Horticulture survey is a multi-purpose survey which is conducted ad hoc by Statistical Center of Iran (SCI). Availability of survey variables in the sampling frame suggests a multivariate stratification in each province based on its desired variables for acquiring a higher efficiency. There are several ways to stratify the sampling frame considering all stratification variables, such as using sum of observation on all variables, clustering, using first principal component, and specially an almost new method which uses multiple-frame techniques. We introduce, the equivalent linear composition factor, and illustrate how it works more efficiently then the other methods in this particular survey.