Amazon Mechanical Turk (MTurk) has generated considerable interest as a low-cost platform for survey and experimental research. A well-known limitation of MTurk is that its population of workers is not representative of the US population. It is often overlooked that individual samples of workers are also not representative of the population of MTurk workers as a whole. To examine potential differences across MTurk samples, a large (N ~10,000) sample of workers was recruited through postings on different days of the week and at different times of day (10am, 3pm and 10pm) over a period of two months. Workers reported demographic information about themselves and completed a brief personality inventory. Workers' prior experience completing research studies was estimated usinga database of ~100,000 previously recorded responses. Worker characteristics were examined as a function of day, time of day and serial position in which respondents completed the survey. Differences in time zone, mobile device use, and demographic characteristics emerged as a function of the day and time of day that data were collected. Differences also emerged as a function of serial position, demonstrating that sample composition systematically changed as sample size increased. These differences may be of particular importance to researchers interested in MTurk workers as a population in its own right (e.g., organizational researchers interested in the “gig economy”) and those seeking to recruit convenience samples of specific compositions. Substantial differences in worker’s prior experience with research emerged across day, time of day, and serial position of responses, reflecting that “professional” respondents complete surveys sooner and are particularly active earlier in the day and in the latter half of the week. This finding is of more general importance to those who wish to limit the proportion of professional respondents in a sample.