Several dummy observation sessions were undertaken as part of the observer training process.
Thus, both SMI and MMI can be used to impute missing values on background variables in large-scale assessments to avoid the conceptual flaws and the possible biases associated with the common approach of dummy coding missing observations.
So linear regressions were estimated with two explanatory variables: the day number as a continuous predictor in one hand, and the period, as a dummy variable separating observations in two groups, before and after, a given date.
In addition, in the BHPS for each observation dummy variables were included that identified the individual's region of residence and the survey wave from which the data were drawn.
In several current large-scale assessment programs (Allen et al. [2001]; Foy et al. [2008]; OECD [2009]) missing observations were dummy coded and the dummy variables were subsequently used in the population model.
9> > -wrap-foot> aA dummy variable in which observations with missing values on income are recoded 1 and all other observations are recoded 0 bThis column reports the correlation coefficients between the dummy and the relevant health variable The original self-assessed health (SAH) variable is a six-point scale variable ranging from very good to bad.
And the identification of a country-pair dummy requires the bilateral observation to be observed at least once for each destination country.
If this is not the case, defining dummy codes for missing observations may be questionable from a theoretical point of view (Schafer & Graham [2002]) and may also lead to biased mean estimates (Rutkowski, [2011]).
A separate dummy variable for these observations was added to the set of dummy variables which the occupational class consists of.
Using an augmented regression analysis with seven dummies for the influential observations, we find that none of the corresponding coefficients is significant at the 5%-level.
Since we had a very limited number of (arbitrary) observation years and we only wanted to control for differences due to contextual factors, we included the year of observation as dummy variables in the analyses.
