It's respondents that are better off with a three member panel, not complainants (although, on occasion, complainants *do* request a three member panel anyway). The reason is that there tends to be a pro-complainant bias in the panelists in general, so a randomly-selected panelist is more likely than not to be in favor of the complainant; a three-member panel, however, gives the respondent a chance to pick one member, allowing a pro-respondent panelist to be picked.