Do Data Collection Procedures Influence Political Knowledge Test Performance?
Sample size: 800
Field period: 8/26/2010-1/4/2011
Valid political knowledge measures are key to comprehending public opinion data. Recent shifts toward collecting data online, although efficient, may introduce biases that effect substantive conclusions. This experiment builds upon previous research that found when open-ended factual knowledge questions were asked, scores were higher when the test was administered online than when administered in pencil and paper form in a classroom. The focus of this experiment is to identify potential explanations for this effect. Plausible explanations include differential time pressure and norm induced motivation to cheat. The sample was composed of 1600 participants from a nationally representative sample collected by TESS. The design was a 2 (online, phone) X 2 (low/high time pressure) X 2 (high/low norms fully crossed and balanced independent groups design). Understanding artifacts introduced in public opinion data based on collection procedures better informs the decision to use these methods and improves measurement for this influential construct.
H1: Participants should score higher on a political knowledge test when the test is hosted online rather than over the phone.
H2: Participants should score higher on a political knowledge test when there is a high norms induction rather than a low norms induction online.
H3: Participants should score higher on a political knowledge test when there is low time pressure rather than high time pressure.
Venue: whether the test was hosted online versus over the phone
Norms: whether participants received instructions indicating the importance and commonality of political knowledge (high norms) versus instructions indicating low knowledge importance and low commonality of knowledge (low norms)
Time Pressure: whether participants were additionally told in the test instructions to hurry and provide answers quickly (high time pressure) versus to take their time before they answer (low time pressure).
Political knowledge test performance (# of correct answers out of 5)
The only main effect uncovered in this manuscript was the finding that time pressure adversely affected test scores, such that participants told to answer quickly did worse on the test than those told to take their time. In addition to this main finding, there were several interaction effects uncovered based on age, time taken to complete the test, and admitting to receiving help on the test. These findings suggest that online surveys do create the opportunity to look up answers, and some people take advantage of this resource. Secondly, experimental instructions can serve to facilitate or inhibit this activity, but might have additional consequences based on the age of the sample. These findings are discussed further at the end of the manuscript.
A question asked participants whether they "used outside help" when responding to the questions. Approximately 6.6% of the sample admitted to receiving help, and this percentage largely came from participants in the online condition and in the low time pressure condition.
Boster, F. J. & Shulman, H. C. (under review). Political Knowledge Test Performance as a Function of Venue, Time Pressure, and Performance Norms. Paper submitted to the annual meeting of the International Communication Association in London, UK.