Fielding Without a Frame: Design, Paradata and Data Quality in a Mobile Survey of Lahore Students

Abstract

Survey research on political attitudes outside high-income democracies usually proceeds without a sampling frame and without platform data against which self-reports can be checked. What remains is the instrument and the record it keeps of its own administration. This article reports the design and the measured data quality of a web survey of university students in Lahore, Pakistan (analytic N = 779, fielded 23 February to 15 March 2026), covering institutional trust, political conflict online and support for extra-constitutional alternatives. A pilot on a general-purpose form platform failed in four ways, and each failure was answered by a feature of a purpose-built application: one experimental list per respondent, section-level timing, completion enforced at the section level while refusal remained available, and local caching for interrupted mobile sessions. The paradata show a sample arriving in institution-led bursts, with 32.7% of responses on a single day; a median completion time of 11.4 minutes against the 10 to 15 announced at consent, with only 35.3% of respondents inside that window; and 98.8% mobile completion. Refusal followed the international pattern for financial items, at 20.9% on a household-asset question against 1.7% on direct support for a military takeover. Consent refusals, breakoff points and duplicate submissions were not recorded at all, and the article specifies the fields that would have captured them. Where a design cannot warrant its sample, the administration is the evidence, and it should be reported to the standard usually reserved for results.