Surveying Surveys and Questioning Questions : Learning from World Bank Experience

Recanatini, Francesca ; Wallsten, Scott J. ; Xu, Lixin Colin

World Bank, Washington, DC

关键词: BANK LOANS;    BANKING SYSTEM;    BANKRUPTCY;    BIDDING;    BIRTH RATE;   

DOI  :  10.1596/1813-9450-2307
RP-ID  :  WPS2307

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

The World Bank has increasingly focusedon firm-level surveys to build the data foundation neededfor accurate policy analysis in developing and transitioneconomies. The authors take stock of some recent Banksurveys, and discuss how to improve their results. Lessonson data issues, and hypothesis testing: 1) Use panel data,if possible. 2) Have enough information about productivityto estimate a production function. 3) Avoid the paradigm of"list the severity of the obstacle/problem on a scaleof 1 to 5". Instead, ask for data on specificdimensions of the problem that will shed light onalternative hypothesis and policy recommendations. 4) Pickparticular disaggregated industries, and sample thoseindustries in each survey. 5) Identify the most importantinterventions of interest, and consider how you willempirically identify specific changes by picking instrumentsuseful for doing so. Lessons on questionnaire design: a)Incorporate only one idea or dimension in each question. Donot ask, in one question, about the "quality,integrity, and efficiency" of services, for example. b)Consider the costs and benefits of numeric scales comparedwith adjectival scales. Scales in which each point islabeled may be more precise than numeric scales in whichonly the end points are labeled. But responses are verysensitive to the exact adjective chosen, and it may beimpossible to translate adjectives precisely acrosslanguages, making it impossible to compare responses acrosscountries. c) Recognize that the share of respondentsexpressing opinions will be biased upward if the survey doesnot include a middle ("indifferent" or"don't know") category, and downward if itdoes include the middle category. d) When askingdegree-of-concern and how-great-an-obstacle question,consider first asking a filter question (such as "Doyou believe this regulation is an obstacle or not?").If the answer is yes, then ask how severe the obstacle is.e) Be aware of the effects of context. The act of askingquestions can affect the answers given on subsequent,related questions. f) Think carefully about how to asksensitive questions. Consider using a self-administeredmodule for sensitive questions. alternatively, a randomizedresponse mechanisms may be a useful, truth-revealing mechanism.