Modeling anchoring effects in sequential Likert scale questions
Surveys in many different research fields rely on sequences of Likert scale questions to assess individuals' general attitudes toward a set of related topics. Most analyses of responses to such a series do not take into account the potential measurement error introduced by the context effect we dub "sequential anchoring," which occurs when the rating for one question influences the rating given to the following question by favoring similar ratings. The presence of sequential anchoring can cause systematic bias in the study of relative ratings. We develop a latent-variable framework for question responses that capitalizes on different question orderings in the survey to identify the presence of sequential anchoring. We propose a parameter estimation algorithm and run simulations to test its effectiveness for different data-generating processes, sample sizes, and orderings. Finally, the model is applied to data in which eight payment instruments are rated on a five-point scale for each of six payment characteristics in the 2012 Survey of Consumer Payment Choice. We find consistent evidence of sequential anchoring, resulting in sizable differences in properties of relative ratings for certain instruments.
AUTHORS: Hitczenko, Marcin
Optimal recall period length in consumer payment surveys
Surveys in many academic fields ask respondents to recall the number of events that occurred over a specific period of time with the goal of learning about the mean frequency of these events among the population. Research has shown that the choice of the recall period, particularly the length, affects the results by influencing the cognitive recall process. We combine experimental recall data with use data to learn about this relationship in the context of consumer payments, specifically for the mean frequency of use of the four most popular payment instruments (cash, credit card, debit card, check). Overall, our analysis suggests that day-based recall is inefficient, with mean-squared errors of population estimates minimized for longer recall periods, although the optimal recall period differs among payment instruments. In addition, for cash, we develop a model relating recalled values to individual frequency of use in order to study the relationship between demographic variables and accuracy at different recall lengths. We find little link between demographic characteristics and accuracy of different recall periods for an individual.
AUTHORS: Hitczenko, Marcin
Battery order effects on relative ratings in Likert scales
Likert-scale batteries, sequences of questions with the same ordinal response choices, are often used in surveys to collect information about attitudes on a related set of topics. Analysis of such data often focuses on the study of relative ratings or the likelihood that one item is given a lower (or higher) rating than another item. This work studies how different orderings of the items within a battery and, in particular, the relative location of items affect relative rating distributions. We take advantage of data from the 2012?2014 Survey of Consumer Payment Surveys, in which item order in six Likert-scale batteries is varied among respondents. We find that ordering effects are real and consistent across years. The most prominent effect relating to relative locations of items is that the farther one item is placed after another item, the more likely that item is to have a lower rating.
AUTHORS: Hitczenko, Marcin
Estimating population means in the 2012 Survey of Consumer Payment Choice
This report examines the effect of adding to a longitudinal panel on estimates of population parameters in the 2012 Survey of Consumer Payment Choice (SCPC) more than 1,000 newly recruited respondents specifically targeted to fill segments of the U.S. population that tend to be underbanked and underrepresented in the longitudinal panel. In many ways, the new respondents have fundamentally different characteristics from the ongoing respondents. To minimize confounding sources of change to annual estimates when making comparisons across years, the official 2012 SCPC publication was based on the longitudinal panel only. This research data report presents estimates based on all available responses and evaluates the effect of the additional panel on these estimates. A complete set of 2012 SCPC tables with estimates based on the full sample accompanies this report.
AUTHORS: Hitczenko, Marcin
Texas Manufacturing Outlook Survey: survey methodology and performance
The Texas Manufacturing Outlook Survey (TMOS) is a monthly survey of area manufacturers conducted by the Federal Reserve Bank of Dallas. TMOS indexes provide timely information on manufacturing activity in Texas, which is useful for understanding broader changes in regional economic conditions. This paper describes the survey methodology and analyzes the explanatory and predictive power of TMOS indexes with regard to other measures of state economic activity. Regression analysis shows that several TMOS indexes successfully explain monthly changes in Texas employment and quarterly changes in gross state product. Forecasting exercises show that several TMOS indexes, particularly general business activity and growth rate of orders, are useful in predicting changes in Texas employment.
AUTHORS: Canas, Jesus; Kerr, Emily
Texas Service Sector Outlook Survey: Survey Methodology and Performance
The Texas Service Sector Outlook Survey (TSSOS) and Texas Retail Outlook Survey (TROS) are monthly surveys of service sector and retail firms in Texas conducted by the Federal Reserve Bank of Dallas. TSSOS and TROS track the Texas private services sector, including general service businesses, retailers and wholesalers. The surveys provide invaluable information on regional economic conditions?information that Dallas Fed economists and the Bank president use in the formulation of monetary policy. This paper describes the survey?s methodology and analyzes the explanatory and predictive power of TSSOS and TROS indexes with regard to Texas employment growth. Regression analysis shows that several TSSOS and TROS indexes help explain monthly variation in Texas employment. In addition, most TSSOS and TROS indexes are also useful in forecasting Texas employment growth.
