Statistics is a way of interpreting the data and applying it to the population being studied. Unfortunately, if the data is collected improperly, the results may be misleading.

In order for the data to represent the population at large, a number of steps must take place:

Step 1: Hypothesis formulation.

There must be a well formed hypothesis which answers a specific question. For example, you might want to know whether older people have slower reflexes than younger people. The question in this case will ask: Is the reflex time for older people slower than for younger people?

Although this seems like an easy task, it often turns out to be much more complex. Without a well defined question, collecting proper information to answer the hypothesis becomes nearly impossible.

Once the hypothesis is formed, the outcome measures must be formulated. In the above example, we need to collect each participant's age as well as finding a way of measuring their reflex time. Other variables might also be important; for example, gender and any physical ailments might play a part in an individual's performance, and therefore must be accounted for in the study. If these measures are not collected at the time of data collection, there would be no way of knowing whether its the age or some other factor which is causing slower reflexes in the older population.

Thus, a researcher with a solid hypothesis statement and well defined and thorough measures will have a better chance of  getting accurate results.

Step 2: Feasibility.

The study must be feasible. First, there must be enough money and time to perform the task necessary to collect the data. Second, all variables that need to be measured must be attainable and not be vague in their interpretation. In the example above, if the patients refuse to tell you anything about their health and you have no way of obtaining this information, then the study is missing a very important component and might not be feasible.

Once the study is planned out, every question asked and measure taken must be carefully considered before the start of data collection. Will the questions be understood by the participants? Will the data collector be able to assign values without  putting his/her own interpretation into the result? Will physiological and psychological measures stay true to their value if the participant knows that he/she is involved in a study? Is the main question of interest attainable? Are there any questions that might produce vague results? Is there enough money to complete the study? Is there enough time to collect the data without jeopardizing the results?

Step 3: Sample Selection.

The selection of participants (observations) is also very important. When the data is collected without following the protocol for proper selection of observations, the results become less stable in their representation. Thus, strict statistical rules must be followed in selection of the participants.

In summary, without all of these steps, your study will not be complete and will not be representative of the population that is being studied. For best results, make sure to consult with a statistician before proceeding with data collection. 

Contact Statistical Consulting Network for help with Study Design