with the edited data.
Standard 5: Surveys must be designed and conducted to achieve the highest practical rates of response to ensure that survey results are representative of the list of recent graduates who received Title IV assistance so that they can be used with confidence to reflect the employment status and earnings of the Title IV-assisted recent graduate cohort. Nonresponse bias analysis must be conducted when response rates or other factors suggest the potential for bias to occur.
5.1: * A completed survey must include sufficient responses to determine whether the respondent has earnings and to support reporting the respondent’s total earnings (including 0 earnings). The RGEES survey will be considered “complete” if the respondent fills out at least one of the earnings items. The survey response rate is calculated as the ratio of the number of completed surveys (S) to the number of graduates in the cohort (C). Potential respondents that are eligible and not interviewed may include refusals, noncontacts, incomplete responses, or other reasons for nonresponse.
Response Rate = RR = S/C
If cohort members who were excluded from the agreed upon cohort list that the U.S. Department of Education submitted to the Social Security Administration to obtain the cohort earnings data are added, the count of those added cohort members should be added to the number of cases in the cohort (C) and also to the number of completed surveys (S) (see Introduction for a description of excluded cases).
5.2: As a minimum requirement, at least one half of the recent graduates surveyed for a specific program must submit a completed survey through one or more modes of data collection. That is, the response rate for an individual program must be at least 50 percent for the data to be used to support an appeal under the Gainful Employment Regulations appeal process.
Standard 6: The potential impact of nonresponse on the quality of information obtained from the survey must be taken into consideration. Nonresponse bias is a serious problem and occurs when there is a correlation between the likelihood of participation in the survey and the responses to the survey variable(s) being measured. Survey estimates may be biased if those who choose to participate (respondents) differ substantially and systematically from those who choose not to participate (nonrespondents). If these differences are related to employment status or earnings, the results from the RGEES may be misleading or even erroneous. The nonresponse bias analysis can indicate the potential impact of nonresponse bias. A nonresponse bias analysis must be conducted if the unit response rate is between 50 percent and 80 percent.
Nonresponse bias is assessed by comparing respondents and nonrespondents on information available from graduates’ student records. Analysis of recent program level data identified three variables that are readily available in student records and are correlated with earnings: graduates with Pell grants, graduates with a zero expected family contribution, and graduates who are female. A nonresponse bias analysis examines these attributes of the program graduates to determine whether response rates are related to those attributes and/or whether the characteristics of respondents and nonrespondents differ on these characteristics.
If excluded cohort members that are not part on the finalized cohort list that the U.S. Department of Education submitted to the Social Security Administration to obtain the cohort earnings data are added to the list of graduates, the student attribute data for those graduates should be included in the data set prior to conducting the nonresponse bias analysis. (If the RGEES platform is used, data for these graduates should be added to the RGEES data base. If the RGEES platform is not used, these counts should be included in the nonresponse bias analysis.)
The first question in a nonresponse bias analysis is whether the response rates vary across the attributes examined (i.e., do the response rates computed for the group of respondents who received Pell grants differ from those who had a zero expected family contribution, or those who are female)?
The second question in a nonresponse bias analysis is whether there are differences between respondents and nonrespondents on the attributes examined—for example, does the percent of responding graduates who received Pell grants differ from the percent of nonresponding graduates who received Pell grants; are there differences between responding and nonresponding graduates in the percentages of graduates with zero expected family contributions; or the percentages who are female).
6.1: *The basic measure of nonresponse bias is the relative bias, that is, the ratio of the bias in an attribute to the percentage of graduates in the cohort with the attribute(s) measured based on data from the graduates’ student records. The relative bias must be computed for the percent of graduates who received Pell Grants while enrolled, the percent with zero expected family contributions, and the percent female and the average relative bias averaged over these three attributes.
If excluded cohort members that are not part of the agreed upon cohort list that the U.S. Department of Education submitted to the Social Security Administration to obtain the cohort earnings data are added to the cohort, they must be added to the data collection data base before the final nonresponse bias analysis is conducted (see Introduction for a description of excluded cases).