AUTHORS: Canas, Jesus; Jordan, Amy
Heaping at Round Numbers on Financial Questions : The Role of Satisficing
Survey responses to quantitative financial questions frequently display strong patterns of heaping at round numbers. This paper uses two studies to examine variation in rounding across questions and by individual characteristics. Rounding was more common for respondents low in ability, for respondents low in motivation, and for more difficult questions, all consistent with theories of satisficing. Questions that require more difficult information retrieval and integration of information exhibit more heaping. The use of records, which lowers task difficulty, reduces rounding as well. Higher episodic memory is associated with less rounding, and standard measures of motivation are negatively associated with rounding. These relationships, along with the fact that longer response latencies are associated with less rounding, all support the idea that rounding is a manifestation of satisficing on open-ended financial questions. Rounding patterns also appear remarkably similar across the two studies, despite being fielded in different modes and employing different question order and wording.
AUTHORS: Hsu, Joanne W.; Gideon, Michael; Helppie-McFall, Brooke
Lining Up : Survey and Administrative Data Estimates of Wealth Concentration
The Survey of Consumer Finances (SCF) has a dual-frame sample design that supplements a standard area-probability frame with a sample of observations drawn from statistical records derived from tax returns. The tax-based frame is stratified on the basis of a "wealth index" constructed largely from observed income flows, with the intent of heavily oversampling wealthy households. Although the SCF is not specifically designed to estimate wealth concentration, the design arguably provides sufficient support to enable such analysis with a reasonable level of credibility. Similar estimates may also be made by using tax-based data directly, as in , by using a construct very close to a key part of the SCF wealth index. Such an approach has appeal as a way of tapping a much larger set of information to improve SCF estimates. Not surprisingly, there are differences in the two approaches, largely as a result of conceptual differences or complications in the survey implementation. This paper focuses on the top 1 percent of the wealth distribution, the group most intensively covered by the SCF list sample and it explores the stability of the relationship between the patterns of concentration in the survey data and parallel patterns in tax-based estimates and considers how those patterns differ across survey participants, the full sample and the entire survey frame. In addition, the paper makes as series of recommendation for further research on the technical support of the survey.
AUTHORS: Kennickell, Arthur B.
Reaching the Hard to Reach with Intermediaries: The Kansas City Fed’s LMI Survey
Reaching hard-to-reach individuals is a common problem in survey research. The low- and moderate-income (LMI) population, for example, is generally hard to reach. The Kansas City Fed?s Low- and Moderate-Income Survey addresses this problem by sampling a database of organizations to serve as proxies for the LMI population. In this paper, I describe why the LMI population can be hard to reach. I then explore potential problems with using a nonrandom survey sample and address the empirical validity of the Kansas City Fed?s LMI Survey. I compare results from the survey using the standard sample to results from the survey using a random sample. I find that the results of the surveys using the standard and random samples are not significantly different and conclude that the use of a nonrandom sample is not a significant problem for the LMI Survey. I find that the series of responses from the LMI Survey are correlated with the things they should be correlated with, suggesting that the survey is empirically valid and does a good job of measuring economic conditions in LMI communities.
AUTHORS: Edmiston, Kelly D.
We Are All Behavioral, More or Less: Measuring and Using Consumer-Level Behavioral Sufficient Statistics
Can a behavioral sufficient statistic empirically capture cross-consumer variation in behavioral tendencies and help identify whether behavioral biases, taken together, are linked to material consumer welfare losses? Our answer is yes. We construct simple consumer-level behavioral sufficient statistics??B-counts??by eliciting seventeen potential sources of behavioral biases per person, in a nationally representative panel, in two separate rounds nearly three years apart. B-counts aggregate information on behavioral biases within-person. Nearly all consumers exhibit multiple biases, in patterns assumed by behavioral sufficient statistic models (a la Chetty), and with substantial variation across people. B-counts are stable within-consumer over time, and that stability helps to address measurement error when using B-counts to model the relationship between biases, decision utility, and experienced utility. Conditional on classical inputs?risk aversion and patience, life-cycle factors and other demographics, cognitive and non-cognitive skills, and financial resources?B-counts strongly negatively correlate with both objective and subjective aspects of experienced utility. The results hold in much lower-dimensional models employing ?Sparsity B-counts? based on bias subsets (a la Gabaix) and/or fewer covariates, illuminating lower-cost ways to use behavioral sufficient statistics to help capture the combined influence of multiple behavioral biases for a wide range of research questions and applications.
AUTHORS: Zinman, Jonathan; Stango, Victor