Evaluating the Integrated Census -2008 Israel Pnina ZADKA

Evaluating the Integrated Census -2008 Israel Pnina ZADKA Central Bureau of Statistics Israel May 12 -15, 2008 1

Content of Presentation § Integrated census – integrated evaluation – Data sources – Methodology in brief § § Evaluation methods Evaluation parameters 2006 evaluation results 2008 Evaluation procedures May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 2

Integrated Census Data Sources Administrative sources (IAF) § Population Register + § Electricity Meters register § Students’ register (Elementary schools, Secondary school and high school) § Vehicles’ register § Geo-coded addresses May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 3

Integrated Census Data Sources Direct Data Collection § Area based sample – Estimating undercoverage within Statistical Area, and collecting socioeconomic information not available in administrative sources. § Population Register Sample – Estimating over-overage within Statistical Area May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 4

Integrated Census Methodology Area sample “U” CAPI survey 11 1 4 5 1 3 2 6 IAF sample “O” CATI survey 2 3 4 4 7 5 5 6 Sampled enumeration cells May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 7 5

Integrated Evaluation Methodology § § Evaluation of the IAF Evaluation of “U” enumeration procedure Evaluation of “O” enumeration Procedure Census weights – reflecting coverage rates Weight =1 - reflecting complete balance between over and under coverage Weight >1 – reflecting more under coverage Weight<1 – reflecting more over coverage May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 6

Evaluation Parameters § § § “Improvement” of residential status “Improvement” of addresses “Administrative Family” construction omitted in the “U” survey & non-response rates Availability of telephone numbers for the “O” sample & non-response rates § Dual model – CV of the final estimates May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 7

2006 Pilot Census § § ≈ 50 thousand persons in the IAF sample ≈ 14 thousand “administrative families” 25 Localities 16 residential institutes (including one prison) May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 8

2006 Pilot Census Evaluation Results § Administrative families – ≈90% correct construction – 0. 5% improper split of family – 1. 4% individuals not unified with their actual family even though living at the same address – 3% improper merge of families – 4. 5% incorrect addresses in the IAF § “Emigrants” – 19 persons not flagged as emigrants (0. 04%) – 9 persons flagged as emigrants improperly (0. 02%) May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 9

2006 Pilot Census Evaluation Results (cont. ) § “U” sample evaluation (“O” as pseudo PES) – ≈ 40 thousand persons were registered in sampled cells addresses – 27. 2 thousand were interviewed in “U” (68%) – 4. 3 thousand living in their registered address(11%)omitted in the “U” enumeration – 5. 7 thousand living at a different address (14%) – 7% not contacted 89% - upper limit of coverage rate for the field enumeration procedure May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 10

2006 Pilot Census Evaluation Results (cont. ) § “O” sample evaluation – 11, 360 families § 1, 413 unknown telephone number – 22, 809 individuals § 2, 032 unknown telephone numbers – 14, 754 full interview (72%) – 1, 524 partial interview or mismatched (7%) – 4, 326 not interviewed (21%) May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 11

2006 Pilot Census Evaluation Results (cont. ) Statistical estimates evaluation § CV calculated by both methods yielded similar results within expected limits. § If combined response rates (“U” + “O”) are less than 50% for a SA the model does not yield valid estimates. May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 12

Implications applied for 2008 § No modification in the Emigrant flagging algorithm § Improve algorithm for “administrative families” § Improve field enumeration results § Improve geo-coding in all sub-files May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 13

Census Evaluation 2008 § All examined evaluation procedures will be repeated § Evaluation of Socio-Economic data with auxiliary sources (National Insurance Institute, LFS, Household Expenditure Survey, Social Survey( May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 14

Thank You May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 15

“Improvement” of residential status The PR consists of all residents with a long term visa who received an ID card (number). Registered residents who live permanently abroad are not deleted from the register. A special algorithm based on border control registrations, updates from the NII, extreme age and 1995 census files, was used to identify “nonresidents” in the PR and flag them in the IAF ☜ May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 16

Improvement” of addresses“ § PR addresses were added from the other administrative files, primary address was selected by an algorithm for the all records IAF. ☜ May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 17

“Administrative Family” construction “Administrative families” were constructed based on kinship relations registered in the PR. Families living in different “improved addresses” were unified using a set of rules and a preference algorithm. ☜ May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 18

Omitted in the enumeration of “U” survey § People registered properly in the IAF who were missed in the “U” survey enumeration and were found in the “O” survey enumeration. ☜ May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 19

O” Sample Survey“ § Administrative families with no telephone number § Wrong number or no response after 15 calls § No response to mail questionnaire ☜ May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 20

Dual model – CV of the final estimates § Impute probabilities for each person in estimation group to reside at their registered address. § Compatibility with current population estimates § Convergence of proportional CV of the estimates calculated using “Taylor expansion” and “Jackknife” methods. ☜ May 12 -15, 2008 UNECE Geneva 2008 Evaluating Integrated Census Israel 21