"the third-child deterrent appears stronger among wealthier families"
So this has both the most and least effect on families that have little reason to care about needing a bigger car (because they can easily afford it, or they can't afford any car at all, respectively). When you go fishing for a correlation, form a post-hoc hypothesis for its mechanism, and then other data fits that hypothesis poorly, it's a strong sign you've found yourself a red herring. Which is the usual result of failing to properly understand the distinction between correlation and causation.
"the third-child deterrent appears stronger among wealthier families"
So this has both the most and least effect on families that have little reason to care about needing a bigger car (because they can easily afford it, or they can't afford any car at all, respectively). When you go fishing for a correlation, form a post-hoc hypothesis for its mechanism, and then other data fits that hypothesis poorly, it's a strong sign you've found yourself a red herring. Which is the usual result of failing to properly understand the distinction between correlation and causation.