Using the word what re-enforces this less than concise formulation of the research question. In fact, predicating your question with what or why doesn't allow your question to ask exactly what you need it to ask.
In experimental design, we're testing for relationships - asking "is there a relationship?". In fact, we're asking a question that allows for the proposal of a hypothesis; a prediction of what that relationship might be. So, we should think about how we can ask a question that reflects the test or experiment we're planning.
In addition to re-framing our question to one which is phrased as a testable question, we want to clearly articulate our population of interest and our main variables of interest. When phrased as
What is the impact of fertilizer x on the growth of mung beans?
the variables ostensibly include fertilizer and plant growth. But plant growth is more nuanced than this, and our study might be too. In fact, arguably, plant growth is not a variable, but a composite of variables; so, we should ask ourselves, "what do we mean by growth? What about growth are we interested in? Germination rate, germination survival, biomass, height, flower set, fruit set?"
Defining the scope of your population and variables is a key consideration when developing a research question; defining these early means that you won't be asking these questions later, once you've already started to collect, or work with, your data.
So, ultimately, we want a question that:
- Is testable;
- Clearly identifies our population;
- Clearly identifies our primary variables of interest; and
- Is concise