There's a tremendous amount of work that goes into research. Like many other worthwhile pursuits, an onlooker will only see the top 1% while the other 99% rests beneath the surface. Society tends to only see the finished product--not the mountain of hard work we put-in to succeed. This post offers some insights into the level of involvement with quantitative research methods (at least in financial planning).
Research in several steps
The total amount of time a particular project requires largely depends on several factors: the researcher's familiarity level with the data set, the quality of the data relative to the research question, and the number of people involved in the project, to name a few. Each step in the process deserves its own blog post (maybe even its own blog!). Broadly, the steps include deciding what you wish to research, what you wish to test, how you wish to test it, and how you (or others) should proceed in light of your findings. That's the simplest way I can sum-up the research steps. Now, let's briefly consider each step.
What do you want to research?
In technical terms, what is your research question? It's a question you must first ask and then try to solve via the scientific method (at least that's essentially what you're doing. It's a bit more involved than the basic steps you learned in elementary school). Depending on the research question, the very beginning can easily be the most time consuming because you must read "the literature" on your research question. This is usually done using university library resources or online sites such as LexisNexis or even Google Scholar.
Often, researchers will filter searches by keywords, and then read the articles that appear in the search. Those articles will cite other, older articles, which the researcher will then read, and so on until (s)he circles back around and stumbles across the same sources using different keywords. For the most part, this step is non-exhaustive meaning you can pretty much spend an endless amount of time here. At some point, however, you acquire enough knowledge to recap for the reader what we collectively know about the topic in question.
What do you wish to test?
In addition to your research question, you pose at least one hypothesis--often in the form of, "There are significant differences between x and y." You can choose to place a direction on your hypothesis, or simply ask whether there's a significant difference. Here's an example of each one:
Directional: "Americans who attain professional degrees earn more income over their lifetimes than those who do not."
General: "There are significant differences in terms of college GPA between males and females."
How do you plan to test your hypotheses?
Hypotheses are only valid if you can actually test them, meaning you can measure whatever you are studying. There are tons of great research questions out there that will go unanswered simply because they cannot be tested. In my own research, I have found fraud to be extremely difficult to measure because there are inherent downsides (at least perceptible ones) for acknowledging that a person has been defrauded. It's embarrassing, disheartening, lowers self esteem, can incite anger from loved ones, reprisal from the fraudster--all sorts of problems come with this subject matter. Researchers must think through these obstacles in order to test the hypotheses and answer the research question.
The two main steps to hypothesis testing is selecting (or collecting) the data and choosing the analysis(es) necessary to measure the object of your study. There are tons of data sets out there. Thus far in my career, I have used the National Longitudinal Survey of Youth (NLSY), the Survey of Consumer Finances (SCF), the Health and Retirement Study (HRS), the National Incident-Based Reporting System (NIBRS) of the FBI, and the World Development Indicators (WDI) of the World Bank. Once you find a data set that contains the variables you want, you then need to familiarize yourself with that data. There should be a codebook that describes each variable as well as general information about how the data were gathered. You will briefly summarize this information in your write-up. As an aside, the first time you use a set of data, it can take a very long time to reach the level where you can employ statistical tools suitable for measurement. Recently, I probably logged over 100 hours on a research project I just submitted to a conference in October. Of that, easily 2/3 was spent wrestling with the data; The struggle is real!
After selecting the data (but before testing), you need to clean your data. This often takes the form of relabeling variables (usually the variables are titled L8179 or V20021 instead of "Household Income" or "Offender Age"). Additionally, you have to filter-out variables that have nothing to do with your hypotheses as well as recode variables so the computer software will be able to run the analysis. Lastly, in many cases, you must transform certain variables depending on the analyses you select in your methodology. It's very difficult to compare things like income, which has very large numbers, with resident status, which may only have a handful of possible values (0 = non-resident, 1 = citizen, 2 = resident, etc.).
The methodology itself usually comprises summary or descriptive statistics, which gives you general information such as mean, median, and mode of each variable. It's important to take note of any inconsistencies or irregularities in the summary statistics because that may drive your more rigorous testing. For instance, if you notice that the participant's age variable has been divided into age bands (usually in increments of 10), and that there are an overwhelming number of 20-29 values and very few 30-39, 40-49, and 50-59, then depending on your population, you may want to group all the non-20's into a single category and compare that to the 20-29 group.
Now what?
After running your tests, it's customary to check your research methodology to ensure you correctly executed the testing, peruse the code book one last time for any last-minute additions you wish to insert into your research model, and then start to make sense of your findings. Often, at least one result will surprise you! After the methodology section of a research paper comes the discussion, conclusion, and future research parts. Here, you provide some rationale for your findings, whether the findings supported your hypotheses, and possible next steps in adding to the body of knowledge on this given topic. Many researchers take note of this final section when conducting literature reviews because past studies that call for investigation into a particular area--one that happens to interest you--is an excellent way to start the research process. What is your reasoning for studying this research question? Because past studies stressed the need for that question to be answered! When approaching the research question, you can either build on a past study, use a newer wave of the same data set to determine whether the findings are the same, or you can test a theory that was raised in another paper--even a different discipline.
Research: time-consuming, all-consuming?
Doing research properly requires an incredible time investment on your part, time when you are not teaching or performing service for your institution or being with your family. From a faculty career standpoint, the more ambitious your research agenda (and the higher the publication expectations are for tenure), the less teaching and service loads you can bear. I am both surprised and pleased that some universities have started emphasizing teaching vs research vs service aspects in their job advertisements and in the position themselves. Traditionally, a faculty member is presumed to be super human with infinite amount of time to teaching a 4/4 load, publish three or more papers in top quality journals a year, and spend several hours each week building courses and serving on various institutional committees. This is unrealistic and unsustainable. Fortunately, many jobs ads will stress "clinical" or "practice" for heavy-teaching/minimal research positions and vice versa. This is a welcomed change in the academy, and perhaps one day the profession will split completely such that teachers will be teachers, and researchers researchers.
