FS Home
Section 1:  Introduction
Section 2:  Coping Strategies
Section 3:  Computing
Section 4:  Analysis Ex. (HLS Bangladesh)
Section 5:  Analysis Ex. (HLS Kenya)

One-Way Analysis    Targeting    Two-Way Analysis    Regression

What is the relationship at the household level of irrigation and cultivation on
food sufficiency in Somalia?

One easy way to see and interpret interactions is through standard graphs. Here, the outcome is (as usual) on the vertical y-axis, and two categories of a factor are on the x-axis; two lines are drawn for the two possibilities of the other factor.

• In order to graph the possible interaction we need to create dichotomous variables instead of continuous variables for cult and irr99 (as we have previously been working with) and create a new interaction term for the new dichotomous variables.

• Once we have created the dummy variables and the interaction, a linear regression model can be run.
  

• Place ( 0 or1) as the value for each variable:
  Food Sufficiency = 1.794 – 0.122(dcult)+ 1.856(dirr99)+ 1.337(dcult*dirr99)
  This equation includes an interaction term for cultivation and irrigation(intirrcu). The term for irrigation is associated with a significant increase in the food sufficiency (p=0.000), but the term for cultivation becomes insignificant (p=0.815) when the interaction term is added in the model, due to collinearity between the variables of cultivation and irrigation. The interaction term will therefore need to be retained in the model to control for this effect which would otherwise lead to inappropriate assumption that cultivation had a lesser effect on food sufficiency. 
  
  Food Sufficiency = 1.794 - 0.122(0) + 1.856 (0) + 1.337(0) (0)   this totals 1.794 and placed in box A
  Food Sufficiency = 1.794 - 0.122(1) + 1.856 (0) + 1.337(1) (0)   this totals 1.672 and placed in box B
  Food Sufficiency = 1.794 - 0.122(0) + 1.856 (1) + 1.337(0) (1)   this totals 3.650 and placed in box C
  Food Sufficiency = 1.794 - 0.122(1) + 1.856 (1) + 1.337(1) (1)   this totals 4.865 and placed in box D        

	dcult
dirr99	0	1
0	A	B
1	C	D

This interaction could be catagorized as LESS THAN ADDITIVE: When the results are less than additive, neither factor shows an effect when improved alone because both are needed to see an improvement.