The "involvement effect" refers to the finding that controlling for gambling involvement often reduces or eliminates frequently observed game-specific associations with problem gambling. In other words, broader patterns of gambling behavior, particularly the number of types of games played over a defined period, contribute more to problem gambling than playing specific games (e.g., lottery, casino, Internet gambling). This study extends this burgeoning area of inquiry in three primary ways. First, it tests independently and simultaneously the predictive power of two gambling patterns: breadth involvement (i.e., the number of games an individual plays) and depth involvement (i.e., the number of days an individual plays). Second, it includes the first involvement analyses of actual betting activity records that are associated with clinical screening information. Third, it evaluates and compares the linearity of breadth and depth effects. We conducted analyses of the actual gambling activity of 1,440 subscribers to the bwin.party gambling service who completed an online gambling disorder screen. In all, 11 of the 16 games we examined had a significant univariate association with a positive screen for gambling disorder. However, after controlling for breadth involvement, only Live Action Internet sports betting retained a significant relationship with potential gambling-related problems. Depth involvement, though significantly related to potential problems, did not impact game-based gambling disorder associations as much as breadth involvement. Finally, breadth effects appeared steeply linear, with a slight quadratic component manifesting beyond four games played, but depth effects appeared to have a strong linear component and a slight cubic component.