Estimating and managing the changing methodological parameters of self-report surveys of addictive behavior - based on the waves of the National Survey on Addiction Problems in Hungary (NSAPH) in 2007 and 2015.

The standard nature of the procedures and tools of sampling and data collection cannot guarantee the stability of data reliability and validity because non-sampling errors are highly sensitive to social conditions. The present study provides a post-hoc attempt to estimate and manage the changing methodological parameters of self-report surveys of addictive behaviours (being highly subjected to changes in social conditions) to make data interpretation easier. The analysis is based on the data of two national Hungarian representative surveys assessing addiction problems in 2007 and 2015 (National Survey on Addiction Problems in Hungary [NSAPH]). Both surveys were conducted using a Hungarian nationwide representative sample aged 18-64 years applying similar procedures in data collection and -processing. Regarding data concerning substance use, both surveys included variables to estimate non-sampling errors in line with current international practices. The methodological parameters of NSAPH2015 showed an increase in non-sampling errors regarding substance use behaviour compared to NSAPH2007. The present paper elaborates an estimation procedure based on the assumption that when following a population, the proportion of people who have ever engaged in a specific type of addictive behaviour cannot be reduced in the given population over time. This also applies to cohorts followed by cross-sectional surveys among national representative samples, as far as lifetime prevalence and data on the age of first use/activity is available. To identify valid trends in different behaviours in epidemiological research assessing addictive behaviours or other sensitive data, researchers should provide the required conditions for controlling or correcting data by cohort analysis.