Pants could not simply fabricate items in order to lengthen their lists. Both coders were not blind to the hypotheses of the study, but they were blind to the initial ratings and therefore could not predict whether the coding of any given item would confirm or deny the hypotheses. Inter-rater reliability was analyzed with a Spearman RankOrder Correlation across individual items, and was good (rs[383] = .884). The codes of the first coder were used for all analyses. Overall, 181 Duvoglustat supplement differences (28.5 of all provided) were coded as invalid across all twelve items and 29 participants, with a maximum of 31 excluded for any individual item (Cucumber ?Zucchini). The exclusions were due to either factual inaccuracy, verified by external sources (e.g., “cucumber has seeds zucchini doesn’t”), or failing to follow the directions regarding acceptable differences (e.g., “Jam can also refer to a sticky situation in which you are stuck.”). As we predicted, H 4065 chemical information adults provided more differences for Known items (M = 1.856, SD = .866) than Unknown items (M = .656, SD = .761), t(58) = 5.698, p < .001. This validates the categorization from our pilot study for these twelve items. 3.2.3. The MM effect--Fig. 2 shows the proportion of participants that showed an MM effect on each item. As predicted, participants generally estimated that they would be able to list more differences than they were actually able to (the "overestimate" bars in Fig. 2). The effect was analyzed in sign tests for each item, to better determine how common the MM effect was rather than how large it tended to be. Eight items were significant and in theCogn Sci. Author manuscript; available in PMC 2015 November 01.Kominsky and KeilPageexpected negative direction (ps < .05), three Known items and five Unknown items. One item was significant in the opposite direction (Wool-Silk, p < .05). For the remaining three items, all were in the expected negative direction, but marginal (Blackbird ?Starling, p = . 057; Rowboat ?Canoe, p = .1) or nonsignificant (Donkey ?Mule, p > .2). While not completely uniform, it is still clear that in the majority of cases adults showed the MM effect. Readers may notice that this effect could be explained as a by-product of the 181 provided differences that were excluded in coding. Participants’ estimates may have reflected the number of things they felt they could list, not realizing that some of the knowledge they possessed was inaccurate. This would be a very different and much less interesting effect – it is no surprise that people are unaware that some of their knowledge is inaccurate (e.g., Fischhoff, Slovic, Lichtenstein, 1977). Rather, according to the MM effect predicted here, people are actually misjudging the amount of knowledge they possess. In order to rule out this deflationary interpretation, a separate set of sign tests were conducted using all of the provided differences (regardless of accuracy or rule adherence). The MM effect was still significant for four items (ps < .05), marginally significant for three additional items (ps < . 1), and only one item showed a significant effect in the opposite direction (Wool ?Silk, p = . 007). In short, while excluding inaccurate responses did make the MM effect more consistent, the effect exists even when accuracy is ignored. For all future analyses and studies, we elected to focus on accurate differences only, again to rule out the possibility that some participants might fabricate additional differences. Turni.Pants could not simply fabricate items in order to lengthen their lists. Both coders were not blind to the hypotheses of the study, but they were blind to the initial ratings and therefore could not predict whether the coding of any given item would confirm or deny the hypotheses. Inter-rater reliability was analyzed with a Spearman RankOrder Correlation across individual items, and was good (rs[383] = .884). The codes of the first coder were used for all analyses. Overall, 181 differences (28.5 of all provided) were coded as invalid across all twelve items and 29 participants, with a maximum of 31 excluded for any individual item (Cucumber ?Zucchini). The exclusions were due to either factual inaccuracy, verified by external sources (e.g., "cucumber has seeds zucchini doesn't"), or failing to follow the directions regarding acceptable differences (e.g., "Jam can also refer to a sticky situation in which you are stuck."). As we predicted, adults provided more differences for Known items (M = 1.856, SD = .866) than Unknown items (M = .656, SD = .761), t(58) = 5.698, p < .001. This validates the categorization from our pilot study for these twelve items. 3.2.3. The MM effect--Fig. 2 shows the proportion of participants that showed an MM effect on each item. As predicted, participants generally estimated that they would be able to list more differences than they were actually able to (the "overestimate" bars in Fig. 2). The effect was analyzed in sign tests for each item, to better determine how common the MM effect was rather than how large it tended to be. Eight items were significant and in theCogn Sci. Author manuscript; available in PMC 2015 November 01.Kominsky and KeilPageexpected negative direction (ps < .05), three Known items and five Unknown items. One item was significant in the opposite direction (Wool-Silk, p < .05). For the remaining three items, all were in the expected negative direction, but marginal (Blackbird ?Starling, p = . 057; Rowboat ?Canoe, p = .1) or nonsignificant (Donkey ?Mule, p > .2). While not completely uniform, it is still clear that in the majority of cases adults showed the MM effect. Readers may notice that this effect could be explained as a by-product of the 181 provided differences that were excluded in coding. Participants’ estimates may have reflected the number of things they felt they could list, not realizing that some of the knowledge they possessed was inaccurate. This would be a very different and much less interesting effect – it is no surprise that people are unaware that some of their knowledge is inaccurate (e.g., Fischhoff, Slovic, Lichtenstein, 1977). Rather, according to the MM effect predicted here, people are actually misjudging the amount of knowledge they possess. In order to rule out this deflationary interpretation, a separate set of sign tests were conducted using all of the provided differences (regardless of accuracy or rule adherence). The MM effect was still significant for four items (ps < .05), marginally significant for three additional items (ps < . 1), and only one item showed a significant effect in the opposite direction (Wool ?Silk, p = . 007). In short, while excluding inaccurate responses did make the MM effect more consistent, the effect exists even when accuracy is ignored. For all future analyses and studies, we elected to focus on accurate differences only, again to rule out the possibility that some participants might fabricate additional differences. Turni.