Title: Museum question-asking data file

Summary: This data contains data collected from a mobile question-asking app implemented at two museums in American Southwest. A total of 1, 331 visitor groups generated 11,909 questions using the app. 

Variables: In this study, we examined three strands of data. The first strand was app data that was directly gleaned from the app, including questions typed in the app, counts of questions, and other log data. The second strand was survey data from pre-post questionnaires. The third strand was researcher-coded data. For the third strand, we scored each question collected from the app based on our coding taxonomy. To address our research questions, we identified three dependent variables (DVs) and five independent variables (IVs) drawn from the literature and the available data. Note that the three DVs in combination represent the question-asking behaviors that revealed information on question-asking quality, frequency, and duration respectively. The five IVs were considered proxy measures to represent cognitive, instrumental, and socio-contextual factors recognized in the literature. All the variables included in this study were listed below and could be found in the column headings of our tabular data. 

Question-asking quality (DV1): To assess visitors question-asking quality, we calculated the average question score per visitor group. Since the app collected all the questions asked by each visitor group, we carefully reviewed each question and assigned them to the corresponding question type or question level based on our coding rubric (See Table 3). Because some groups had more members and thereby asked more questions, we decided to use the average score for each question asked by a group rather than the overall score per group as the proxy measure for question-asking quality. 
Question-asking frequency (DV2): We simply counted the number of questions asked per visitor group. 
Question-asking duration (DV3): We simply estimated the minutes spent in the Dr. D question-asking app per visitor group. Please note that we used a paper timesheet to record the start and end times that each visitor group had a device in their possession when we handed out the mobile device and collected them again. 

Cognitive factor (CF). The cognitive factor was retrieved from the pre-questionnaire where we asked visitors about their frequency of participation in free time science activities. 
* Epistemic curiosity (CF_IV1): Participants were asked In your free time, how often do you 1) read news about science? 2) read books about science? 3) visit websites about science? 4) watch TV about science? 5) talk to others about science? 6) visit science-related museums? 7) do science activities/projects? and the choices were Never/almost never, Once a year, A few times a year, Monthly, Weekly, Daily. We scored the choices from 1 (Never/almost never) to 6 (Daily). We calculated an average composite score as a proxy measure for epistemic curiosity per group. This variable was continuous in our analysis. 




Instrumental factors (IF). The instrumental factors were sourced directly from the app. 
* App mode (IF_IV2): Participants were randomly assigned to two app modes: Ask mode and Game mode. 
* App performance (IF_IV3): The app used a tiered response system to document how certain it was in responding to each question. For the analysis, we estimated the average tier score for each visitor group as the proxy measure for app performance. This variable was a continuous one. 

Socio-contextual factors (SCF). The socio-contextual factors considered survey data such as self-reported group question-asking frequency collected from the post questionnaire. 
* Museum layout (SCF_IV4): Participants were from two different museums which featured different layouts. In Museum A, the Dr. D app was used in a nonlinear exhibit whereas Museum B implemented the Dr. D app with visitors in a linear exhibit.   
* Group inquiry frequency (SCF_IV5): Participants were asked to rate how often did your group discuss what questions to ask Dr. D? on a 1 to 5 Likert scale (1 = not at all; 5 = every time). This variable was included in our analysis as an ordinal variable. 

