We started with running extensive forward searchers on the first study to use what is now considered the delay time marshmallow test with both rewards present (e.g. Mischel & Ebbesen, 1970). Each study was then filtered for includion using the following criteria: Children under the age of ten Both rewards present Data only coded as delay time No or minimal additional changes to procedure (e.g. no accumulating rewards, not done in an fmri scanner, etc.). We also sent an email request to the cognitive development society asking for unpublished data. The email was as follows: John Protzko <protzko@gmail.com> 9/27/16 to cogdevsoc To the CDS listserv: Please post the following message to the Cognitive Development Listserv Dear Cognitive Development Colleagues, We are currently carrying out a meta-analysis of studies using the Mischel 'Marshmallow' task of delay of gratification. The goal is to look for secular changes from the original data to the present day in children's ability to delay gratification. If you have completed a study in children of any age where you have administered the marshmallow delay of gratification task (note: can also have used other stimuli, such as toys, stickers etc.) or if you have any information that would be relevant to this analysis, please contact us at either protzko@gmail.com or protzko@psych.ucsb.edu Thank you very much. Best; John Protzko This resulted in an additional four studies that met the inclusion criteria. We plan to run the meta-analysis using a meta-regression. The DV will be delay time. We will condition on participant's age, due to the increasing ability to delay gratification with age. We will concurrently test whether year the data was collected alters childrens ability to delay gratificaiton. For year data collecdted, we will use the following rubric: If the year data was collected is in the article, that is the year the data was collected. Unless the data is archival or from a separate study, peer-reviewed research articles will be coded as collected two years prior to publication year. Masters these and Dissertations will be coded as one year before publication date. Any other decisions will be coded in the data table. Our code for running the meta-analysis will be metareg delaymin age yearc, wsse(se) yearc is year data was collected centered by subtracting 1968 from the year (1968 being the first study). standard errors are standard errors of the mean (sd/(N^.5)). Any missing data will be imputed using multiple imputation in the mi package in R.