Methods in Human Growth Research: Ordinal longitudinal data analysis

摘要: Introduction Growth data and longitudinal in general are often of an ordinal nature. For example, developmental stages may be classified into categories behavioural variables repeatedly measured by discrete scales. Consider the set presented Table 15.1. This table contains information on marijuana use taken from five annual waves (1976–80) National Youth Survey (Elliot et al ., 1989; Lang 1999). The 237 respondents were 13 years old 1976. variable interest is a trichotomous ‘Marijuana past year’ during consecutive years. There also gender respondents. Ordinal like this analysed as if it continuous interval level data, that is, means methods imply linear relationships normally distributed errors. However, 15.1 essentially categorical at ordinal, not level. Consequently, much better way to deal with such response treat coming multinomial distribution; nature then account imposing particular constraints odds responding, i.e. choosing one category rather than another. As will further explained below, analysis can based cumulative, adjacent-categories, or continuation-ratio (Agresti, 2002). form equality inequality these types odds.