Best explanation I heard about this... Aikido locks are intended to be body locks, not appendage locks. So at some point, the joint lock of shiho nage should not be limited to the arm. Once this happens, we should not be limited to the mechanical requirements of creating torsion through a largely horizontal position of the arm. Once we don't have a need to squat under a horizontal arm that is more or less only as high at the shoulder of your partner... you have more freedom for your partner to move out of your way, rather than move around your partner. Without that desperate feeling that you are going to rip off your partner's arm. Not that you can't...
Having struggled with shihonage for a long time, a few years ago I discovered a couple of essential things about this technique. First of all, if you have to resort to ducking under your partner's arm you have basically lost by that point. The whole technique becomes a lot easier and safer if you manipulate your partner's body before then, so that their elbow rises over your head (I like to think of the image of uke's arm making a kind of pitched roof over your head). More recently, I realised that most of shihonage is generated in the initial contact - if you get this right, this weakens your partner's frame and posture so that everything from that point follows almost effortlessly. Kanetsuka Sensei always says "attack partner's knees": when he does this to me at the start of the technique, I feel my feet become uprooted, and there is no chance to counter the technique at all.
This also, of course, helps to even out any adverse height difference between you and your partner (though I do still occasionally struggle in extreme cases!).