*The nonresponse bias analysis must be conducted both with and without the excluded cases. The relative bias must be calculated as follows:
An estimate of the bias due to nonresponse for each of the three attributes examined can be computed as the product of the nonresponse rate and the difference between respondents and nonrespondents. An estimate of the bias for each of the three attributes examined is given by:
_ _ _ _ _
B(YNR) = YR – YT = (NNR /N) (YR – YNR )
Where:
_
YT = the percent based on all cases;
_
YR = the percent based only on respondent cases;
_
YNR = the percent based only on the nonrespondent cases;
N = the total number of cases; and
NNR = the number of nonrespondent cases.
An estimate of the relative bias is then computed as the ratio of the bias due to nonresponse to the actual value derived from student records.
_ _ _
RelB(YNR ) = B(YNR )/ YT
The average relative bias due to nonresponse, computed as the average of the absolute value of the relative bias due to nonresponse measured for each of the three attributes examined, is used to measure the relative bias due to nonresponse present in of the RGEES data.
_ _ _ _ _ _ _
Average RelB(YNR ) =(│(B(Y1NR )/ Y1T)│ + │(B(Y2NR)/ Y2T)│ + │(B(Y3NR)/ y3T )│)/3
6.2: *In the case of nonresponse to the RGEES, the following decision rule applies:
As a minimum requirement, the average relative bias due to nonresponse (equation 6.1.3) in the data for recent graduates surveyed for a specific program must not exceed 10 percent. That is, the average relative bias computed over the percent of graduates who received a Pell grant, the percent with a zero expected family contribution, and the percent female within an individual program must not exceed 10 percent in order for the earnings data from RGEES to be used to support an appeal under the Gainful Employment Regulations appeal process.
Standard 7: Established procedures must be used to compute the mean and median of the earnings data collected using RGEES.
7.1: *The mean is the average, computed as the sum of all earnings across graduates who responded to the survey divided by the number of graduates who responded to the survey. Nonrespondents must be excluded from the computation, but respondents who reported zero ($0) earnings must be included in the computation. (The mean will be calculated automatically by the RGEES platform. If the RGEES platform is not used, the mean must be computed as described here.)
7.2: *The median is the midpoint of the distribution of all reported earnings, including zeros ($0). One half of the respondents with earnings reported (including zeros ($)) are above the median and one half are below the median. . (The median will be calculated automatically by the RGEES platform. If the RGEES platform is not used, the median must be computed as described here.)
Order the respondents’ earnings from the lowest ($0) to the highest ($999,999). (If multiple graduates have equal amounts of total income, enter that amount once for each graduate with that amount of income.)
Identify the middlemost earnings value which is the median earnings value. If the number of respondents is even, there is no one middle value. In this instance, the median is computed as the mean of the two earnings values closest to the middle.
Standard 8: Complete documentation must be developed for each program that uses the RGEES in an appeal of the program’s graduates’ earnings under the Gainful Employment Regulations. Documentation includes those materials necessary to replicate and evaluate each survey.
8.1: *Survey documentation must, at a minimum, include the elements that follow.
Final data set(s), final instrument(s) or a facsimile thereof;
Definitions of all variables, including coding;
Data file layout;
Descriptions of constructed variables on the data file that are computed from responses to other variables on the file;
List of variables that could be used (alone or in combination) to uniquely identify an individual graduate in the data file;
List of all edits applied to the data;
Frequency counts of all survey items, including counts of the number missing.
8.2: If the data will be used for any purposes other than the appeal under the Gainful Employment Regulations, the documentation must also include the elements that follow.
General description of disclosure avoidance techniques and
Descriptions of restrictions on the use of data (e.g., limited to statistical uses; when using restricted data files, all unweighted counts included in any release must be rounded).
8.3: *System documentation must include the elements that follow. If using the RGEES Platform, and any alternative modes are used, 1 and 2 should be modified to include relevant information for modes other than the web collection.
All instructions to respondents and interviewers about how to properly respond to a survey item or about how to properly present a survey item;
Description of the data collection methodology;
Data editing plan specifications (see Standard 4 and the RGEES Best Practices Guide);
Data processing plan specifications and justifications for why they were implemented;
Response rates; and
Nonresponse bias analysis, if applicable.
| File Type | application/vnd.openxmlformats-officedocument.wordprocessingml.document |
| File Modified | 0000-00-00 |
| File Created | 2021-01-24 |