The major factor here is foreshortening. Unless a photo is taken exactly perpendicular to the shadow cast by an upright object, the apparent length of the shadow will be shortened by the perspective. In your example, the photo seems to be taken somewhere around 45-60 degrees off perpendicular. Taking the cosine of that tells you the shadows appear at somewhere between 0.5 and 0.7 times their actual length, which gets you near enough to your expected 3.7:1 stretch factor.
Here's a top-view diagram of how that works.

A smaller factor is the unevenness of the lunar surface. It's notoriously difficult to estimate the elevation changes from the Apollo photos. Shadows falling on an uphill slope will be shorter than expected.
You also reference a near-perpendicular-shadow picture:

I am able to nearly reproduce the geometry of the image in a 3D tool using a 53.5º field of view and a 15.3º sun angle. It's not a perfect match, but I don't know the exact dimensions of the box or geometry of the terrain:

(Clearly, The Conspiracy has reached Unity Technologies.)
Additionally, the debunking site Clavius.org has some nice sets of photos showing how the same shadow can look very different from different directions. You may enjoy some of the other photo analysis pages on that site.