Attitudes towards Norwegian development aid

Name: Attitudes towards and knowledge about Norwegian development aid
Topic: Public sector

Responsible division

Division for social surveys

Whole country, but broken down at province level.

Statistics Norway has conducted the survey in 1972, 1974, 1977, 1980, 1983, 1986, 1990, 1993, 1995, 1996, 1999, 2001, 2006, 2010 and 2013.

Microdata are transferred to the NSD - Norwegian Centre for Research Data.

The purpose of the study is to examine people's attitudes to Norwegian development aid. The survey consists of a series of questions that have been asked several times since 1972. There has been some changes to the questions throughout the survey’s history, also in 2013. The Norwegian Agency for Development Cooperation (Norad) has funded the studies.

The main users of statistics are public authorities and others working agencies on Norwegian aid.

The statistics is compiled on the basis of a more comprehensive report. Both are funded by NORAD. The client has access to the content of the report, but only for internal use, before the statistics and the report are published simultaneously by SSB and NORAD.

Voluntary participation

The Norwegian population aged 16-79 years, registered as resident inNorway. The unit is person.

In 2013, the survey on attitudes towards Norwegian development aid was conducted as a separate survey, as it was2006. In2010, the data were collected as part of the Travel andHolidaysurvey. From 2001 to 1993 it was part of Statistics Norway’s Omnibus survey, while it was in the labor force survey from 1972 to 1992. The Norwegian Population Registry was used to define the population and to draw the sample. In addition, information from the education registry was used in the analysis.

After 1993 the survey has been based on a representative sample of 2,000 persons aged 16-79 years. From period 1972 to 1990, the surveys’ sample was aged 16-74 years. Until 2006, the sample was drawn using Statistics Norway’s two-stage selection plan. In 2013, the sample was drawn as a self-weighting probability sample in one step.

The bulk of the data collection was conducted by telephone. Respondents who were not interviewed in the telephone survey were offered to answer a web form. 132 of the 1166 interviews were conducted on the Internet.

The interviews were conducted as computer assisted interviews. The program has controls to prevent incorrect answers or registration errors during the interview. In some cases, the interviewer gets warnings when they enter an answer. In other cases there are limits that can not be exceeded.

Nothing is ever published that makes it possible to detect the identity of the respondents.

A number of questions has been kept unaltered since 1972, but there has been several changes to other questions.

We try to avoid measurement error ( respondent providing incorrect information ) and processing errors (errors related to coding of responses) by using different control systems. One can however not guarantee that all errors detected.

Unit Non-response (persons that do not participate in the survey) have recently varied between 32 percent (1996) and 46 percent (2010 ) . Unit Non-response in 2013 was at 41 percent. For more information about bias due to non-response, see the publications for each survey.

Since the results are based on a sample of the population the survey covers ,there will always be some uncertainty in the data. This is called the sample variance. Because the sample is drawn randomly, we can calculate how large the sample variance is expected to be.

A commonly used measure of the uncertainty of a variable is the standard deviation of the observed value of this variable. The size of this standard deviation depends on the number of observations, the way the sample is drawn and the distribution of the relevant variable in the population. The distribution in the population is not known, but it is possible to estimate the standard deviation of sample distribution using the observations in the sample.

There were no separate calculations of such estimates for this study. Table 1 shows, however, the size of the standard deviation of the observed percentages for different sample sizes. The table shows that uncertainty increases as the number of observations decreases and when the percentage approaches 50.

Tabell 1. Expected standard deviations for observed percentages at different sample sizes