2.3 Topic Review: Process of Science
Types of Scientific Inquiry
What is a Scientific Study?
A scientific study always follows a rigorous, methodical process, but it doesn't always follow the common perception of a scientist in a lab coat testing a hypothesis
What scientific question is driving this inquiry?
All scientific studies are looking to answer a question about the natural world
Which type of scientific inquiry is the best for this question?
Questions are asking about what exists, or how to describe something new often drive observational descriptive studies.
If there is very little known about something, there needs to be a rigorous description of it to enable developing hypotheses to test.
Questions that focus on how something works, or which factors affect it are usually addressed in hypothesis-driven experiments.
A hypothesis proposes an explanation of an observation, and an experiment then rigorously tests whether evidence collected during testing supports this explanation.
Questions that focus on how to use mathematics & computational algorithms to describe a phenomenon involve developing models or theory to build simulations.
These types of scientific inquiry use the evidence collected in other scientific studies to build these models. These types of studies involve significant elements of both biology & computer programming.
Hypothesis-Driven Experiments include:
treatments, replicates, and controls that help focus data collection on specific variables
Keys to a Scientific Question
A scientific question seeks to explain something about the natural world.
A scientific question meets three criteria:
It is rational.
Scientific questions are logical and relate to the reality in which we live.
Scientific questions do not involve any deities or supernatural beings.
It is testable; it can be answered by collecting evidence.
It is repeatable; if different people or groups collected evidence related to the same question, the results would be consistent.
Scientific Method
Make an observation.
Ask a question.
Form a hypothesis that answers the question.
Make a prediction based on the hypothesis.
Do an experiment to test the prediction.
Analyze the results
If the hypothesis is supported, report the results.
If the hypothesis is NOT supported, report the results & try again (form another hypothesis).
Planning & Methods
All scientific inquiries begin with a scientific question.
Each of the 3 types of inquiry involves different types of planning, and requires different approach to their questions in different ways.
Both observational descriptive studies & development of models & theory deviate from flow charts of the scientific method that are usually presented.
These types of studies are strongly focused on the scientific questions they are addressing.
The specific details will depend on the question.
Hypothesis-driven Experiments Design
Decide how data will be collected:
Will there be data collected at multiple points in time? Continuously throughout the experiment? Is the data descriptive, such as color, or numerical, such as height or temperature?
What treatment is being applied, and how is it being applied?
The treatment explains how the independent variable is going to be manipulated.
For the fruit fly experiment, for example, will temperature be changed using water baths? temperature-controlled rooms? heat lamps?
How many replicates are needed?
If you only try something once, you don’t have much confidence that you’ll get the same results next time. Replication included multiple copies of the same experimental conditions.
If all replicates with the same conditions produce similar test results, then you have increased confidence that your experiment has produced consistent results.
There are statistical results to evaluate the similarity between results, as well as plan for how many replicates would be ideally needed to evaluate the hypothesis
Logistical concerns, such as space, time, or cost, often affect how many replicates will be included.
Identify the Variables
Independent variables - are the things that the experimenter plans to change. A good experiment is designed to allow ONE variable at a time.
For example, if you were performing an experiment on the effect of environmental conditions on fruit fly hatching success, you wouldn’t want to change temperature AND the food source fed to fly larvae. If your results differed, was this difference due to the temperature or the food?
When they want to test the effects of multiple independent variables, scientists use multiple trials. For the fruit fly experiment, you could have a trial that changed just temperature, another trial that changed just food source, and a third trial that changed both to see if these two variables interact with each other.
Dependent variables - are the things that the experimenter plans to measure. An experiment may collect data on moe than one dependent variable.
In the fruit fly experiment, for example, you might collect the percentage of eggs that hatched, as well as how long it took eggs to hatch.
Determine Appropriate Controls
In any hypothesis-driven experiment, data are needed to evaluate whether observed differences are due to chance, or due to the changes made to the independent variable.
A control provides this “baseline” data.
The design of the experiment determines the controls that are necessary.
Controls setup a trial where no independent variables are changed.
If an experiment is evaluating interactions between independent variables, multiple controls will be usually be needed.
In the fly experiment, for example, having replicates at room temperature might be an appropriate control, whereas two controls would be needed if evaluating the effects of both temperatures & food source.
Data Analysis
Before collecting data, good scientific practice is to decide how these data will be analyzed. Often, different types of analysis require collecting different types or amounts of data, so making these decisions as part of the study design ensures that the correct data will be collected. These decisions are important for all types of scientific inquiry.
For example, a “bio blitz” is a type of observational descriptive study that seeks to identify ALL organisms that are found in a particular geographic area. In these types of studies, including information about when & where an organism is found is critical.
There are also statistical methods that will tell you how large an effect size (difference between treatments) may be measured using different amounts of data. Practical considerations nearly always limits the amount of data that can be collected, so understanding the relationship between amount of data & effect size provides some guidance when designing a hypothesis-driven experiment.
Results Presentation
Communicating results of a study to a broader audience is critical part of all types of scientific inquiry. Data presentation is where the rigorous process oof science meets the flexible creative process of design.
Things that need to be considered in presenting the conclusions of a data:
Will a graph or a table be best for illustrating the results?
Graphics are eye-catching, but sometimes mask important details.
Tables are good for reviewing details, but often difficult to see patterns when looking only at a table.
The most effective is a mix of graphics & tables. Tables should represent the important details that a graphic might mask, while graphics are better at illustrating patterns or trends in the data.
What message do the overall patterns in the data show? Is there a trend? Do values between the measured data points have any meaning?
If using a graphic, which type?
Maps, bar charts, line graphs, pie charts, timelines, two-dimensional coordinate plots… The options for graphics are numerous, but they each have a particular type of data or message they convey best.
The type of data is the first thing to consider.
Does the data consist of discrete values, such as counts, or does the data vary continuously, such as temperature?
Displaying discrete data values on a line graph, where each data point is connected, doesn’t make sense, but the pattern is hidden if a bar graph is used to show continuously varying data.
When graphing variables on two aces, independent variables are always plotted on the x-axis (horizontal), and dependent variables on the y-axis (vertical)
Each type of graphic also involves other choices, such as various labels, scale of aces, colors, patterns, etc.
When looking at graphs, always make sure to look for the scales being used.
Does the y-axis scale start from zero, or some other point?
What units are used?
If multiple graphs are combined (such as a bar with a line graph), how do the scales compare?
There is no one “correct” answer to any of these questions, since they depend on the data & the message that is being conveyed. You should always look to see that the choices used make sense for the type of graphic & data being presented.