Expectations for lab reports

Data analysis expectations

In real research, we start writing our papers after we have finished the bulk of our analysis. That’s because we don’t know what our paper should say until we have thoroughly studied our results and found useful data analysis interpretations. There is not only one way you should analyze your data – it may be context dependent – but here is a basic structure that will work in most cases:

  • Define your model Consider what you measured and what your hypothesis is for the results of that measurement. This should lead you to a mathematical model to compare to your expected results. For example, if I am studying the speed of a block sliding down a ramp, then I expect the final speed to be related to the release height by \(v = \sqrt{2gh}\). That’s my model and that is what I will compare my data to.
  • Identify parameters What are the parameters (things that are fixed but you maybe don’t know perfectly) and what are your variables (things you vary, things you measure). In the previous example with the sliding block, my variables are height (what I varied) and final speed (what I measured). Those together are just my data points. The parameters are less obvious, but I should consider what kinds of deviation from the model I might see. In this case, I might see an offset so I would add that as a parameter. There might also be some scaling problem with the height (what if my ruler was off by 5%?) so I could add that as a multiplying parameter. My new model would be \(v^2 = a\cdot(2g)\cdot h + b\).
  • Fit model to data Use some kind of fitting routine (on the computer) to fit the model to your data and find values for your new parameters. This way we can quantify how far off our data is from the ideal case where \(a=1\) and \(b=0\).
  • Compare with uncertainty Hopefully you took some uncertainty measurements with your data and used a fitting method that incorporates uncertainty. If you did, you can check the uncertainty of your parameters and see whether they are significantly different from the ideal case. If you didn’t, you can still report the best fit metric (ex: reduced chi-2 or least squares) and comment on whether the model is a “good” fit for your data.

This would be a very complete treatment of your data analysis, and it’s what would be expected of you in a real research context.

Writing

Lab reports should be written to a similar length and quality as a research paper in your history or literature classes. In fact, they share many similarities: in a literature course you would write an argumentative essay supporting some thesis about a work of writing, and the bulk of your essay would consist of detailed supporting evidence for your thesis. A lab report is exactly the same, but your supporting evidence is your measurement. Make sure you are writing in full, correct sentences that connect to each other smoothly. Style matters!

You should write for an audience that is somewhat familiar with the subject but was not present when you performed your experiment. You will need to carefully explain your goals and motivation, your process, and your conclusions.

I have included descriptions of each part of the lab report below, and indicated the percentage points.

Abstract (10 pts)

Abstracts are a concise paragraph that summarize your paper. They tell your audience what they should expect to see (basic methods and results) and why they should bother to read it (motivation, novelty, etc.). You should report any numerical values that you have found and state your conclusion clearly (AFTER describing the experiment and its motivation).

Introduction and Background (15 pts)

This is where you introduce the topic, why you are studying it, and any necessary theoretical background. Assume the reader is somewhat unfamiliar with the topic. Remind them of important background and context. Briefly derive any important equations (an example of an important equation is your model!). If you do not include the critical equations needed to understand your later analysis you will not get credit for the introduction.

Methods (15 pts)

The goal here is to explain what you measured and how you analyzed your data.

Explain your experiment in enough detail for someone to replicate it. Describe your measurement system carefully but do not include irrelevant details. What is an irrelevant detail depends on the experiment, but ask yourself if you need to know it to do something equivalent. Describe the range of inputs to your system and their source (in our case, usually a signal generator, a type of waveform, and some range of frequencies). You should include what values you measured and how you measured them (basic description of equipment and its use).

Describe any data analysis methods used (linearization, fitting technique, error analysis, etc.)

Results and Analysis (40 pts)

The bulk of your score will be for completing the data analysis described earlier.

Include your results, typically as a plot of data with a fitted model. Plots should be big enough to see clearly but not so large that they take up a full page. Plots should be thoroughly labeled (axes, legends, etc). Plots should be efficient and directly compare data and models as appropriate. Captions should point out any important features in the plot and explain the contents.

Carefully describe what you see in the plot and point out any important features of the data. Comment on the fit of the model and report the parameter values and their significance. Interpret those parameter values as physical characteristics of the system. Suggest any changes that would improve the model. Comment on any experimental difficulties and their effect on your results.

Conclusion (20 pts)

Briefly recap the previous sections, restating the purpose of the experiment and the theoretical background and context. Make a definitive conclusion about whether your data supports your model, and support it with evidence from the results and analysis section. The conclusion you reach should be clearly supported by your data analysis, and you must clearly state this connection. Brief does not mean incomplete. Someone should be able to read the conclusion and basically understand your entire experiment and its results.

I will deduct points from the total score for poor quality writing and style.