Title | Comparing RE and PA models | |
Author | William Sribney, StataCorp | |
Date | January 1999; updated June 2013 |
Random-effects estimators (or other cluster-specific estimators) fit the model
Pr(Y_{ij}=1 | X_{ij}, u_{i}) = F(X_{ij} b + u_{i})
whereas population-average estimators fit the model:
Pr(Y_{ij}=1 | X_{ij}) = G(X_{ij} b*)
The subtle point is that b and b* are different population parameters. Hence, the estimators are estimating different things. In practice, however, b and b* are often very close.
The population-averaged model does NOT fully specify the distribution of the population. The cluster-specific model DOES fully specify the distribution (u_{i} is either given a distribution—i.e., a random-effects model—or is considered fixed like X_{ij}—i.e., a fixed-effects model). The population-averaged model specifies only a marginal distribution. Hence, the term “marginal” is often used for GEE estimates.
The subtle difference between b and b* is best explained with an example.
Suppose that you are looking at
Outcome Y_{ij}: employment/unemployment Predictor X_{ij}: married/unmarried
Then, under the cluster-specific model
logit Pr(Y_{ij}=1 | X_{ij}, u_{i}) = a + X_{ij} b + u_{i}
the odds ratio
Pr(Y_{ij}=1 | X_{ij}=1, u_{i})/Pr(Y_{ij}=0 | X_{ij}=1, u_{i}) OR_{cs} = -------------------------------------------- = exp(b) Pr(Y_{ij}=1 | X_{ij}=0, u_{i})/Pr(Y_{ij}=0 | X_{ij}=0, u_{i})
represents the odds of the person being employed if married compared with the odds of the SAME person being employed if not married.
Under the population-averaged model
logit Pr(Y_{ij}=1 | X_{ij}) = a + X_{ij} b*
the odds ratio
Pr(Y_{ij}=1 | X_{ij}=1)/Pr(Y_{ij}=0 | X_{ij}=1) OR_{pa} = ------------------------------------ = exp(b*) Pr(Y_{ij}=1 | X_{ij}=0)/Pr(Y_{ij}=0 | X_{ij}=0)
represents the odds of an AVERAGE married person being employed compared with the odds of an AVERAGE unmarried person being employed.
Rather than saying “AVERAGE”, sometimes I speak loosely and say the odds of a married person “picked at random” being employed compared with the odds of another unmarried person “picked at random” being employed.
Let me now show that b and b* are, in general, different population parameters.
Here is my definition of the population DISTRIBUTION. (It is NOT a dataset.) The total population consists of five subjects:
subject i j X_{ij} u_{i} Z_{ij} Pr_{cs} Pr_{pa} --------- --- ---- ---- ----- ------ ------ 1 1 0 -0.2 -0.10 0.4750 0.5249 1 2 1 -0.2 0.50 0.6225 0.6674 2 1 0 -0.1 -0.00 0.5000 0.5249 2 2 1 -0.1 0.60 0.6457 0.6674 3 1 0 0.0 0.10 0.5250 0.5249 3 2 1 0.0 0.70 0.6682 0.6674 4 1 0 0.1 0.20 0.5498 0.5249 4 2 1 0.1 0.80 0.6900 0.6674 5 1 0 0.2 0.30 0.5744 0.5249 5 2 1 0.2 0.90 0.7109 0.6674
Here Z_{ij} = a + b*X_{ij} + u_{i}, with a = 0.1, b = 0.6, and u_{i} as given.
The cluster-specific probability Pr_{cs} is given by
Pr_{cs} = exp(Z_{ij})/(1 + exp(Z_{ij}))
For this population, the population-averaged probability, Pr_{pa}, is simply the average of Pr_{cs} for each X_{ij}. That is,
Pr_{pa}(X_{ij}=1) | = | (1/5) * |
| Pr_{cs} | (x_{ij}=1) | |||
= | (1/5) * | (0.6225 + 0.6457 + 0.6682 + 0.6900 + 0.7109) | ||||||
= | 0.6674 |
Cluster-specific odds ratio = exp(b) = exp(0.6) = 1.8221.
This is, of course, the same as the odds ratios computed within subject:
Subject 1: (0.6225/(1 - 0.6225))/(0.4750/(1 - 0.4750)) = 1.8221 Subject 2: (0.6457/(1 - 0.6457))/(0.5000/(1 - 0.5000)) = 1.8221 Subject 3: (0.6682/(1 - 0.6682))/(0.5250/(1 - 0.5250)) = 1.8221 Subject 4: (0.6900/(1 - 0.6900))/(0.5498/(1 - 0.5498)) = 1.8221 Subject 5: (0.7109/(1 - 0.7109))/(0.5744/(1 - 0.5744)) = 1.8221
Population-averaged odds ratio is
exp(b*) = (0.6674/(1 - 0.6674))/(0.5249/(1 - 0.5249)) = 1.8169
Solving for b* gives
b* = 0.5972
so b* is closer to the null, as the theory predicts (see the Neuhaus papers).
b and b* above are the TRUE population parameters, not estimates.
If we had a dataset consisting of a sample from this population distribution, and we used xtgee on this dataset (with the logit link and binomial distribution), xtgee would be estimating b*. If we used regular logit, we would also be estimating b* (one would want to specify the vce(cluster clustvar) option to correct the standard errors in this case).
If we used clogit on this dataset or a random-effects logit estimator, (one that assumes normally distributed u_{i}), we would be estimating b.
(Aside: The random-effects logit estimator described in the Neuhaus papers assumes a distribution for u_{i} different from that of the random-effects logit estimator implemented in Stata. My theory discussion here assumes one is using the “correct” distribution of u_{i}. I do not want to digress on this subject, but random-effects estimators that assume different distributions for u_{i} are technically different estimators; hence, there is more than one “random-effects logit estimator”.)
Here b and b* are almost the same number (b = 0.6 and b* = 0.5972), so it is easy to obscure the fact that the cluster-specific and population-averaged estimators are estimating different parameters. In other cases, the difference can be greater, so it is important to keep in mind which one you are estimating.
The bottom line for someone thinking about using the GEE estimator is to think about whether the averaging procedure makes sense for the type of inference you want to make. If you want to estimate how marriage makes a person get his act together and get a job (or else leave it to the spouse to bring home the groceries), then you want to go after b. If you want to look at employment for the average married person compared with the average unmarried person, then you want to go after b*.
Sometimes you might argue b* and b should be close, so the distinction is not worth making. But you had better be sure of your argument. Zero correlation (u_{i}=0) makes them the same; big Var(u_{i}) makes the difference greater.