Multivalent intrinsically disordered proteins (IDPs) bound to multiple protein ligands are found in numerous cellular systems. The 'beads-on-a-string' architecture that is common amongst such multivalent IDPs, consists of a highly flexible IDP "string" bound to multiple regulatory or scaffold protein "beads". The inherent conformational flexibility of the IDP, coupled with the potential compositional heterogeneity of ligand assemblies due to low binding affinities has made these systems difficult to characterize structurally. Electron microscopy (EM) has emerged as a powerful tool for structural characterization of heterogeneous protein complexes; however, in cases of continuum dynamics traditional "class averaging" effectively washes out the heterogeneity of primary interest. Furthermore, recently deployed methods in EM for characterizing such highly dynamic systems are not suitable for small proteins (e.g., < 50 kDa), due to a low signal-to-noise ratio. Here, we report automated analysis for a particular class of multivalent IDPs bound to ~20 kDa regulatory 'hub' proteins, which exhibit not only a multiplicity of bound species but also continuous conformational flexibility. The analysis (i) identifies oligomers and provides 'direct' counts of all species, (ii) statistically corrects the direct population counts for artifacts resulting from random proximity of unbound ligand 'beads', and (iii) provides conformational distributions for all species. We demonstrate our approach on a synthetic multivalent four-site IDP, which binds in a parallel duplex fashion to the ubiquitous hub protein, the LC8 homodimer. The duplex IDP architecture allows for potentially greater heterogeneity due to the possibility of off-register assemblies, which could in principle lead to runaway polymerization. We employ negative-stain EM (NSEM) because of its high contrast, which enabled direct visualization of individual LC8 homodimers for single particle analysis, although fundamentally our approach should be applicable to other 'beads-on-a-string'-like systems whenever there is sufficient contrast within the EM dataset. The automated analysis shows a heterogeneous population distribution of oligomeric species that are consistent with manually analyzed data. The statistical correction suggests that five-bead 'off-register' complexes identified in both automated and manual analysis, likely are four-bead oligomers extended by a randomly distributed free LC8 particle. Finally, significant conformational heterogeneity is resolved and characterized for the oligomeric assemblies that were not resolved by traditional 2D class averaging methods.