Data Editors, Survey Design, and Data Cleaning
Chart Creation and Data Analysis
Sophia Grace Capili
Our survey was conducted over a period of 16 days, from June 23 to July 8, 2022. On three occasions, The Daily Princetonian emailed questions, assembled in a Google Form, to members of the Class of 2026 included in the Residential College Student Facebook as of June 21 — a total of 1,528 people. All told, we received 925 responses, comprising just over 60 percent of the class: a record-high response rate.
Every question on the survey was optional. On select sensitive questions, respondents could explicitly decline to answer. Although all responses were fully anonymous, the Google Form was limited to individuals with a University- issued email address, and a feature from Google enabled the ‘Prince’ to prevent multiple submissions. To protect respondents’ anonymity and data, only select members of our survey team received access to the raw data, and none made any effort to identify individual respondents. As part of our commitment to comprehensive privacy, we collected and aggregated all raw data on Google accounts unaffiliated with the University.
Upon receiving an abundance of submissions, we embarked on a months-long process of compiling and checking each tabulation. A minimum of two ‘Prince’ editors from the Data team verified every datapoint.
We exercised discretion when presenting the data in question. Our team carefully considered the implications of various cross-tabulations and sought to ensure that the responses of individual students may not be identified. When utilizing percentages, rather than using raw numbers to represent data, our team rounded the value to the nearest tenth of a percent. As every query was optional, many questions garnered sample sizes one to two percent smaller than the overall respondent pool; in our narratives, “respondent” refers to any student who responded to a given question.
Despite our best efforts, our survey includes limitations. We did not complete statistical analysis of data points cross-tabulated from two or more questions, nor did we make any attempt to increase or decrease the weight of various demographic groups to more accurately match the overall picture of a larger group. All the same, a check of our respondent pool’s composition against data previously released by the University indicates a fairly representative and large sample.
In sincerity, all which we report is a little out of date; between the survey and now, some respondents have likely had their first drink, experienced changes in their financial situation, and changed their concentration. Then switched it back. Then changed it again. Our survey constitutes a glance at the Class of 2026 as it existed in June and early July, unresponsive to other life interruptions which may have taken place since.