[PPT] - Cam ambodia odia Voter er Reg egis istr try y Audi udit t 20 PowerPoint Presentation

SLIDE 1

Phnom Penh, 21 March 2013

Cam ambodia

dia Voter

er Reg egis istr try y Audi udit t 20 2013

Supported by:

SLIDE 2

Pr Prese esenta ntati tion

n Ou

Outl tlin ine

 Introduction  Methodology & Design of the Audit  VRA Management & Process  Election Data Analysis: Overview of the 2012 Voter Registry

& List of the Polling Stations

 Field Audit findings on the quality of the voter registry:

1.

Comprehensiveness: registration gap, does the voter registry contain all eligible voters?

2.

Accuracy: the completeness of the information of voter particulars and the existence of duplicate records

3.

Currency (Validity): is the information in the voter registry up-to-date and does it contain valid records?

 Voter behavior & opinion which may influence the registry  Conclusions & Recommendations

SLIDE 3

Introduction

SLIDE 4

The e im impor

rta

tance nce of

f voter

er reg egis istr trat ation ion

 Voter registration is a fundamental part of the election process

– voter registry is the key to franchise – and thus should be checked for accuracy and validity.

 The maintenance and upkeep of a voter registry can be

particularly challenging in countries with insufficient records, transient populations or weak infrastructure.

 Voter registries are susceptible to manipulation for electoral

advantage.

 Inaccurate voter registries have led to numerous post-election

conflicts in elections held around the world and have disenfranchised many eligible voters.

SLIDE 5

Rem emem ember er

Voter registration is not perfect. Every voter registration exercise will result in some eligible voters still being

disenfranchised. However, voter registration provides

important and necessary safeguards to ensure that eligible voters will be able to vote.

SLIDE 6

Pu Purpose rpose of

f th

the au e audi dit

 To address concerns raised in regard to the quality of the

voter registry through the undertaking of an independent audit

 To ascertain the quality improvements made to the

registry by the recently concluded voter update and clean up exercise

 To build public confidence in the electoral process

SLIDE 7

Wh What at is is a V a Voter er Reg egis istr try y Audi udit ( t (VRA RA)? )?

 A scientific, systematic, and independent assessment of

the voter registry using a specific statistical methodology

 VRA is scientific and emphasizes facts, and NOT about

perceptions/ opinions.

 A VRA is a method that is used by independent

election observers around the world and has been proven reliable and accurate internationally

SLIDE 8

A VRA cannot…

 Register voters  Speculate about causes or opinions. VRA is scientific

and emphasizes facts.

 Verify whether citizen identification documents were

handed out correctly

 Analyze the quality of elections on election day

SLIDE 9

Methodology & Design of Audit

SLIDE 10

What t is th s the e 2013 3 Voter er Reg egistr stry y Audi udit t (VR VRA) ) meth method

dology?
logy?

1.

Electoral Data Analysis (Partial Computer Audit):

a)

to assess how current or updated the registry was, a demographic comparison was made between the voter registry and current census population projection figures obtained from NEC & the National Institute

f Statistics (NIS)

b)

to make general analysis on the 2012 voter registry and the list of polling stations which will be used in 2013 general elections

2.

Statistically rigorous two-way field test:

a.

List to People  to ensure that every name on the voter registry is that

f an actual person who is eligible to vote and that his/ her information is

correct, as well as deletion list to people test to check whether the deletions records were valid.

b.

People to List  to determine if there are people who are eligible to vote and/ or who appear to have attempted to register to vote, but whose names are absent from the voter registry and to ensure their particulars are correct.

SLIDE 11

Three ee ma main in crit iteria eria for evaluatin luating g th the v e voter er reg egistr stry: y:

1.

Comprehensiveness/Completeness: assess what proportion of eligible voters is included on the voter registry (People to List test)

2.

Currency (Validity): evaluate whether the information in the voter registry is updated in timely fashion (List to People test – by checking the validity of the names on the registry)

3.

