   _____   _____   __  __ 
  / ____| / ____| |  \/  |
 | |     | |  __  | \  / |
 | |     | | |_ | | |\/| |
 | |____ | |__| | | |  | |
  \_____| \_____| |_|  |_|

README

This is the Readme file for the code for Tables 1-5 from Crandall, Garavaglia, and Mongold (2026). 

The data were obtained through three experiments conducted using Qualtrics. These three experiments 
are described in Crandall, Garavaglia, and Mongold (2026).

The analyses for Tables 1, 3, 4, and 5 are completed in R. Because Table 2 involves binary data,
SAS's proc catmod function is used for these analyses.

The R code first imports the libraries to be used in the program. Next, the data are read into the
environment. The code filters out incomplete entries, test data, and responses from participants 
without investment experience, as our intended population does not comprise these participants. Next,
the code applies aesthetic/readability changes to variable names, combines counterbalanced measures 
into a single variable, creates variable labels for each individual factor, and adds an interaction term
for use in contrast coding. Additionally, because Experiment 1 has a +1 condition (the direct method
condition), the code subsets the data for Experiment 1 to include only the four primary conditions of 
the 2x2 factor design, for contrast testing and ANOVA. 

After the data are in a correct and readable format, the remaining code computes descriptive statistics 
and performs, where applicable, the ANOVAs, contrast testing, simple effects tests, and the planned contrast 
test between the direct method condition and our theoretically-motivated condition (linked presentation and 
Income Statement labeling). This process is performed on four data sets. dfE1 and dfE1noDM (the subset of 
dfE1 without direct method condition observations) are used for Tables 1 and 3, dfE2 is used for Table 4, 
and dfE3 is used for Table 5. 

Data for Table 2, coding of free responses, are entered manually as counts into SAS. The SAS code 
computes descriptive statistics and performs the categorical modeling procedure, simple effects tests, 
contrast testing, and the planned contrast between the direct method condition and our theoretically-
motivated condition (linked presentation and Income Statement labeling). 

For Table 1, our dependent variable is firm risk assessment, and our independent variables are whether 
the Statement of Cash Flows uses unlinked versus linked structure and Balance Sheet versus Income
Statement labeling. For Table 2, our dependent variable is whether participants mention risky receivables 
or payables increases in their free response, and our independent variables are whether the Statement of 
Cash Flows uses unlinked versus linked structure and Balance Sheet versus Income Statement labeling. For 
Table 3, our dependent variable is participants' likelihood of investing in the firm, and our independent 
variables are whether the Statement of Cash Flows uses unlinked versus linked structure and Balance Sheet 
versus Income Statement labeling. For Table 4, our dependent variable is firm risk assessment, and our 
independent variables are whether the Statement of Cash Flows uses the indirect method or the theoretically-
motivated cash flow presentation method (linked and Income Statement labeling), and whether the firm's 
operating liquidity risk is lower or higher. For Table 5, our dependent variable is firm risk assessment, 
and our independent variables are whether the Statement of Cash Flows uses the indirect method or the 
theoretically-motivated cash flow presentation method (linked and Income Statement labeling), and whether 
the participant is or is not primed to think of receivables risk before providing their response to the firm 
risk assessment dependent measure.