The data files uploaded here include: **ADOScodes.csv** This spreadsheet shows participant scores on each behavioural code from A1 to E3 on the ADOS-2; then scores are shown for the 11 items contributing to the ADOS-2 algorithm are repeated. SubtotalCOMM is the total score on communication items (participants need to score at least 2 for an autism spectrum classification and 3 for an autism classification). SubtotalSOCIAL is the total score on the social interaction items (participants need to score at least 4 for an autism spectrum classification and 6 for autism). TOTAL is the sum of the two subtotals. Classification is the outcome of the algorithm. Z-scores are based on normative data collected in the Adult Psychiatric Morbidity Survey (APMS; National Centre for Social Research, 2017). Brugha et al. (2011) report on prevalence of autism based on a representative study of more than 7000 UK adults, whose data were weighted according to demographic variables. The ADOS-2 was administered to 617 of these individuals, who were oversampled based on probability of autism according to their score on the Autism Spectrum Quotient (AQ, Baron-Cohen, Wheelwright, Skinner, Martin, & Clubley, 2001). Since the sampling weights (based on demographics and probability of being selected for interview) are known, it is possible to generalise scores to the general adult population, and so we computed percentiles and z-scores based on the data. See script calculateNormsADOS_APMSdata.R uploaded here for our method. **AwkwardDialogues.csv** This spreadsheet shows scores on each of the five awkward dialogues and associated factual recall questions. The factor score is the summary score for the test, and was generated through item response theory modelling. Outliers include any individuals who answered more than one invalid score on an awkward dialogue. **Backchannel.csv** This spreadsheet shows item-level data and total scores on the Test of Fillers/Backchannels. **CCSR.csv** This spreadsheet shows item-level data on the Communication Checklist - Self Report; 0 to 3 indicate frequency, and NA represents a missing item. For the structural language and pragmatic language composites, we present raw total scores, number of missing items on the composite, corrected score and z-score from the CC-SR manual. Where participants missed one or two items were missing for a composite, scores were prorated as "corrected scores". Where participants missed more than two items, corrected score is recorded as NA. Note that two items were omitted due to a technical error on the structural scale, and so participants effectively needed to answer all questions on this scale or their total was recorded as NA. **DemographicsAutistic.csv and DemographicsNonautistic.csv** These spreadsheets show gender, student status and highest level of education for each participant. Note that the gender variable has been slightly altered to preserve anonymity. **Grammar.csv** This spreadsheet shows item-level data and total scores on the Grammaticality Decision Test. Outliers scored lower than 2.2 times the interquartile range below the first quartile. **Implicature.csv** This spreadsheet shows accuracy, response and reaction time on all items of the Implicature Comprehension Test. It also includes total score on the implicature items (imp_total), total score on the explicit-response control items (exp_acc), number of time-outs on implicature items (Ntimeouts), and coding for outlying participants (0 or 1) based on whether a participant scored at least 7 on the control items of the test. **Inferencing.csv** This spreadsheet shows item-level data and total scores on the Test of Local-level Inferencing. **IQ.csv** This spreadsheet shows item-level data and total scores for our measure of general cognitive ability, the Sample Test from the International Cognitive Ability Resource (ICAR). **TOManimations.csv** This spreadsheet shows item-level data on the five theory of mind animations and four control animations. Summary scores on this test are a factor score extracted from the theory of mind items by item response modelling and raw total on the control items. **Vocab.csv** This spreadsheet shows item-level data and total scores on the Synonymns Test.