I wouldn't say that weight or shape has no effect, but rather that the impact is small enough, and other factors are more significant, that it is harder to separate out weight and body style. The 2 door Golf is lighter, but if you have a full tank, a full trunk, and a passenger, you weigh the same as the 4 door (or more). Some factors really hit you in city driving, but once you're on the highway, an ALH in good condition in a car that has been properly maintained, is impacted less notably by weight or CD.
Many years ago, when ALH engines dominated the thread discussions, some people would refer to a car that easily rated high FE as "one of the good ones." I'm a believer in the idea that some ALH engines just do better than others, even when other factors are equal. Our Jetta is slightly more efficient than the Golf, but the Golf has a better avg MPG yearly, because of the driver. This is my evaluation when the majority of other key variables are accounted for- but obviously some variables will remain. True comparisons are really hard- same route, same distance, same time, same fuel, same weather, same wind, same speed, etc. This evaluation only happens when both cars have to head to the same long distance destination, and do so by following one another (but not closely enough to impact FE). Then the cars themselves: Same tires? Same PSI? Engine oil? Transmission oil? Etc. I've been able to match these for our cars for a few, but not all, long distance trips. What about maint and condition? Type and condition of suspension components? Exhaust? Mileage on the body? These are tough to match exactly. What else is drawing power? If both cars follow one another across country, does one have AC on, stereo at full blast, all lights at full power, and three devices plugged into the 12vDC? I can look at my FE records and know which trips required AC just by the MPG numbers. What about weight? The Jetta weighs more than the Golf, but if you're traveling, has the weight been equalized? Do you know what the weight difference between the car really is? Once you've "matched" all of these variables, you're left with a host of difficult variables: Has the car been modded? Age, condition, and type of nozzles? Tuning modified? Owners will claim FE changes due to mods, but it's often very difficult to be exact about the impact of modifications unless you establish a base line, and then track the changes over a significant number of miles and conditions (which many owners here will do).
Lastly you have the driver: Driving style is key for FE, and very hard to compare across platforms, which means you need the same driver with a fleet of different cars. Luckily, Fred's has a bunch of those. Even with the same driver, conditions change from day to day, so unless the same driver, runs the same route, with different cars on the same day, we introduce variables that make comparisons difficult. Of course, these variables are smoothed by logging hundreds of thousands of miles over the last decade, so we do in fact get results that are reasonably accurate, even if the precise variable is difficult to pinpoint, along with the extent of its impact on FE.
So, with our cars, when I've accounted for most of the variables, what do I find? On the same long distance trip, cars following each other, the Jetta does better by a just a bit. Is it CD? Well, I've also found that, even on the highway, same long distance follow-each-other trip, the car that I drive does better for FE. The Golf has a better overall lifetime MPG number b/c I drive it more, but under the same circumstances, I can get better hwy FE with the Jetta. It's tempting to just say that a tiny CD difference between the two is being seen on the hwy, but the numbers reveal my right foot ends up being the final factor.