DATA

To determine the loudest cereal, volume of milk, and type of milk combination, we measured the sound level produced by the cereal in milk over the first 15 seconds and the sound level produced over the first 90 seconds. We completed a total of 45 trial runs: 3 trials per combination of controllable variables (13 combinations) with an additional 2 baseline tests bookending the experiment. A noise meter iPhone application, DecibelX, allowed us to record the sound level produced by the cereal in decibels every 200 milliseconds for up to 3 minutes. The raw data was further analyzed after exported to Microsoft Excel. The data collected in this experiment will indicate the cereal combination with the loudest initial sound level and loudest sustained sound level, respectively.

METHODS

Each trial run was prepared with 1 cup of dry cereal in a plastic bowl (Video 1). We measured out a volume of milk, the amount and type specified by the trial. The measured milk was poured over the cereal in a circular motion to uniformly distribute the liquid in the cereal. Immediately after pouring all of the milk into the cereal, we began to record the decibels produced by the cereal immersed in liquid. We continued recording for 90 seconds to prepare for a comprehensive analysis of the sound level produced over time.

Video 1: Demonstration of the procedure completed for each trial of the experiment.

Figure 9: Flowchart depicting data collection procedure, beginning with measuring the purchased cereal and milk and then uniformly pouring the milk into the bowl of cereal. Immediately after pouring in the milk, the sound level was recorded in decibels over a total of at least 90 seconds.

To begin our data collection process, as outlined in the above Figure 9, we needed to obtain the necessary materials: three family size boxes of each cereal, one gallon of each type of milk, measuring cups, and bowls. We also prepared and measured one cup of cereal in a clear bowl in the first level of our flowchart. In the second level, we prepared the independent variables for each cereal, which was also a controllable variable. For each of the three cereals, we poured a one type of milk with varying volumes in at a time.

Data was collected via the DecibelX noise meter application on an iPhone 6s. Its app logo is shown in Figure 10. DecibelX makes use of the iPhone’s microphone to record sound level every 200 milliseconds in units of decibels. The application can record up to 3 minutes of data at a time, highlighting the average, minimum, maximum, and peak decibels of a space.

In our experiment, we used DecibelX to record sound level produced by the cereal for approximately 90 seconds. We could then export the data for the trial run as a .csv file to import into Microsoft Excel. The imported file contained the average, minimum, maximum, and peak decibels collected as well as the entire log of data collected every 200 milliseconds. Its interface screen can be seen in Figure 11. Using features in Excel, we determined the average decibels over the first 15 seconds after adding milk to the cereal and recorded the results in our data table.

Excel also provided resources to plot results and compare the impact of each controllable variable on the respective response variables through Design of Experiment Means comparisons.

Figure 10: DecibelX application logo.

Data collection took place over the course of about three hours on a single day. We chose a quiet group study room in the Murray Library to conduct our experiment with minimal background noise. To mitigate this uncontrollable variable, we recorded the average sound level within the room. Once the space was set up, we took a baseline test. From there, it took some time to find the most efficient way of assembling the bowls of cereal and measuring the resulting data.

We concluded that filling three bowls of the same cereal, same type of milk, and the same volume of milk for each trial run was quickest. We assembled three bowls of one cup of cereal, Rice Krispies for example, and then measured out a volume of one type of milk, such as a 1/2 cup of 2% cow milk. We then poured a 1/2 cup of 2% milk into the cereal and immediately began recording the sound level produced by the now wet cereal once all of the liquid was dispersed.

Once one trial was run, we moved on to the next trial until 3 runs (a,b,c) were run of each trial number (2,3,4,...,13). Trials 1 and 14 were baseline tests. After each trial number set, we disposed of the cereal, cleaned out the bowls, and reset the experiment with a different variable. This method continued through each trial set over the course of almost three hours.

Figure 11: Data collection in action using the DecibelX app for each type of cereal (Rice Krispies, Cheerios, Cookie Crisp).

Table 3: Data collection tasks assigned to each team member to carry out and maintain an efficient process.

