Is it still possible to join ISRD3?
Yes, it is. Although we originally decided that ISRD3 would be between 2012 and 2015, we have decided to continue data collection through 2019. We continuously update the larger data file to include the more recent data. Since the ISRD is primarily a theory-testing survey, rather than simply an effort to estimate prevalence of delinquency and victimization, it is less crucial that all country data are collected at exactly the same time. The international data set does include a variable indicating the year of data collection.
Is it possible to have more than one ISRD research team in a country?
Yes, it is. This is particularly relevant in large countries like China, or India. However, even in smaller countries it may make sense to have multiple research teams (located in different regions of the country). We do expect though that there will be coordination and communication between the national teams.
Is there central funding available for the ISRD?
Unfortunately, each national research team has to try to secure its own funding. We have been successful in obtaining EU-based funds to support some of the work in groups of countries, but there is no overarching funding scheme that includes all participants. However, the questionnaires and other research tools are available to participants, so the actual implementation of the survey does not need to be very costly. This is particularly the case when making use of Unipark, the electronic data collection method that ISRD uses. The ISRD Steering Committee is also willing and available to participate in national funding proposals.
Can anybody use the ISRD questionnaire?
Yes, but one needs to clearly identify that it is the ISRD questionnaire ISRD3 by using the following reference:
Working Group (2013). Questionnaire ISRD3: Standard Student Questionnaire (ISRD3 Technical Report Series #2).
Who can use the combined ISRD3 data?
Currently, the data can be used by ISRD data collectors who have submitted a technical report that is approved by the Steering Committee.
After having signed the ISRD3 data access agreement you will receive an individual invitation to the depository together with instructions on how to get first access and how to use the files.
In a later stage the data will be placed in a data archive that is public for scientific research.
Where can I find the most updated version of the ISRD3 data?
Is SPSS format supported?
Yes. In the data depository (see FAQ “Who can use the combined data”) you will find a “ReadMe.pdf” explaining how to create your own SPSS data file using the files provided.
Is the SPSS missing data specifications syntax available?
Yes, when downloading SPSS format data from the depository you will receive a command file specific for defining the missing values of the dataset you want to use. A “ReadMe.pdf” located in the depository explains the necessary steps.
Is Stata format supported?
Yes, the depository contains data for Stata version 14+ and Stata version 11. It is recommended to use Stata version 14+ data because it uses very long strings and UTF-8 character encoding, able to display non Western language characters correctly.
What is meant by “standard” ISRD dataset?
This expression refers to data which includes only grades 7 to 9.
What is meant by “expanded” ISRD dataset?
This data version additionally includes respondents from grade levels other then 7 to 9 (e.g. grade 6 or grade 10+).
Should I use weights in analyses comparing countries?
Some countries have weights. If you want to compare estimates of means (or proportions) across countries, you should use these weights. You can use weights even though you include countries with no weighting, since for those countries the weight of each respondent is always 1.
Note that if you examine variables of online follow-up questions (available only in online data collection), it is better not to use the weight variable because some countries used several data collection modes.
What weights are in the data?
The data currently contains three weights:
a_weight (analytic weights) adjusts for sample/population ratio without changing the number of observations. This weight is 1 for countries which did not submit weights. This is the weight to use if you work with SPSS and don’t use the SPSS module Complex Samples.
p_weight (population weights) adjusts for sample/population ratio and projects the number of respondents to population levels. This is the weight to use if work with Stata or with the SPSS module Complex Samples.
m_weight (specific) is a weight that equalizes the number of respondents in each country so that countries with large samples do not dominate the combined data – the number of cases per country are weighted to be equal. Such a weight should only be used if you combine estimates of means (or percentages) of several countries. Currently this weight is only available to those who can run their own command files to create these weights.
Examples for SPSS and Stata will be provided in the future.
How can I adjust for the clustered structure of the data?
The ISRD data sampling did sample students clustered in school classes, so observations are not independent from one another, as they would be in a simple random sample of individuals. In analyses, it is advisable to use the SPSS Complex Samples module or corresponding Stata commands to adjust for the design effects due to the clustering of data within classes.
If I adjust for clustered data structure, what level of clustering should be adjusted for?
The ISRD3 is a sample of classes, so in most cases it is advisable to adjust for class-level clustering.
Should weighting be used when analyzing the online follow-up responses?
If you are using or comparing online follow-up responses of several countries is better not to use weights because some countries used several data collection modes. The variable “srvmode” shows the survey mode used in a country. If in the set of countries you are analyzing paper & pencil questionnaires have not been used, weights can be applied.
If I want to compare prevalence levels for regions or countries combined into groups using standard variables, how can I adjust for the differential sample sizes and sampling ratios in different countries?
You can create a weight that equalizes the number of respondents in each country so that countries with large samples do not dominate the combined data – the number of cases per country are weighted to be equal. Such a weight should only be used if you combine estimates of means (or percentages) of several countries. Currently this weight is only available to those who can run their own command files to create these weights.
Examples for SPSS and Stata will be provided in the future.
When will IRSD4 start ?
We hope to start the fourth sweep of the International Self-Report Delinquency study (ISRD4) in 2020.
Will ISRD4 be different from ISRD3?
No and yes. The ISRD project wants to maintain a core standardized set of questions, both related to delinquency and victimization, as well as basic socio-demographic variables and basic theoretical constructs. However, we also want to be flexible and incorporate novel and emerging issues (i.e. new crime types, different theoretical perspectives) For more explanation, see ISRD3 Technical Report Series #1
Is it possible to provide suggestions for ISRD4?
Yes, by all means. We are interested in suggestions related to survey administration, questionnaire content, and possible other ways to improve the project. Contact the ISRD Steering Committee with your suggestions. We also discuss ISRD4 at the ISRD participant meetings at the European Society of Criminology, American Society of Criminology and other venues.
How do I sign up for participation in ISRD4?
Express your interest to the ISRD Steering Committee. We will announce all relevant deadlines on this website and through other relevant venues.