Directorates
·	Director General's Office
·	Data Development
·	Strategic Evaluations
·	Partnership and Continuous Evaluations
Publications
·	Evaluation Reports
·	Evaluation Briefs
·	Labour Market Development Agreement Series
·	Lessons Learned Series
·	Evaluation Tool Kit Series
·	Evaluation Forum Newsletter
·	From Theory To Practice

The evaluation of the Small Weeks Pilot Project relied primarily on Human Resources Development Canada (HRDC) administrative data. Fundamental claim data, including summary claim information, personal attributes, and geographic location, were derived from the Status Vector file. This file is maintained by Data Development for purposes of research, analysis and evaluation. It is derived from the Benefits and Overpayments File (the file used by HRDC to administer unemployment claims and pay benefits) and is updated quarterly.

The evaluation also required information from claimants' Records of Employment (ROE). The ROE file maintained by Data Development was used but was supplemented by more detailed earnings data from HRDC Systems. Specifically, Systems provided data from the Support System for Agents (SSA) database.

SSA is a microcomputer-based system tool used by insurance staff to assess claims for Employment Insurance benefits. It has two main functions: (1) a rule based program to assess and calculate a claim for benefits based on current policy and procedures; and (2) a user interface with the On-line Insurance System (OLIS). SSA input data and resulting calculations are stored on the LAN of the local Human Resources Centre (HRCC) offices.

Calculation of a small weeks claim relies heavily on the SSA facility. The system retains information on earnings in each week of the rate calculation period. The weekly earnings are derived from Block 15C of the ROE, the application for EI benefits, the small weeks questionnaire, from earnings while on claim, or from pay records of the claimant or the employer. The agent reviews the weekly earnings to distinguish small weeks from regular weeks, to compare total weeks with the regional divisor, and to identify those small weeks eligible for exclusion. The weekly earnings and the results of the agent's assessment are stored on the LAN.

Two years ago, during an initial evaluation of the 1997-1998 Small Weeks Adjustment Projects, Evaluation and Data Development (EDD) obtained the weekly earnings data from the LAN of each local office. Close to 100 offices forwarded small weeks data to NHQ. Each file was converted from the SSA software to an ASCII format using a program written by Systems. Once converted, the local office data were transferred from microcomputer to the mainframe for processing at EDD.

In 1999, Systems began to centralize the small weeks files. They gradually transferred the data from the local offices to the four Information Technology Centers (ITCs) across the country. In August 2000, about half of the transfer was complete.

For the present evaluation, it was necessary to acquire the data using two approaches. For those offices whose data had not been transferred, EDD followed the same method as in the initial evaluation, namely receipt directly from the local office. For those offices whose data had been transferred, a new set of procedures was implemented. Systems wrote additional programs to extract the necessary data from the SSA database and to convert the data from a UNISYS format to an IBM readable format. Systems applied rigorous standards to their extract programs before forwarding them to the Information Technology Centers for execution. They put the programs through a series of tests using a working data file and also required EDD to apply a further set of tests to ensure that the files were readable and contained the requisite information. Once the Systems' program was approved, program documentation was prepared and an implementation date was arranged with the ITCs. The program was run at the ITCs and the data were copied to cartridges and forwarded to EDD.

EDD combined the files from the four ITCs and the approximately 45 non-transferred local offices. At this stage, the combined file contained administrative data — not research data. It was data based on working files, files where fields could be used to flag a potential situation but where agents would not necessarily go back and make a correction to the flag if the situation was not realized. The implication was that the data could not always be taken at face value. Furthermore, the data were not in a format for analysis. Earnings were reported by ROE, not by rate calculation period. Substantive reformatting had to occur to isolate weeks within the rate calculation period, to combine earnings from multiple jobs, and to identify weeks between ROEs with no earnings. Reconciliation of ROE and BNOP total earnings with weekly earnings presented another challenge. Considerable data “cleaning” had to occur before the file was ready for analysis.