Elements of a Science Investigation


Figure 1, below, is a diagram showing some of the rhetorical components of a science paper. These components, and how they can be used to improve your understanding of science writing, will be discussed. Please carefully examine the blocks in figure 1 and notice what is required for a good scientific investigation. Notice that there is a strong connection between a “theory” or “model” and that your data must be used to show that your model is consistent with the data. Sometimes more than one model is supported and you then need to find data that will distinguish between the two.



Figure 1. Elements of a science paper. The elements in this diagram show critical elements of a science paper, and can be used to more precisely evaluate student writing.




I like to make the analogy between a scientific investigation and a courtroom trial. When a person is being accused of a crime, evidence is marshaled to support the assertion that the defendant has actually committed the crime. The prosecutor will find witnesses who will testify that the accused was at the crime scene, that he/she took the cash, pointed the gun, etc. There may be video from a security camera. The video would be shown to the jury, in support of the assertion. The defendant might call witnesses that will testify that he/she was somewhere else at the time of the crime. Perhaps there is doubt about the identity because the perpetrator was wearing a mask, or the image was blurry, etc. The point here is that evidence is brought forth to support a theory. The evidence is described, the validity (or accuracy) of the evidence is described, alternative interpretations are addressed, and the way the evidence supports the theory is described. These are all elements that must be addressed in a scientific investigation too.



Scientific evidence: interpretation or observation?

For the purposes of your scientific writing assignments, observations are derived from plots of earth data. It is important to distinguish between simple cartoon-like figures and figures that have been generated from data. Figure 2 shows a figure generated from actual temperature and wind data. Figure 3 is a cartoon that illustrates the basic features of the data, but has been synthesized and put into an illustration. Images captured from the MAP screen of “Our Dynamic Planet” can be considered as data representations and used in your observations.


It may not always be easy to determine if a figure is an observations or a cartoon. Check with your instructor or TA if you have doubts.

Figure 2. A graphic image of El Nino sea surface temperature (SST) and winds, generated from sattelite and buoy data.

Figure 3. Cartoon showing El Nino conditions, illustrated by a graphic illustrator.


How do scientists tell observations from interpretations?

Figure 4 shows how scientific knowledge can be classified, and some indications of how observations and interpretations can be distinguished. Knowledge can be classified on a scale according to how well it is accepted. Even within a single scientific context, there is a shifting of categories as new discoveries are made.


Can you think of any knowledge that you once considered “fact,” yet reconsidered after gaining new information?


Figure 4 Science knowledge structure and the difference between observations and interpretations.


The Rhetoric of a Science Paper: How to put together a scientific argument

Figure 1 suggests important elements of science writing. The effectiveness of an argument can be measured by classifying sentences into a relatively small number of sentence types. A good scientific argument contains each of the kinds of sentences that are listed below. Note that ALL of the types of sentences must be used for a complete scientific argument.


A good scientific argument contains sentences that:

1.       include an observation, or description of an observation.

2.       name or classify an observation in terms of geological features.

3.       describe a feature that has been classified.

4.       describe relationships between classified features.

5.       describe a model or theory and/or a relationship between model features.

6.       describe relationships between features or data and a theoretical model.


1. Observation: Observations are at the top of Figure 1, and have already been discussed in detail. They form the foundation building blocks of the scientific argument. If observations have been made by others, references must be made to publications that describe the experimental methods and errors in the observations. In the context of the “Our Dynamic Planet” CD, examples of observations are the maps of quakes, elevation profiles, quake profiles, and the map with data plotted on it. The learner should be aware of the accuracy of the data, and would report it in the “methods” section of the paper. An example the importance of understanding the data is contained in earthquake data. Quake depths for regions in the middle of ocean basins, far from land, often show concentrations at certain depths, making it appear that horizontal faults are occurring. However, this is an artifact of the earthquake location software, which puts quakes at a specific location when the arrival time data (used to locate quakes) are not complete enough to calculate an accurate depth.


