In order to write a Research Paper on a particular and specific question, the IMRaD format is often the preferred choice. The acronym IMRaD stands for Introduction, Methods, Results, and Discussion. The fundamental objective of adopting the IMRaD format in Research work is to present the facts substantiated with collected figures and data and how they represent and relate to the Hypothesis or Research Question through Results. Therefore the Results section plays a pivotal role as it determines how the readers are to be drawn towards the conclusion through the best possible presentation of tables, graphs, charts, photographs and related texts in a logical manner. Once it is done properly, then only a discussion could be written analyzing such data and the relevance of their findings to the Research Question.
The result section is the third major part of the research paper and it’s probably the most important part because it contains actual facts about your experiment. The other sections contain a plan, hope and interpretations but the result section is the actual truth of your study. In the result section, one should aim to narrate his/her finding without trying to interpret or evaluate them. Basically, the result section explains any problem you have with your data collection, the main result of the experiment and any other interesting trends in the data. Interpretation of the meaning of the result section is done in the discussion section. With the result, we want to convey our data in the most accessible way, so we usually use visual elements like graphs and tables to make it easier to understand.
Essential elements for writing a proper Results Section:
The results section should be strictly used for presenting the key results without any interpretation. The facts, figures, and findings are to be presented in a logical manner leading to the hypothesis and following the sequence of the method section. Mention must be made for the negative results as it would substantiate the discussion section later on. Any unusual correlation observed between variables should be noted in the Result Section. But any speculation about the reason for such an unusual correlation should be avoided. Such speculations are the domains of the discussion section.
Resultant data are to be presented either through text, figures, graphs or tables or in a combination of all of the best suited for leading to the hypothesis. Care should be taken to prevent any duplication of the text, figures, graphs, and tables. If any result is presented in figures or graphs, it need not be explained through text. Similarly, any data presented through the graph should not be repeated in the table.
Each table and graph should be clearly labelled and titled. Each different finding should be made in a separate sub-section under the proper sub-heading following the sequence adopted in Method Section.
Comparisons between samples or controls are to be clearly defined by specifically mentioning the common quality and the degree of difference between the comparable samples or controls. Results should always be presented in the past tense.
Tables are labelled at the top and are labelled as Table 1, Table 2 and so on. Every table must have a caption. It’s good if one can put independent variable conditions on the left side vertically, and the things you have measured horizontally so one can easily compare the measurements across the categories. But you need to decide for each table you make, what is easiest to understand, and what fits on the paper. Anything that is not a Table is a Figure. In the case of figures, the captions should come below, called Figure 1, Figure 2 and so on. In graphs, the independent variable is X-axis and the dependent variable is Y-axis.
You can use various types of graphs in your results like a line graph, bar graph, scatter plot, a line graph with colours, a box with whiskers plot and a histogram. Tables are good for showing the exact values or showing a lot of different information in one place. Graphs are good for showing overall trends and are much easier to understand quickly. It also depends on your data. In general, continuous variables like temperature, growth, age, and time can be better displayed in a line graph on a scatter plot or maybe on histograms. If you have comparative data that you would like to represent through a chart then a bar chart would be the best option. This type of chart is one of the more familiar options as it is easy to interpret. These charts are useful for displaying data that is classified into nominal or ordinal categories. In any case, you need to decide which is the best option for each particular example you have, but never put a graph and a table with the same data in your paper.
Now, when it comes to the organization of the result section, start with the paragraph, not a Table or Figure, and make sure you show the Tables and Figures after they are mentioned in the text. Also at the beginning of the result section, you should explain any missing data or problems you had, while collecting the data. Then explain the main results and address your hypothesis. Finally, explain all the other interesting trends in your data.
Some Common Mistakes Observed in the Results section
Let’s look at some of the common mistakes which can be observed in the result section.
i. One should not include raw data which are not directly related to your objectives. Readers wont to be able to interpret your intentions and may unnecessarily collect unwanted data while replicating your experiments.
ii. Don’t just tell the readers to look at the Table and Figure and figure it out by themselves, e.g “The results are shown in the following Tables and Graphs”.
iii. Do not give too much explanation about Figures and Tables.
Example: For brevity let me first show the Abstract of my work on a paper titled “An Optimized Fuzzy Based Short Term Object Motion Prediction for Real-Life Robot Navigation Environment”
Short-term motion prediction for moving objects in a real-life Robotic navigation environment involves uncertainty and temporal validity of the results. Prediction of the accurate position of a moving object and responding with quick action are the main objectives of Robot motion planning. This paper proposes a novel algorithm for short-term motion prediction involving a fuzzy-based predictor. Because of the multi-valued nature of the fuzzy logic, this approach enjoys high robustness in dealing with noisy and uncertain data. The knowledge captured by the Rulebase has been optimized by defining directional space. In the proposed work the predictor has been evaluated with three well-known defuzzification techniques. Based on the analysis of results, it has been found that the Mean Of Maximum defuzzification technique has a low response time and better accuracy. The predictor is tested for various motion patterns in a real-life environment.
Keywords: Short Term Motion Prediction, Fuzzy Rule base, Rule base Optimization, Directional Space, Fuzzy Predictor Algorithm, Defuzzification.
Example: Here is the Results Section of My work:
Object motions with different motion patterns are generated by a simulator in different directions to generate the initial rule base. The rules generated are clustered based on the direction of the motion pattern into the directional space clusters. Table1 shows the number of rules that remained in each directional space after removing inconsistencies and redundancies.
Our predictor algorithm is tested for a real-life benchmark dataset available at: http://homepages.inf.ed.ac.uk/rbf/CAVIAR/. (EC Funded CAVIAR project/IST 2001 37540) to check for relative error. The data set consists of different human motion patterns observed at INRIA Lab at Grenoble, France and Shop Centre. These motion patterns consist of frames captured at 25 frames/second. A typical scenario of the INRIA Lab and the Shop Centre is shown in the Figure below.
We define the relative error (RE) for M sample test data as
Where da is the actual distance, dp is the predicted distance. The Fuzzy predictor was run on 1.66GHz machine in VC++ environment. The figure below represents the average relative error observed for the prediction algorithm for various test cases using Min-Max, MOM and COA defuzzification techniques.
Fig. Average Response time and Relative Error of the predictor at prediction step: 02 seconds
For each test case, the average response time is also calculated to find its suitability for a real-life environment. The prediction algorithm is tested by processing the frame data of moving human patterns stored in the database at intervals of 50 frames (02 Seconds). The navigation environment is presented in the form of a Prediction graph where the x-axis represents the Range parameter and the y-axis represents the Angle parameter. The predicted Angle and Range values are compared with actual values obtained from the real-life environment.
Figure. Prediction graphs showing the some of path prediction solutions with MOM
The performance of the predictor is tested when more than one object is sensed by the sensor. The tests are carried out assuming at most 6-8 objects can be visible and can affect the decisions to be made regarding robot traversal(Figure below).
Figure. Response Time of the Predictor for Multiple Number of Objects