Grant Project Translating and Testing a Victimisation Survey Module

Grant Project “Translating and testing a victimisation survey module”

Interim Report

Central Statistical Bureau of Latvia

INTERIM REPORT

Grant agreement No. 11002.2008.003-2008.224

Title:

“Translating and testing a victimisation survey module”

2009 February

TABLE OF CONTENTS

I. / Grant Project “Translating and testing a victimisation survey module”: activities and results of the first phase. / 3
II.  / Grant Project “Translating and testing a victimisation survey module”: activities and results of the second phase. / 3
1.  / Methodology / 3
2.  / General comments on the questionnaire module / 4
3.  / Sample design / 6
4.  / Choice of interview mode / 7
III.  / Grant Project “Translating and testing a victimisation survey module”: activities planned and tasks to be performed during the commoning phase of the project. / 20
ANNEX1 Agenda of the expert seminar / Enclosed
ANNEX2 participant list of the expert seminar / Enclosed
ANNEX3 Detailed comments on the questions in Sections A-F and Section Q / Enclosed
ANNEX4 Questionnaire in latvian Language / Enclosed
ANNEX5 Questionnaire in Russian Language / Enclosed

I. Grant Project “Translating and testing a victimisation survey module”: activities and results of the first phase.

The Central Statistical Bureau of Latvia (CSB) started the project on “Developing comprehensive and coherent EU strategy to measure crime and criminal justice: An EU Action Plan 2006-2010” in September 2008. The main activities include translation of the draft module of victimisation into the national language and testing of it in the fieldwork environment.

During September – November 2008, the CSB performed activities according to the Terms of Reference and adhered to the expected timetable. The first stage of the project was the preparation of the national questionnaire, translating it from the English to the Latvian and Russian languages. Ms. Ieva Peskova, an experienced translator of surveys, did the translation in both languages.

II.  Grant Project “Translating and testing a victimisation survey module”: activities and results of the second phase

In 18 February 2009 the Living Standards Statistics Section of the Central Statistical Bureau of Latvia jointly with experts involved in the project organised a wider seminar of experts, where in the review and evaluation of the translated questionnaire were involved experts from other government institutions, who work daily in the respective field.

The project experts I.Lāce and A.Vilks chaired the meeting.

Participants were informed about the point of the project and the possible survey in the future (in 2013), as well as about the possibility to organize a similar meeting before the survey.

12 experts participated in the meeting (a list of participants is attached in ANNEX 2).

The translated questionnaire was considered (agenda is attached in ANNEX 1). Project experts informed also about the results of cognitive testing. It was decided that the participants of the meeting would send technical corrections to be made in the questionnaire by e-mail to people involved in the project. Thus in the meeting were considered only questions related to the correct translation and interpretation of terms and variants of questions and answers in point of fact. Experts also suggested a possibility to shorten and to improve the questionnaire in the future so as to be able to use it in the survey.

The meeting was recorded in the minutes (minutes are available in Latvian). Suggestions of experts will be included in the project reports – both in the Interim report and in the Final report.

1.  Methodology

In general, it should be acknowledged that the EU victimization survey is a unique research that allows establishing the real level of criminal danger, making a comparative analysis of criminal statistics data, and fixing the common situation with latent crime, as well as the latency of crime by separate kind of offence, studying the reasons of latency. It should be also acknowledged that in the course of research of great importance are correct interviews, processing of obtained data and interpretation of them.

The testing of the questionnaire module involved three steps. First, comparisons of translations of the module from English to Latvian and from English to Russian were revised and compared (English-Latvian, English-Russian, and Latvian-Russian).

Second, the cognitive interviews with selected respondents of different age, gender, educational groups etc. were conduced, 10 interviews in total (see Table1 for aggregate data on respondents). In the process of the cognitive interview, respondents were asked questions as if in a real interview setting. A paper questionnaire was used. Respondents were encouraged to “think aloud” about the questions and the answer options were provided. The aim of interviews was to examine, if the questions and answers are easy to comprehend and are relevant to the situation.

Table1. Data on respondents of the cognitive interviews (n=10)

Number of respondents
Gender
Male / 5
Female / 5
Nationality
Latvian / 6
Russian living in Latvia / 4
Age
Aged 18 - 45 / 6
Aged 46 - 74 / 4
Education
Higher / 6
Secondary vocational / 4
Place of Residence
Capital Riga / 5
Other town or rural area / 5

The average length of an interview was from 50 to 150 minutes. However, as respondents were encouraged to reason aloud about their understanding of questions, the actual length of an interview should be slightly less.

Finally, an expert seminar was held on the clarification of various legal and police work issues arising from the cognitive interviews and localisation of the questionnaire. In the expert seminar participated representatives from the Central Statistical Bureau, the Police Office, the Prosecutors’ Office, the Criminologists’ Association, the Information centre of the Ministry of the Interior (responsible for the official registered crime statistics) and the Public Health Agency, Riga Addiction Prevention Centre. Issues related to legal terminology and legal processes were discussed, as well as methodological and measurement issues regarding victimisation.

1. General comments on the questionnaire module

2.1. Introductory part on the aim of the survey, its international character and guarantees of anonymity should be provided in the beginning of the questionnaire module.

