The role of different intermolecular interactions in the aggregation of amphiphiles in an organic solvent is studied for systems of relevance to liquid-liquid extraction (LLE), a chemical process used to selectively recover metals from complex mixtures. Of specific interest is the role, or lack thereof, of hydrogen bonding, which is often assumed to be a main driver of the organic phase structural organization that has been linked to separation efficacy. Toward that end, a series of malonamide extractants in n-dodecane have been studied in the absence of any extracted aqueous solutes, including water. The series of extractants includes N,N'-dimethyl-N,N'-dibutyltetradecylmalonamide (DMDBTDMA), two of its homologs, and N,N'-dimethyl-N,N'-dioctylhexylethoxymalonamide (DMDOHEMA). This simplified model LLE system enables systematic investigation of the role of dipole-dipole and alkyl tail steric interactions in amphiphile aggregation. Small-angle X-ray scattering (SAXS) profiles computed from molecular dynamics trajectories are in good agreement with experimental SAXS data. Molecular dynamics simulations show that malonamide aggregation results from dipole-driven self-association and lacks characteristic aggregate sizes. Mid-q correlation peaks in the SAXS profiles emerge at high concentration for each malonamide. In those densely packed solutions, the correlation peaks are observed to result from alkyl tail-induced spacing between electron-rich polar head groups, with peak positions determined by the different alkyl tail lengths present in the malonamide molecule. This explanation of the SAXS correlation peaks contrasts with the prevailing literature, which attributes mesoscale features observed in small-angle scattering to the formation of microemulsions. Instead, this work finds that these features are present in the absence of water or any reverse micellar organization of the malonamides. As such, molecular-scale malonamide self-association and packing, rather than microemulsion-based colloidal-scale descriptions, is a more appropriate framework for these LLE systems.