2. Classification: Based on the observations, classifications are developed. For example, a series of elevation profiles across a particular geological feature will result in its classification as a trench, ridge, mountain, or other topographical feature. In short, the data have been used to name or classify a feature according to accepted terms. In some cases, the separation between observations and classifications is somewhat arbitrary. A geologist would argue strongly that he/she has directly observed a mountain range, rather than a linear shape of darker brown or purple color on the horizon. However, in the context of the plate tectonics investigations, the use of several profiles to infer a mountain range is called a “classification.” This classification could also be called a “feature.” A feature is somewhat more general and could refer equally well to a mountain range, a linear valley, or an arcuate pattern of earthquake epicenters. The “Profile Game” tool of the CD helps you acquire the skill of using profiles to make these classifications. Please note that a single profile is NOT sufficient to identify a topographic feature.


3. Describing a feature: Data have been acquired and a “feature” has been identified. It may be long and linear, dipping into the crust, or it may be localized in some way. The feature’s depth, length, and trend should be measured and reported. It is important to describe features quantitatively, rather than in vague terms such as “large,” “small,” “long,” etc.


4. Patterns and relationships: The next level of scientific argumentation involves statements showing relationships between features. An example would be noting that a topographic trench on the seafloor often has volcanoes parallel to it. The search for the understanding of patterns and relationships provides many puzzles that a theory or model must explain. For example, the investigator might look at many trench-like structures and determine whether a parallel row of volcanoes exist in each instance. Most often, the simple pattern will have exceptions and variations. Why is there an exception? What determines the distance between the line of volcanoes and the trench? Maybe the pattern of earthquakes can help solve the puzzle. Are patterns of quakes different in regions where a parallel line of volcanoes is absent? The search for patterns, with the associated process of comparing and contrasting different regions or regimes, is a powerful method for testing theories in most sciences.


5. Describing a model or theory: Ultimately, the goal of research is to find (or test) a model or theory that explains all of the observations. This category includes all sentences that describe a model and describe relationships between “features” of the model. For example, a model of a subduction zone might discuss the topographic trench, volcanoes parallel to the trench, and a descending pattern of quakes as the tectonic plate descends into the Earth. A cross section could show how melting at the upper surface of the subducting slab creates molten rock that surfaces as volcanoes, and how the descending pattern of earthquakes is predicted by relative motion between the moving slab and more static plate above it. A clear description of the model makes it easier to point out how the data support or disagree with the model.


6. Relationships between features and a model: In a well-crafted investigation, this section can be the most satisfying one. It is at the highest rhetorical level and can demonstrate student understanding of the investigation. Here the author discusses the relationship between the data and the model they have created. Diagrams and illustrations are used. For example, the model may have the locations of the observed trench and volcanoes explicitly drawn on it (to scale, of course), which explicitly shows the important relationships. The important point is that the model must match with the data. The drawings should clearly show this correspondence. One common mistake that students make is drawing a subduction zone showing the descending slab plunging to the right, while the earthquake data clearly show it descending to the left. Clearly, this student has not thought about how the data and model must agree.


Related research and findings: Here the results of previous research by other investigators are discussed. The author might point out where the new observations expand or provide further support for a particular theory. Other investigators' work must be referenced. The format for these references is discussed later.


The iterative nature of a science investigation: Learners often have the idea that they will take data, interpret it, then write it up and they are done. In science practice, the act of interpreting data most often leads to more questions that can be investigated. A good investigation will include serious application of thought, reflection, and refinement of the experiment.


Summary: Several kinds of argumentation statements have been discussed. Support for a theory flows from a recognition of the implied patterns and relationships of features. The features are identified and classified using the observations. A theory or model must then be introduced and statements that describe how the data agrees with the theory. For example, the theory may predict that earthquakes will be in a certain location. The writer must explicitly explain how the earthquakes shown in the data plot are where the theory predicts them to be. The argument is strengthened by comparing multiple kinds of data with the predictions of the theory. Elevations, volcanoes, earthquakes, and seafloor ages can all be used to support plate tectonics theory.