Accuracy: measure the rate of error in the data entered for individual voters (Both tests)

SLIDE 12

Sampling process…

Population of interest Sample Sampling frame Inference

SLIDE 13

Sur urvey y Des Desig ign

People to List List to People Population Voting age population in Cambodia a) Registered voters in the 2012 voter registry (for the 2013 elections) b) People in the 2012 deletion list Respondents 2,472 individuals of eligible voting age 1,656 voters from the 2012 voter registry + 765 voters from the deletion list Margin of Error ± 2.0% margin of error and 95% level of confidence ± 2.5% margin of error and 95% level of confidence , ± 3.5% margin of error for deletion list

 Sampling: a statistically representative random sample (nationwide,

urban-rural, gender)

 Total observers: more than 450 observers were trained and deployed

to do face-to-face interviews

 Data collection: February 5-8, 2013 and February 16-19, 2013 using

structured tracking questionnaires

 Coverage area: 24 provinces, 176 districts, 414 communes, 830

villages and reached both urban and rural areas

SLIDE 14

Sam ampli ling ng Al Alloca

cati

tion

n of
f Com
mmu

munes nes

SLIDE 15

Res espondent pondent sel electi ection

n

List st to P

Peo

eople ple

415 communes & Villages selected using a statistically representative sample A computer program selected randomly one polling station in each village & the deletion list associated with the selected location

The trained

bserver went

to the location

f PS

Observer met with the village chief to confirm the location

A computer program selected randomly: a. 4 voters from the list b. 2 people from the deletion list

Observer would attempt to find & meet the selected respondents

Successfully verified, met and interviewed the respondents Can’t met the respondent in person -

Talked & received information

n the respondent’s

whereabouts from relatives, neighbor or village chief

If the selected respondent refused to be interviewed, we randomly identified a replacement Confirm the respondent’s whereabouts

SLIDE 16

Res espondent pondent sel electi ection

n

Peo eople ple to L

List

st

In the selected 415 communes, 415 additional villages selected using a statistically representative sample

The trained

bserver went

to the location

f village

Observer met with the village chief to confirm the location Observer selected the sampling start point Select four households Interval - Every 10th house in rural areas and every 20th house in urban areas

Select randomly a respondent within each household using a ―lucky draw‖ Interview the respondent Two quota respondents

Disabled eligible voters
18 years old or turning

18 before election day

Recently moved to the

village;

Recently died

Check to the registry

SLIDE 17

Li Limi mita tati tion

ns of

s of au audi dit

 Voter registry including the deletion list used for this audit is the 2012

National Voter Registry and the 2012 Deletion List made available for us as of December 31, 2012. Therefore, any changes to the register made after this date are not taken into account by this audit report.

 The NIS figure of the eligible voting age population used to assess how

current or up to date the voter registry is a projection made by the NIS, taking into consideration migration, mortality rate, etc., which means that the figure also contains a margin of error which is yet to be established.

 Because we don’t have access to the whole voter registry in an analyzable

format :

 The registration status was checked only in the location we found the respondents

and the location information given by the respondents

 We can’t do analysis on the duplication rate since we only can check whether the

respondents are still registered in the previous address and not the whole registry

SLIDE 18

VRA RA Man anag ageme ement nt & Pr & Proc

cess

ess

SLIDE 19

VRA RA Man anag agement ement Str truc ucture ture

NICFEC VRA Data Collection

NICFEC Executive Director VRA Project Coordinator 24 NICFEC Provincial Coordinators 414 Interviewers 4,893 VRA Respondents

CAS and NDI Quality Control

CAS and NDI Technical Staff and Master Trainers 52 CAS Supervisors assisting and

verseeing work of 414 NICFEC

interviewers 6 CAS and 6 NDI Auditors conducting parallel interviews of sample of respondents

SLIDE 20

VRA RA Pr Proc

cess

ess

Survey Design & Instrument Development Observer Recruitment

Criteria:
Age from 18 to 45 years old
Minimum of high school

education

Has their own transportation

and mobile phone

Not an activist of any party
Selected 469 (154 female)

interviewers from target communes, 54 reserve observers

Observer Trainings

Conducted 12 trainings, 36 – 45

participants in each training

Training content:
Guidelines for interviewer
Review of questionnaires
Interview methodology
Practice/Role play
Evaluation

Deployment

1st deployment: 5-8 February 2013 (15 provinces)
2nd deployment: 16-19 February 2013 (9 provinces)
Interviewer needs to interview 12 respondents (6 for

L2P and 6 P2L).

