Soc 128 – Fall 2013 TF: Asad Asad Constructing Causal Arguments 1. Hypothesize a relationship between a dependent and independent variable 2. Run a two-by-two table with dependent variable down (rows) and independent variable across (columns) 3. Do you observe a relationship? (read across the table – left to right. Also, remember to sum down the rows) 4. Hypothesize a third variable you could control for (or hold constant) 5. Run the table again with the control. 6. Does the association between the dependent and the independent variable go away? If the association does not go away, then you have replicated your findings and the relationship between your dependent and independent variable is said to hold. If the association does go away, one of two things could be happening – the tables won’t tell you which – use logical thinking! i. You may have found that the control variable explains the relationship between your dependent and independent variable. This indicates that it is an antecedent variable. In other words, your control variable (Z) affects both your dependent variable (Y) and your independent variable (X) but there is no direct relationship between X and Y. X Z Y Basically, variation in Z is producing variation in X and in Y, making it look like X covaries with Y. Once we control for Z, this relationship disappears since it is a spurious one. ii. You may have interpreted the relationship between X and Y. Here, the control variable functions as an intervening variable so that you say that your independent variable affects your dependent variable through your control variable. In other words, you show that X affects Y indirectly through Z. Z tells you how X affects Y. when you hold Z constant, there is no direct relationship between Z and Y and hence no association. X Z Y 7. If you find that the relationship between X and Y holds for only one of the values of the control variable, then you have specified the relationship and shown that your independent and control variables interact to affect your dependent variable. In other words, whether X affects Y depends on the value of Z.
© Copyright 2026 Paperzz