The quantitative study on knowledge, attitudes and practices of the Georgian general public regarding violence against children took place in summer 2012. The research organization ACT Research was selected as a partner to undertake the preparation, fieldwork and provide UNICEF with a cleaned and weighted data set.

Questionnaire

A questionnaire was designed by UNICEF Georgia based on instruments used in various studies on violence against children. Please see Questionnaire in Annex 2. The instrument was piloted and administrated by ACT Research in multiple settings and covered Tbilisi and all 10 regions under control of GoG. The target segment of the survey was a Georgian citizen of 18 years and above.

Sample Design

The survey was designed as a two-stage cluster sampling with preliminary stratification by region, ethnicity and type of settlement. A two-stage cluster analysis was applied in sample design. The primary sampling unit (PSU) was a census unit and the secondary sampling unit (SSU) was a household. In total, ten respondents were interviewed in each PSU and one respondent was interviewed in each SSU. The selection of households was conducted according to a random walk principle. The respondent within each household was chosen according to the last birthday principle.

Both the sampling and the fieldwork were conducted in two stages. During the first stage, the sample size was set at 2950 respondents, stratified by regions and settlement type. There were 295 clusters sampled, with ten interviews in each cluster on average. The original sampling frame included the quotas for ethnicity as indicated below:

After the first stage, the number of respondents was counted according to ethnicity: Georgian, Armenian, Azeri and Other. During the second stage, an additional sampling was conducted taking into account the shortage of respondents in each ethnic group. During the second stage, several of the interviews were conducted in regions with dense ethnic minority populations with random cluster sampling. The respondents from the clusters were chosen by ethnicity. In the regions with low density of ethnic minority populations, the interviews were conducted using the “snowball” method.

In addition to these two stages, the dataset includes interviews with the focus group participants. The questionnaire used for this group is identical to the field questionnaire. This group is stratified and weighted using the same methodology as the general sample, although the focus groups’ analysis did not include a cluster variable. The sampling weights were calculated according to region, sex and ethnicity. In total, 3345 questionnaires were completed: 3284 from the general public and 61 focus group participants.

Both the sample size and sampling error (SE) in each region are presented in the table below:

The confidence level was defined as 95 percent.

The sample size provided the opportunity to analyze the data according to settlement type:

In the regions that are densely populated with ethnic minorities, the quotas for respondent selections were distributed proportionally in order to guarantee the sufficient representation of ethnic minorities in sample.

The sample size provided the opportunity to analyze the data according to ethnicity:

Survey Documents and Preparation for the Fieldwork

At the preparatory stage of the survey, the following documents were prepared: training manual, instructions for field personnel, instructions for completing the questionnaire, questionnaires (in Georgian, Russian, Armenian and Azeri languages) and show cards in four languages. UNICEF Georgia approved these documents in advance.

Selection and Training of Field Staff

Each interviewer participated in two types of training. Research specialists, the fieldwork department and trainers developed the ACT general training module. It covers the following subjects:

• General communication skills;
• Techniques of interviewing;
• Techniques for working with different vulnerable segments and sensitive issues;
• Main methods and techniques of the research.

After the general training was completed, ACT gave the interviewers a training manual, which was developed for the purposes of fieldwork, to better acquaint them with fieldwork techniques.

The project and field managers used the training manual to conduct a training for the interviewers. The training process mainly focused on the following issues:

• Subject of the survey;
• Study tools;
• Sampling design;
• Questionnaires and show cards;
• Detailed instructions to ensure the respondents understand the importance of participation in the study and ensuring the confidentiality.

It should be also noted that the team of interviewers was comprised of Georgian, Russian, Azeri and Armenian speaking interviewers. Questionnaire coding, revision specialists and control group members also attended the trainings.

The training of the interviews was conducted in two days on the 6th and 9th of July, 2012. The first day of the training was devoted to the regional supervisors and the second day to Tbilisi interviewers. After the training process was over, all interviewers were equipped with all documents necessary for fieldwork including the following:

• Training manual;
• Questionnaire;
• Show cards;
• List of sampling points.

The regional fieldwork was conducted from 7 -23 of July while activities in Tbilisi were conducted from 10-25 of July.

Primary Control of Questionnaires

Before sending the questionnaires to the head office in Tbilisi, primary control specialists in each region were responsible for checking the quality of each completed questionnaire. In the event of any inconsistencies, the control specialists gave the questionnaires back to the interviewer for correction. The control specialists followed previously defined procedures when checking the questionnaires including the following:

1. Identify omitted or incomplete sections;
2. Check accuracy of completed questionnaires and identify errors.

During the fieldwork stage, the primary control specialists also ensured that the interviewers were obeying questionnaire completion rules.

Quality Control

All quality control team members conducted field quality control processes in accordance with ACT quality assurance procedures. The quality control process ran simultaneously with the fieldwork activities to ensure the data obtained was of high quality. The quality control process was conducted from 15-30 of July, 2012.

