<p>Geoengineering represents a set of options to reduce the greenhouse effect that causes global climate change. While public support is necessary for effective governance and implementation, the lay public has low awareness of its existence, types, and mechanisms. Prior studies have found that providing information to the public can affect support, but there are conflicting results. This study examines to what extent these differences may be the result of survey sampling rather than the information itself. Survey respondents were randomly assigned to receive either basic or extensive information about five types of geoengineering: space mirrors, stratospheric aerosol injection, ocean fertilization, direct air capture, and afforestation. This design was replicated across four online survey samples gathered from address-based sampling, an opt-in online panel, Amazon Mechanical Turk (MTurk), and a college classroom. Across analyses, there was general consistency of effects between the address-based sample, the opt-in sample, and the classroom sample, with limited exceptions. Effects from the MTurk sample, however, were uncorrelated with the other samples, diminishing their utility, even when filtering out responses that failed two attention checks. As new climate technologies come out, it is important to test which information is given to the public in surveys to generate useful opinions. In general, we found that researchers should be able to use many kinds of samples, besides MTurk, for initial studies.</p>