Provincial coordinators in each province to monitor
bservers.

Data Reporting & Cleaning

Each questionnaire was checked in multiple levels
Unclear information  call to the respondents/

village chief/ interviewers

Registration status of the respondents was checked on

the VR website (www.voterlist.org.kh) and the PDF both voter registry and the deletion list

Re-check team was doing further data cleaning to the

respondents/ village chief to locate missing info

SLIDE 21

Qu Qual alit ity y Con

ntr

trol

l

Survey Design & Instrument Development

Observer Recruitment Observer Trainings Deployment Data Reporting & Cleaning

Review & recommendations

from stakeholders

Multiple levels review from

election experts

Pilot test on how to

administer the questionnaire

Translation checked by

different people

Longer training period for both

master trainers & interviewers

More interactive trainings
Pre-test and post-test to

identify misunderstanding and follow up with interviewers 1. Multiple levels oversight done by three different levels: 1. NICFEC provincial coordinators 2. CAS supervisors 3. Independent auditors/ spot-checkers by NDI/ CAS 2. Parallel independent audit to 25% of the NICFEC interviewers using the complete questionnaire 1. Re-interviewed more than 7% respondents 2. Found that all interviews had taken place

1. NICFEC checked completeness of all questionnaires 2. Multiple checks using NEC website in location, Voter List PDF and deletion list, using different phonetic spelling to attempt to find voters on the list 3. For abnormal findings, NICFEC, CAS and NDI recheck data by:

Re-check logic and data entry
Re-deploy in Phnom Penh to try to find respondents in

person

For outside of Phnom Penh, directly contacted respondents,

informants (family, village chief, etc.) by phone to confirm whereabouts and status.

For voters not found on the list, re-checked two times by two

different people.

For suspected incorrectly deleted voters, checked in voter list

in the polling station they were deleted from, as well as other possible registration locations.

SLIDE 22

Ele lect ction ion Data a Ana naly lysis: sis:

Ov Over ervie view w of

f the

the 2012 12 Voter er Regi egistr stry

Sources: NEC – 2012 Voter Registration Results NEC – List of Polling Stations for the National Election 2013 NIS – 2013 Demographics Projection

SLIDE 23

Elect ction ion Data a Anal alysi ysis s - Compar ariso son n bet etween en NEC C and NIS Data: a:

Di Dist stribution ribution of

f Votin

ting g Age ge Pop

pulat

ulation ion

Sources: NEC – 2012 Voter Registration Result NIS – 2013 Demographics Projection

SLIDE 24

Elec ectio tion n Data a Analysis ysis

Di Dist stribution ribution of reg egis istered ered voter ers s by Province ince

Total number of registered voters in the 2012 Voter Registry = 9,675,453

Source: NEC – 2012 Voter Registration Result

SLIDE 25

SLIDE 26

Elec ectio tion n Data a Analysis ysis

Reg egistration stration Rate e by Province: vince:

Compa paring ng number mber of regist stered ered vot

ters

ers with h the vot

tin

ing g age populati ulation

n infor
rma

mati tion

Sources: NEC – 2012 Voter Registration Result NIS – 2013 Demographics Projection

SLIDE 27

Elec ectio tion n Data a Analysis ysis – Resu sults lts of the Voter er Registr gistry y Revision sion

Per erce cent ntage age Ne Newly wly Reg egist stered ered Voter ers s & D & Dele eleted ed Voters s by Pr Province ince in th the e 2012 12 Voter r Registr istry

Source: NEC – 2012 Voter Registration Result

SLIDE 28

Elec ectio tion n Data a Analysis ysis Compari parison

n bet

etween en 2011 and 2012 Voter er Registr istry

Per ercentag centage e increase crease in reg egist stered ered voter ers

Source: NEC – 2012 Voter Registration Result

SLIDE 29

Pa Parti tial al Comput puter er Aud udit

Pol

lling

ling Sta Station tion An Analysis lysis

SLIDE 30

Election tion Data An Analy lysi sis

Di Dist stribution ribution of

f 144 Ne