The fieldwork control procedures and techniques were conducted in the following way:

• Attendance at the interview: 3-4 per cent of the sample size;
• Telephone control: 15 per cent of conducted interviews were checked by using a special mini questionnaire through telephone to re-interview the respondent;
• Visit the respondent: 10 per cent of submitted questionnaires were verified by visiting the respondent.

In addition, all the questionnaires were revised and edited by the field coordinators and revision specialists before being submitted to the ACT head office for coding and data entry.

Once the fieldwork activities were completed, the results of the quality control process were summarized and given to the head of the field department.

In conclusion, no significant inaccuracies were discovered during the quality control process. In the event information was missing from a questionnaire and could not be retrieved by the interviewer, the head office called the respondent to accurately complete the form

Logical Control

The field manager organized the questionnaire’s logical control and editing processes. The revision specialist controlled all questionnaires in order to identify any missing information or inaccuracies in the questionnaire.

Coding

The coding specialists and coding group coordinators conducted coding of open-ended questions after the questionnaires had undergone all quality assurance procedures including primary control, quality control and logical control.

Data Entry

For data entry and archiving of the corresponding documentation, the following procedures were performed:

• Receiving the revised and coded (open-ended questions) questionnaires from the field department;
• Distributing the questionnaires to the data-entry operators;
• Collecting the entered questionnaires;
• Merging the databases.

In addition, four data entry operators entered the data in SPSS 16.0.

Data Cleaning

The ACT database manager and database specialist applied the following macros in order to exclude inconsistencies from the data:

• Data (sort cases, transpose, restructure, aggregate, split file, select cases);
• Transformation (compute variable, count values within cases, recode into same variables, recode into different variables)
• Analysis (descriptive statistics, frequencies, descriptive, explore, crosstabs)
• Reports (case summaries)
• Tables (general tables, table of frequencies)
• Compare Means (means)

Macros designed by ACT:

1. Filter – these macros are the key cleaning macros. For proofing, any type of image can be indicated. Where valid, macros provide visual selection of the respective questionnaires.
2. Equals – these macros provides identification and visual selection of similar values in the set of variables.
3. Delsys - these macros provides identification of omitted values and their placement in the right side.
4. Include - these macros links two sets of variables and identifies cases where values of one set are repeated in the values of other set.
5. New - these macros provides right side variables to be replaced into positive variables.

In the case an inconsistency was identified, the respective data was verified with the questionnaire.

Data weighting

In order to generalize the data to a general population and provide statistically significant conclusions, statistical weighting of data was conducted. Region, gender and ethnicity were taken into account in data weighting to ensure final weights provided correct data interpretation.

Data analysis

The statistical software “StataSE 12” was used to analyze the data. The “SVY” procedure used in the analysis adjusts for the impact of the weights on the standard errors, as well as point statistics. “SVY” computes the standard errors of the estimates taking the survey design into account. It is then used for computing variances based on the first-order Taylor series approximation also known as linearization. It also contains the stratum identifier variable to take into account the multistage structure of the survey.

The “svy: tabulate” procedure was used for one and two-way tabulations for complex survey data, with confidence level set to 95 per cent.

To analyze what affects the likelihoods of involving the police, a school, school staff or social workers/social services in response to child abuse cases, “stepwise, pr (.05)” procedure was used. This procedure allows for choosing the variables that have coefficients significantly different from zero in the model out of a set of independent variables.

After analyzing the quantitative research data, the discussion meetings with youth ranging in ages from 14 to 18 were conducted to evaluate Georgian youth’s knowledge and attitudes regarding child abuse and neglect. During the course of the youth group activities, the findings from the study were also shared. Three meetings were held in April 2013 in Tbilisi, Kutaisi and Telavi, which are three major cities in Georgia. In sum, 76 children were involved with an average age 15.6, 60 per cent were girls and 40 per cent were boys. The young people came from public as well as private schools, disadvantaged communities, ethnic minorities. In addition, youth with disabilities as well as those from internally displaced families were involved. The regional centers of the Georgian Ministry of Education and Science the Anti-Violence Network of Georgia, a local NGO, to host these group discussions.

The aim of these meetings was to gain insights into the way Georgian youth view child abuse issues. The youth in attendance were requested to express their positions and discuss the following issues:

• Knowledge of different forms of domestic violence and violence against children;
• Profiles of victims and perpetrators;
• Where does violence take place;
• Physical punishment as a method of child rearing; harsh versus positive parenting;
• Reporting cases of child abuse;

The youth members in each focus group were randomly grouped into 4 sub-groups. The subgroups were given different questions to discuss in and then later present their findings to the entire group. These four presentations were followed by discussions guided by professional moderator.

For a detailed guide, see Appendix 3.

Although it is not possible to directly compare the quantitative data from the adult population with the data findings from the youth focus group discussions, the findings provide an opportunity to observe general age-related trends regarding child abuse in Georgia. Findings of the discussion groups are presented in the chapter “Knowledge, Attitude and Practices of Youth (aged from 14 to 18)”.