expectations about future trends in local e-government development. Two of those statements illustrate the point, ‘‘For the next few years at least, most local government Web sites can be expected to remain mainly informational with limited transactional capabilities.’’And,‘‘as has been the case with IT in government in general, payoffs will lag adoption.’’
The following questions are suggested by Norris and Moon’s expectations ‘‘How rapidly do local governments adopt e-government applications that have demonstrated payoffs?’’or, ‘‘What are the payoffs that local governments realize from implementing e-government?’’and,‘‘Do local governments adopt applications that have demonstrated payoffs more quickly than other types of applications.’’
A common difficulty in developing a research question consists of defining a solution into the problem statement. Bardach provides examples that illustrate this problem and how to avoid it. For example,‘‘Don’t say:‘There is too little shelter for homeless families’.’’This implies one specific solution. A better research question is:‘‘There are too many homeless families.’’Or, for example,
‘‘Don’t say:‘New schools are being built too slowly’.’’This problem statement implies a solution.
A more effective alternative leads to exploration of many different solutions and evaluation of them on their merits. In light of these criteria, a better alternative is‘‘There are too many schoolchildren relative to the currently available classroom space’’(Bardach, 2000, pp. 6–8).
How questions are framed determines their usefulness in focusing a research project. This is usually an iterative process in which one problem after another is eliminated from the draft of a question until the researcher is left with a usable question. A series of questions and discussion of their revision illustrates this process starting with the following question:
Question 1: How rapidly do governments adopt IT applications?
As this question stands, it is too vague to be useful. The words or phrases‘‘rapidly,’’ ‘‘govern- ments,’’and‘‘IT applications’’are too broad to give the research a clear enough focus to be useful. Thus [vague] [type of government?] [type of IT?]
Q1: HowrapidlydogovernmentsadoptIT applications?
Revision of a research question is an iterative process in which vague words are replaced with more specific ones and the focus is made more exact or sharpened. Here the general word
‘‘governments’’is replaced with‘‘local governments’’and the general phrase‘‘IT applications’’is replaced with‘‘e-government.’’
The focus on local government comes from the fact that the International City and County Managers Association (ICMA) already collects relevant data on IT and e-government at intervals from cities with populations greater than 20,000, and will sell it to a researcher for a reasonable price. This means two important things. It would probably not make sense to collect new data on your own and an investigation could reasonably be limited to those cities. This data provides the focus on e-government which is a hot topic in the public administration literature today.
[vague]
Q2: Howrapidlydolocal governmentsadopte-government [what payoffs]?
applications that havedemonstrated payoffs?
The second version still contains vague words such as‘‘rapidly’’and‘‘demonstrated payoffs.’’
The phrase‘‘demonstrated payoffs’’is derived from Norris and Moon’s discussion in their conclusion.
Q3:‘‘Are local government adoption rates for e-government applications that have [a specific grouping factor]
proven to work successfully in other cities [a specific measure]
faster than adoption rates of other e-government applications?’’
The third version of the question focuses more narrowly on comparison of the adoption rates of e-government applications that have or have not been successful in other cites. Copying what works is one explanation for the spread of technology. This explanation may or may not explain the spread of specific e-government applications. Other factors such as size of a city, demand, and cost might limit the adoption rate too. Demonstrated payoffs could be measured in other ways as well such as cost-effectiveness (lower cost per transaction) or improved=lowered response times or lowered complaint rates as well.
Another way to focus this question might be to make it even more specific. For example, this question could be focused more tightly by specifying which aspects of e-government to examine.
For example, if the question were limited to the adoption of full e-government portal capabilities the following question might result:
Q4: Doescity size impactadoption ofe-government portal capability?5
This question contains three issues. What is meant by city size? What is meant by impact? And, what is meant by e-government portal capability? City size can be measured in terms of population, size of budget, size of the IT budget, size of the IT budget relative to the total general operating fund and land area, and in other ways as well. Impact could be the relationship between two variables or one variable could be used to predict another. For example, do cities spend larger percentages of their budgets on IT to adopt more portal transaction capabilities? E-government portal capability could mean nonfinancial transactions, orfinancial transactions. So the question could be revised to make it more specific in the following manner:
Q5: Does city size impact adoption of the following e-government transactions?
One way to address this issue of e-government transactions in more detail might be to use subquestions, one per service, such as
Q6: Does city size impact adoption of permit application and renewal transactions through a Web portal?
Rather than address a series of similar questions, a single question could be asked if there were a single measure of transaction portal capabilities. One way to do this would involve data reduction and scale construction. This would start with survey data such as those collected by the ICMA in which respondents indicated all the different kinds of nonfinancial andfinancial services provided through their Web portal. Data reduction techniques could be used to create a scaled measure of transaction portal capabilities in local e-governments. A question like the following one could then be asked:
Q7: Does city size (percent of budget spent on IT) impact adoption of the e-government transaction portal capabilities (TPC scale)?
To be fair to the authors, Don Norris and Jae Moon did not ask these questions in their article because it would not have made sense to do so. The data distributions they report tell the reader that there were too few cities offering interactive transactions through their portals to support analysis of questions like these. Perhaps, the International City and County Manager’s Association’s latest e-government survey data collected in 2004 which became available for purchase in 2006 will support this type of analysis.
Every research question may contain subquestions. Leedy and Ormrod (2004) describe them as follows: Each subquestion should be a completely researchable unit. Each subquestion must be clearly tied to the interpretation of the data. The subquestions must add up to the totality of the problem. And, subquestions should be small in number. They suggest identifying subquestions by going through the following paper and pencil steps:
1. Copy the problem onto a clean sheet of paper, leaving considerable space between the lines.
2. Read the problem critically to discover the areas that should receive in-depth treatment before the problem can be resolved.
3. Make sure every subquestion contains a word that indicates the necessity to interpret the data within that particular subquestions (e.g., analyze, discover, compare). Underline that word.
4. Arrange the entire problem . . . into a skeletal plan that shows the research structure of the problem. You now have a structure of the whole research design (Leedy and Ormrod, 2004, p. 58).
The reader may have reacted to the words‘‘paper and pencil’’much like the author by thinking
‘‘I can do this paper and pencil process better with a word processor.’’So be it. Alternatively, they point out that one could use brainstorming software (Leedy and Ormrod, 2001).
More research questions and subquestions may be developed than can be examined in a single study. This problem can be resolved by prioritizing them to determine which ones to examine in a single study (McNabb, 2002, pp. 67–71).