Adjustment Formula

How can we translate the abstract concept of Graph Surgery into something that can compute actual numerical values? In fact, we can work directly with graphs, in the form of Bayesian networks, and use BayesiaLab to perform Graph Surgery and simulate interventions.

However, before we illustrate that in the next section of this chapter, we want to formally conclude the line of reasoning that connects the pre-intervention distribution $P$ to the post-intervention distribution $P_m$ and introduce the Adjustment Formula. We paraphrase Pearl, Glymour, and Jewell (2016), p. 56f. to develop this formula.

In our example, we can easily estimate the pre-intervention distribution $P$ from the available data, but we need the post-intervention distribution $P_m$ to calculate the causal effect. The key lies in recognizing that $P_m$ shares two essential properties with $P$.

First, the marginal distribution $P(Z=z)$ remains invariant under the intervention because the process of determining $Z$ is unaffected by removing the arc $Z \rightarrow X$. In our example, this means that the share of men and women must remain the same before and after the intervention:

${P_m}(Z = z) = P(Z = z)$

Secondly, the conditional probability distribution $P(Y=y \mid Z=z, X=x)$ remains invariant under the intervention because the process that determines how $Y$ responds to $X$ and $Z$ stays the same, regardless of whether $X$ changes naturally or through external intervention. We can state this formally as follows:

${P_m}(Y = y \mid Z = z, X = x) = P(Y = y \mid Z = z, X = x)$

Furthermore, $X$ and $Z$ are marginally independent in the mutilated graph. This means that the conditional probability distribution of $Z$ given $X$ in the mutilated graph is the same as the marginal probability distribution of $Z$ in the pre-intervention graph:

${P_m}\left( {Z = z \mid X = x} \right) = {P_m}\left( {Z = z} \right) = P\left( {Z = z} \right)$

Since the Adjustment Criterion is satisfied in the mutilated graph, we have the following:

$P(Y = y \mid do(X = x)) = {P_m}(Y = y \mid X = x)$

By conditioning on $Z$ and summing over all values $z$, we obtain:

$P(Y = y \mid do(X = x)) = \sum\limits_z {{P_m}(Y = y \mid X = x, Z = z){P_m}(Z = z \mid X = x)}$

Furthermore, $X$ and $Z$ are independent in the mutilated graph:

$P(Y = y \mid do(X = x)) = \sum\limits_z {{P_m}(Y = y \mid X = x, Z = z){P_m}(Z = z)}$

Using the two invariance equations above, we obtain what is known as the Adjustment Formula. It expresses the post-intervention distribution exclusively in terms of the pre-intervention distribution:

$P(Y = y \mid do(X = x)) = \sum\limits_z {P(Y = y \mid X = x, Z = z)P(Z = z)}$

The Adjustment Formula computes the association between $X$ and $Y$ for each value $z$ (or strata of $z \in Z$) and then produces a weighted average. On this basis, we can now estimate the Average Causal Effect (ACE):

$ACE = P\left( {Y = 1 \mid do(X = 1)} \right) - P\left( {Y = 1 \mid do(X = 0)} \right)$

Simpson’s Paradox Resolved

Returning to Simpson’s Paradox, the equation

$P(Y = y \mid do(X = x)) = \sum\limits_z {P(Y = y \mid X = x, Z = z)P(Z = z)}$

gives us the answer to our question of whether we need to look at the aggregate data table or the gender-specific data table for determining the true causal effect of treatment on the outcome: “Conditioning on $Z$ and summing over all values $z$” means that we need to utilize the gender-specific table. More specifically, we need to compute the association between $X$ and $Y$ for each value of $Z$, i.e., each stratum of $z \in Z$, and then calculate the weighted average. This estimation method is also known as stratification.

Aggregate Table

Gender-Specific Table

The $ACE$ turns out to be negative: it has the opposite sign of what we would have inferred by looking only at the aggregate association between treatment and outcome. Bias can reverse the sign of the effect! Under our causal assumptions, a therapy promoted as beneficial actually reduces the probability of recovery by 10 percentage points. Acting on the misleading aggregate result could therefore expose an entire patient population to a harmful treatment. Encountering Simpson’s Paradox unprepared can turn an apparent medical breakthrough into a public-health disaster.

The good news is that we have a theory stipulating that gender is a confounder, and this variable is observed. If it were not recorded in our dataset (hidden variable), we would not be able to compute the causal effect of treatment. We can also imagine situations where we do not know that confounders exist and, therefore, do not measure them. This can lead to substantially wrong estimations of causal effects and lead to policies with catastrophic consequences.