This is a similar veined question / discussion to one I raised a few months back, then regarding postgraduate teaching. I am starting to teach a course on "Numerical Ecology" this semester, to 3rd year biology students, and I wonder what other folks think are the topics such a course should cover. More generally, what are the quantitative topics that students from an undergraduate general course in Biology / Ecology should be exposed to before they finish their degree. I would also appreciate links to other similar courses in Biology undergraduate courses (say to "Biostatistics", "Statistics for biologyists", "Numerical Ecology", "Quantitative methods" etc.
Excellent question! I believe I earned a great deal of experience in conducting science and writing manuscripts throughout my undergraduate degree but felt one of the largest aspects pertained to data management and analyses. As a professional in the field, I find having sound skills in database development, curation, and querying are essential at all levels of science. Having a good understanding of data structure provides an excellent foundation for future lessons in analyses and graphing.
Depends on the discipline. In a survey you have asked 1200 people five Likert-type questions each might need different methods relative to a faunal survey where you set out five light traps in each of eight biomes within the Ohio River watershed.
Organizational skills related to keeping data. Five years from now can you look at the data and recreate the analysis?
How to enter data. Does it make any difference if replicates are in rows or columns?
Does consistency matter? I enter the date as 3-17-18 or sometimes 17-Mar-2018, or sometimes March 17, 2018. It is all the same, right? This is a simple case, but sometimes data has notes, or the treatment could be A or a. We both know that this is the letter "a" but does the computer?
Types of data: continuous, categorical, fixed, random
Approaches to collecting data: trapping methods, transect sampling, observational, treatments.
Sample size: what factors are important? Is a sample size of five ever a good idea? Sampling bias.
Cleaning data: Find the errors, or at least as many as you can. If the temperature at 8:05 yesterday evening was 23.4 C and half an hour later it was 233 C, there might be a problem with the data. This might be a missing decimal point, or some other issue.
Using simple descriptive statistics: the minimum and maximum functions might highlight bad values.
Plot the data, graphing methods.
Missing data: imputation is common in survey methods.
Data Ethics: When can I delete a data point that I don't like? What is the difference between cleaning data versus rigging the results? If I can just get this variable to be significant I am sure my boss will be happier and give me a raise.
From this point on, the methods get more diverse. Comparing means, and multiple comparisons are often used. Analysis of variance (ANOVA, MANOVA, ANCOVA). Assumptions of the models. Regression, and multivariate methods like PCA, factor analysis, discriminant analysis, cluster analysis.
Cluster analysis in an introductory course? It is probably not a good idea to assume that all high school graduates have had a course in linear algebra. However, one starts by recognizing that this technique exists and it can help answer specific biological questions.
Basically, if I do not know that a method exists for answering "this" type of question, then I will not ask questions in that way. The answers that we get from our experiments are restricted by the methods we have available. If the biological results are (in part) a consequence of the methods we use to ask the questions, then science will stagnate when all the manuscripts only have multiple comparison procedures.
You might at least introduce Bayesian analysis.
My personal favorites are computer intensive methods: randomization tests, permutation tests, bootstrap, Jackknife.
This is a fun topic! How about getting your students to design their own research questions (maybe in teams of 2-3 or individually) using data available on the web. For example, I teach a course in which I have my students look through the data sets available on one of the US Long Term Ecological Research (LTER) Network sites (specifically, I use the Palmer Antarctica LTER because I teach Polar Oceanography, but there are many others - terrestrial, fresh water and marine - that you could choose from) and come up with a research question they could answer using the data. Then I have them go through the steps of answering those research questions by analyzing and interpreting the data. They learn basic statistics, data visualization, data interpretation, etc. Their final term product is a paper reporting their study.
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology