Household Environment Survey 2015

The concept of data quality covers many aspects, starting from the initial planning of the survey to the dissemination of the results and how well users understand and use the data. There are seven dimensions of statistical quality: relevance, accuracy, timeliness and punctuality, accessibility and clarity, comparability, coherence and completeness.

Accuracy

This includes many aspects of the survey, mainly statistical errors due to the use of a sample, and also non-statistical errors from workers and survey tools. It also includes the response rates in this survey and their effect on the assumptions. This section includes:

Sampling Errors:
Data of this survey may be affected by sampling errors due to use of a sample and not a complete enumeration. Therefore, certain differences are expected in comparison with the real values obtained through censuses. Variances were calculated for the most important indicators and the variance table is in the final report found in the related external resources under downloads. There is no problem with the dissemination of results on national and regional level (North, Middle, South West Bank, Gaza Strip) or by locality type.

Non Sampling Errors:
The non-sampling errors are possible to occur at all phases of implementing the project, through data collection and entry which could be summarized as non-response errors, and responding errors (respondents), and interview errors (fieldworkers) and data-entry errors. To avoid errors and reduce the impact, it had been made great efforts through extensive training of fieldworkers on how to conduct interviews, things that ought to be followed during an interview, things that should be avoided, making some practical and theoretical exercises during training session, in addition to providing them with a manual booklet for fieldworkers which contained a private key questions of questionnaire, mechanism to fill questionnaire and methods of dealing with respondents to reduce refusal rates and providing correct and non-biased data. Also data entry staff were trained on the data entry program, which was tested before starting the data entry process.

As for office work, they had been trained for a special auditing of questionnaires and error detection, which greatly reduced rates of errors during field work. In order to reduce the percentage of errors during data entry, the program was designed to enter data so as not to allow any mistakes during the process and contained many of logical terms. This process led to disclosure of most of errors that had not been found in earlier phases of the work, where they were correcting all the errors that had been discovered.

After the completion of the aforesaid audits, data consistency was examined by computer using frequency and cross tables as turned out to be quite consistent. Errors impact was not detectable on data quality. This in turn gave a good impression of those in charge of the survey that we could rely on this data and extract reliable statistical and high significant indicators on the reality of corruption in Palestine.