Measurement day

Objective: Explore concepts of measurement error and data collection

• bias: any non-random error associated with a measurement
• biological error: individuals differ for some measured trait
• censored data: value of an observation is only partially known, e.g., true value is beyond instrument measuring limits
• error: differences between observed and true value
• interobserver error: reproducibility of repeated measures of same object, different observers
• intraobserver error: reproducibility of repeat measures of same object, same observer
• missing value: observed value is unknown
• truncated data: values beyond a boundary limit are excluded, either when measured or when analyzed.

refer to chapter 3.4, Mike’s Biostatistics Book

To do

For each of the projects, briefly discuss, agree upon measures to make and how to make them, and record these decisions to the meta sheet. Review together the concepts listed above and how they are — or are not — relevant to the data collection planned for the day. Include as part of your design ways to estimate the different sources of error. Discuss possible sources of bias with each measure.

Setup and utilize a spreadsheet to record data. Setup new worksheets for each project. Include a meta sheet —  also known as a data dictionary or codebook — to include standard kinds of information needed to recreate the work you do. See Table 1 and Table 2 for examples that follow the row and column format. For more help with proper data entry and data management in spreadsheets see Spreadsheets and data entry.

Make sure that your data are submitted to DrD via CANVAS. Your results will be collected and returned to the class for use in Homework 2A: Measurement Day results.

Hint 1: Think about how to assess inter- and intra-observer error (hint: one trial or many trials?)

Hint 2: Think in terms of “threes”. For example, if you wish to compare performance of a ruler vs the calipers for measuring shell length — at a minimum — record three independent measures of the shell with the same ruler, and another three measures with the calipers. Repeat for as many samples that you have.

Hint 3: Think about steps you can take in obtaining measures to minimize sources of bias.

In-person, eight projects (remote = three projects)

Remote students: Do projects

1 (online dart game)
2 (shell images)
6 (internet bandwidth speed test).

1. Dartboard: How to assess accuracy? Precision?

Figure 1. Results of toss of three darts

In-person

General: each student should toss one set of darts (3 darts). Each student should toss darts.

For tossed darts, record the distance between the center (bulls eye) and the location of the tossed dart. Use results to calculate accuracy and precision, per discussions in Chapter 3 of Mike’s Biostatistics Book.

Take a digital picture of each results — include a unit measure marker of known length with the image (e.g., dartboard diameter). From the image, extract information, e.g., length between dart toss and center of the target, where accuracy is judged as close to the bulls eye (center of target). Although you could measure the distances on your screen (get diameter length on the screen and correct if to actual diameter of the dart board), best to use software designed for this purpose, e.g., imageJ or WebPlotDigitizer.

Note: The boards used in class have a diameter of 11.5 inches (29.21 cm). Additional details provided in class, which should be recorded as part of your meta data.

Remote

General: Instead of tossing darts in-person, try an online dart game: https://www.transum.org/Maths/Game/Darts/ . Each student should “toss” darts, record each trial by screenshot. From the image, extract information, e.g., length between dart toss and center of the target , where accuracy is judged as close to the bulls eye (center of target).

2. Shells, shell length as measure of “size”

In-person

Figure 2. Shells, calipers, and rulers.

I have set out an assortment of shells for you to measure, including

• Shelled gastropods (snails)
  • Giant African snails (Achatina fulica)
  • Conus species
• Echinoderms
  • Sea stars (Protoreaster nodosus and Linckia laevigata)
• Bivalves
  • Scallops (Pectinidae)
• Urchins
  • Sand dollars (Dendraster excentricus)
    • Question: What’s causing that rattling sound when you shake (gently, please!) a sand dollar? Ask Google and find out 😉

Rulers and calipers are provided.

Take images and extract information, e.g., length (longest dimension) and width (widest dimension) of shell.

Suggested worksheet organization

* Don’t assume my measurements are “correct.” Do your own work!
** Units should be reported in metadata, not as part of variable name.

Remote

In place of physical access to the shells, get length (longest dimension) and width (widest dimension) of shell from images. Figure 3 contains 17 Conus spp shells; Figure 4 has 18 Dendraster excentricus shells.

Figure 3. Seventeen Conus shells. Unit measure 1 cm. Image by M. Dohm

Figure 4. Eighteen Dendraster excentricus shells. Unit measure 1 cm. Image by M. Dohm.

Bonus

In-person activity. Record length and width of object with a ruler or caliper. Next, take image of same object, include a unit measure marker, and get length and width from the image with imageJ or WebPlotDigitizer.

3. Measure your lung vital capacity

Covid-19 update: We may skip this measure. Assessing proper lung function is an important clinical measure. A spirometer is the device of choice, but we can approximate the device reasonably well with a simple balloon and an appreciation for how to apply experimental design principles to overcome limitations of an imprecise instrument (e.g., a balloon!). Grab a deflated balloon, take a deep breath, then expel all of your air into the balloon. (Improved function may be had by first inflating the balloon before a breath trial to improve the elasticity of the balloon.) The size of the inflated balloon is a measure of your lung forced vital capacity, FVC. (A related measure, FEV1, is measured the same way, but expel breath for one second only. The ratio is a key clinical indicator of lung function).