2.2. The structure of the questionnaire seems to be rather complicated:

2.2.1.) Some of the sections overlap to some extent – for example, robbery related questions on violence experienced, injuries, and items stolen in the incident appear both in Section D as well as in Section G. In order to save the interview’s time and prevent the respondent’s annoyance from answering the same questions twice (in the case if the last robbery incident in the last 5 years was caused by an unknown person (R5=3) is the last violent incident the respondent has suffered from an unknown person (Q16)) it should be possible to arrange the structure of an interview so as if the two incidents are the same, the respondent does not have to answer twice the same questions in Section D and Section G.

Suggestion: To rearrange the structure of the questionnaire so that in the case the questions on robbery (R1-R19) are not repeated (Q17 – Q32), if the last robbery incident is from an unknown person (R5=3) and is the last violent incident from an unknown person the person has suffered from (Q16). This could be done, by inserting a probing question Q16A, if this was the event the respondent had told about before as robbery.

2.2.2.) Subsections on the ex-partner violence as well as on violence by acquaintances are overlapping. It might be that for the measurement reasons the indicators of the existence of violent incidents caused by ex-partners and other acquaintances is important, but it is rather annoying for respondents to look at the types of violence for 4 times.

Suggestion: If for the measurement reasons it is not inevitable and crucial to record the ex-partner and acquaintance-caused violent incidents separately, these sections could be integrated into one by adding an answer “ex partner (ex-boyfriend, ex-girlfriend)” to the list of answers of Q93.

2.2.3.) Questions Q109 and Q110 (on experiences of sexual harassment before the respondent reached the age of 15) should be moved to the respective section (as for now Q1 –Q8 on the experiences after the age of 15 and the most recent experiences).

2.3. Changes in the time limits in measuring some indicators in the questionnaire should be considered:

2.3.1.) Victimisation experiences longer than 5 years are rather difficult for respondents to remember correctly, and therefore the measurement can reveal data biased as to age and specific processes of memory of the respondents. The cognitive interviews with people in different age groups revealed that events that had happened longer in past easily faded, and respondents did not remember details and did not consider incidents worth mentioning. It might lead to false conclusions that some time ago there was much less violence than nowadays. On the other hand, more recent events (last 5 years) allow for a more reliable estimation of data (as the time period is rather limited, both the young people and the elderly will recover their experiences in a more reliable way). The opportunity to detect the general victimisation rate of a person is lost in this way, but the cognitive interviews revealed that younger respondents tended to remember more events than elderly, and the data would be biased anyway.

Suggestion: In Section Q, questions on sexual harassment and violence should be asked (except Q108 - Q110) only for the time period of last 5 years and last 12 months, so that all the respondents cover the same time (instead of the time period since the person was 15 years old).

2.3.2.) The time limits for indicators in Section C (C10, C15, C20, C25, C30, C35, C40, C45, C50, C55) are rather confusing and with high requirements for the respondents’ memory, and thus time consuming in the interview. Besides, the data obtained would heavily depend on the time of the year when the survey is conducted – if it is January-March, the possibility of some incident having occurred “This year” will be much less than if the survey would be conducted in the end of “This year”, in November or December. In this case it will also be much more difficult for the respondent to recover the incidents of the “Last year” and distinguish them from “Earlier”.

In Latvia now the level of detail in the questionnaire would not provide any added value for the measurement and comparison with the official statistics. The official statistics is arranged according to the date of registration of a criminal offence, not with the date of occurrence, so these two dates can differ. Therefore the data of the registered criminal offences in a time period in Latvia will not be comparable to the data on registered crime.

Suggestion: If the comparability with the yearly official crime statistics is not so important and is not reasonable, for the sake of simplicity and saving time, C10-C13 (C15-C18, C20-C23, C25-C28, C30-C33, C35-C38, C40-C43, C45-C48, C50-C53, C55-C58) should be changed as follows:

C10. When did the most recent incident occur? 1)During the last 12 months, 2)Earlier, more than 1 year ago to 5 years ago?

C11. How many times has it happened during the last 12 months?

C12. How many times has it happened earlier, more than 1 year ago to 5 years ago?

2.4. Questions related to religious affiliation are not permitted in official statistical survey forms in Latvia. Comment: Q111-Q116 might not be permitted to ask.

2.5.In the existing form, the questionnaire module does not prescribe, in which cases the answers to the question should be read aloud for the respondent, and in which cases the respondent should give a spontaneous answer, which is later coded by the interviewer according to the answers given. It is important for the measurement and interpretation of data. On the one hand, spontaneous answers are better, as they are closer to the respondent’s experience, but on the other hand, in some situations respondents do not think themselves of all the relevant answers or cannot formulate them, so reading aloud helps to clarify. For example, in questions like B2, B15, we believe, the research would gain more from spontaneous answers of respondents coded by interviewers into the relevant answer category rather than from the answers read aloud for respondens to consider all of them.

Suggestion: It is important to revise the questionnaire and decide for the questions, where it might be important, whether to read the answers aloud for the respondent or the respondent gives a spontaneous answer and the interviewer codes it.

2.6.Several questions in Section G and R1-R19 need answers “refuse to answer”, “don’t remember”, and other (see the detailed comments in the respective sections).

2.  Sample design

In accordance with the application for the project, in the pilot survey the questionnaire will be tested both CAPI and CATI. It is planned to carry out 150 CAPI interviews and 100 CATI interviews.

Besides, it is envisaged that in the case of face-to-face interviews (CAPI) part G is filled in by the respondents himself. As we cannot be sure of the respondent’s computer skills, a possibility is provided to choose, whether the respondent is able to fill in the electronic questionnaire on a computer or a paper questionnaire is filled in.