“…Positive bias is also plausible in cross-sectional surveys where the amount of exposure to violence is usually correlated with aggression scores, holding constant background vanables that are thought to correlate with viewing and a^ression The adequacy of this approach depends on how completely selection differences between heavier and hghter viewers are modeled and how well the constructs in the mo^l are measured Campbell and Bomch (1975) and Mumane, Newstead, and CHsen (1985) have argued that "underadjustment" is likely to occur m this situaticm, though not everyone agrees (e g , Cronbach, Rogossa, Floden, & Pnce, 1977) The synchrtmous correlation between viewing and a^ressKm is positive, althm^ usually small (m the 15 to 25 range, withmit rehabdity COTrectirais) If Campbell and Boruch (1975) are COTrect, s(»ne of the l^ckgnMUKl differences between heavier and lighter viewers will nrt be removed m covanence type analyses and will inflate estimates of tl^ rdatit»iship between television videix« and aggression Huesmann (1982) n^ntimis six reports of field expenments lnvdving raodran assignm^it, altlxM;^ Ik^rts and Maccxhy (1985) (nte ten In Huesmann's work fiMir of the studies w^re said to have yielded evidence of a positive relation between violence viewing and aggression, the exceptions being Fesbbach and Singer (1971) and Wells (1973) However, the study by Loye, Comey, and Steele (1977) is also an exception since tbe data sbow tbat tbe group assigned to a television diet of bigb violence was no more likely to commit hurtful behaviors than were controls Moreover, the expenment of Stem and Fnedncb (1972) showed weak effects confined to a subgroup of initially more aggressive children and, according to Parke, Berkowitz, Leyens, West, and Sebastian (1977), &iled to replicate (Sawin, 1973) Of tbe two other reports, the work of Leyens, Parke, Camino, and Berkowitz (1975), is included as one of three expenments m Parke et al (1977), which therefore emerges as the crucial report claiming to find effects of television violence However, these field expenments deal vwth institutionalized children where regular television viewing was forbidden, and in two studies the boys assigned to view aggressive films were initially more aggressive In our view and that of Freedman (1984), the field expenments on television violence produce little consistent evidence of effects It IS widely agreed that the best metbod for studying television effects involves longitudinal panel studies Unfortunately, the state of the analytic art has recently changed The formerly advocated analysis based on cross-lagged panel correlations is misleading when test-retest correlations differ between vanables Then, the less reliable measure will spuriously appear to be causal (Cook & Campbell, 1979, Rogosa, 1980 Since television viewing is nearly ahvays less stable than aggressiveness and since cross-lagged analysis was used in all but two of the past panel studies, a bias may have operated to inflate the television-aggression relationship Reliance on evidence from multiple heterogeneous methods of st...…”