Figure 5. Filled up balloon.

Tie off the balloon and retrieve a nylon cord. Wrap the nylon cord around the middle of the balloon and note where the end of the cord meets the wrapped cord. Holding the place with your finger transfer the cord to a tape measure to record the circumference of the inflated balloon. Share with the class your age, height, gender, and average FVC.

Example in R

Copy/paste code to your Script Window in R Commander.

#Dr Dohm's values, inches, for three balloons (inches). Equations require liters and centimeters
balloon <- c(22, 23, 23.5)
radius <- (balloon*2.54)/(pi*2)
#volume of a sphere
FVC <- ((pi*radius^3)*4/3)/1000;FVC

#Predicted FVC
Age <- 58
Height <- 68*2.54
FVCpred <- (27.63 - (0.112*Age)*Height)/1000

#Print the results
cat("FVC estimates: ", FVC, "\n")
cat("Predicted FVC: ", FVCpred)

Optional

Try a digital flow meter to record PEF and FEV1, Compare to balloon measurements.

4. Predict core body temperature from measures of skin temperature

Core body temperature is an important clinical measure. The best way to get core body temperature is with use of oral or rectal thermometers, but obviously, these are less than convenient. Under typical settings we would expect that skin temperature would be proportional to core temperature. Assume core body temperature is 37 °C. The purpose of this measure then is to acquire temperature readings from skin temperatures and construct equations to predict core body temperature (see Lenhardt and Sessler 2007). We have three infra red thermometer guns; fresh 9 volt batteries will be made available.

Figure 6. IR handheld temperature device.

R Example

Used three devices, took three measures of back of the throat and three measures of skin temperature from the forehead.

Copy/paste code to your Script Window in R Commander.

device <- c("1", "1", "1", "2", "2", "2", "3", "3", "3")
throat <- c(95, 95, 95.2, 97.5, 98.4, 98.6, 97.4, 96.5, 96)
forehead <- c(85, 85.1, 84.5, 86.7, 86, 86.3, 85.5, 85.2, 85.3)
tmp <- data.frame(device, throat, forehead)
scatterplot(throat~forehead | device, regLine=TRUE, smooth=FALSE, boxplots=FALSE, by.groups=TRUE, data=tmp)
abline(lm(throat ~ forehead))

5. Accuracy and precision of volumetric pipettor

• Open a spreadsheet file.
  • Create a meta data sheet, which would include a description of what you are doing. Include also room temperature and barometric pressure.
  • The second worksheet would be for data recording
• Grab distilled water
• Place a weighing boat onto the balance and tare.
• Select the first pipettor and pick an appropriate volume. Pipet that volume of distilled water into the pipettor, then dispense the amount into the weighing boat.
• Record the volume and the corresponding weight into the cell of the worksheet and also into your notebook.
• For each volume, repeat for a total of four readings.
  • Use the same volume, change tips between readings to avoid carryover.
• Import the spreadsheet into R, then calculate appropriate statistics (mean, corrected mean, standard deviation, coefficient of variation, and percent accuracy).

6. Speed test internet

Get repeated readings of your Wi-Fi bandwidth access speed. There are many sites to check your internet latency (milliseconds), download (megabytes per second), and upload (megabytes per second) speeds; we’ll use “run an NDT test” at Measurement Lab (NDT stands for network diagnostic test. as opposed to other industries, nondestructive test). Internet speed can vary for a number of reasons, including time of day, number of users on a single connection, location, to name a few.

Here are a few observations I made on my office desktop, connected to campus Wi-Fi or via hotspot on my cell phone.

Table 2. Bandwidth speeds by Wi-Fi type and time of day

Try this yourself. Think about how you would gather observations to separate factors that affect internet speed. For example, location vs. time of day.

7. Constant volume (bomb) calorimetry

Will provide students with an aluminum calorimeter (Eisco) to measure amount of energy released through burning food. Dried bananas and peanuts provided.

Figure 7. Simple food calorimeter. Not shown, thermometer or thermocouple to monitor temperature change.

The principle is straightforward. Burn a known quantity of food onto a nail in the placed cork directly under flask with known volume of water, record the change in temperature of the water.

where Q is the amount of heat gained by a sample (e.g., calories), m is the mass of water, c is the specific heat of water (1 calorie/g), and is the temperature change. To obtain the calories per unit mass of a food stuff, divide Q by the change in mass of the food item after burning.

An example calculation

125 ml of water (= 125 g) placed into the flask. Initial water temperature, , was 21.2 °C. Initial mass of food sample (+ cork + nail), , was 10.0 g. After the food item completed burn, end water temperature, , was 39.8 °C and end mass of food sample (+ cork + nail), , was 9.7 g. Thus, .

With the formula we have , which should be converted to kilocalories (common units for food items), or . And finally,

8. Skittle count

One mini bag per student, tally the number of beans of each color

Green

Orange

Purple

Red

Yellow

References

Lenhardt, R., & Sessler, D. I. (2006). Estimation of mean body temperature from mean skin and core temperature. The Journal of the American Society of Anesthesiologists, 105(6), 1117-1121.