Using the DecibelX noise meter iPhone application, we took sound level measurements in the unit of decibels over a specified amount of time. When exported into Microsoft Excel, we had the ability to manipulate the raw data consisting of decibels recorded every 200 milliseconds. Microsoft Excel allowed us to organize the data by trial run, take the average of the first 15 seconds of each trial, and take the average sound level over the first 90 seconds. These calculations, shown in Equation 1, define our response variables: initial sound level of the cereal in milk and the sustained sound level over time.

Equation 1: Sample calculation of the average sound level of Rice Krispies in ½ cup of 2% cow milk over the first 15 seconds. This calculation was done using the AVERAGE function within Microsoft Excel.

Excel also allowed us to compare response variables of each controllable variable, as seen below in Figures 12 and 13 in the “Spoonful of Data” section.

COST

Table 4: Actual costs of the official experiment performed.

Table 4 portrays the actual costs of the final experiment that was performed. It differs from the original table of costs shown in Table 2 in that the costs above take into account the labor costs of the experiment. Also some more measuring utensils were included as well as only 3 cereals were used. The total cost also more than doubled, which was expected since modifications were made.

A SPOONFUL OF DATA

After collecting our raw data, we were able to perform some initial organization and analysis in a design of experiment (DOE) plot, depicted below in the Design of Experiment plots in Figures 12 and 13. The plot displays the effect of each independent variable by averaging all trials employing that variable. We can clearly see in the plot of Figure 12 that Rice Krispies have the greatest average sound level among the cereals, almond milk produced louder sound than 2%, and ½ cup and ¾ cup of milk were nearly identical in sound level. Figure13 exhibits each factors' sustained average sound level. More information on this analysis will come in the "Analysis" section of our website.

Figure 12: Design of Experiment Means plot illustrating the effect of each controllable variable on our response variable, initial sound level, measured in decibels.

Figure 13: Design of Experiment Means plot illustrating the effect of each controllable variable on our response variable, average sound level over 90 seconds, measured in decibels per second.

Below, we have plotted the raw data of Rice Krispies' sound level in decibels when the cereal is immersed in a 1/2 cup of 2% milk over the course of 90 seconds. Figure 14 shows the relative consistency of the data points as well as an overall trend as time goes on from 0 to 90 seconds.

Figure 14: Plot of traditional x vs. y experiment demonstrating the sound level (dBA) of Rice Krispies immersed in a 1/2 cup of 2% milk over the course of 90 seconds.

DATA CONCLUSIONS

Preliminary Conclusions: During our data collection, we were able to infer some preliminary conclusions. The most obvious was the difference in sound level between Rice Krispies and the other two cereals. Whether in almond milk or 2% milk, the Rice Krispies had the longest lasting sound with the slowest deceleration of decibels, as well as the loudest measured sound level in general. Another conclusion that may not have been that obvious at all, was the effect that the type of milk had on the cereal. For example, the almond milk appeared (by best judgement) to have made the cereals a bit quieter. However, this was extremely difficult to distinguish, so this might mean that the different types of milk did not make a significant difference in sound level.

What Went Well: After a few of the first runs, we eventually formed a rhythm that kept the procedure moving. Also, we were able to collect, understand, and interpret data efficiently with the decibel app that was being used (Decibel X).

What Could’ve Gone Better: One complication was determining a sufficient benchmark decibel value for when to stop the running time. This was due to the massive difference in sound level and endurance between the dominant Rice Krispies and the other two cereals. This was also a contributing factor as to why it took a while to get a rhythm started as far as running the tests.

Recommendations for Future Teams: For future teams that wish to perform this experiment, it is recommended to have more than two people involved in the experiment. It is manageable with two, however the procedure and division of tasks would be much more efficient and much quicker. More recommendations would be to test other cereals that were not tested in this particular experiment (mostly out of curiosity of those results), and also to conduct the experiment in another location such as a study room to see how background noise levels compare to ours.

FUTURE WORK

The plan moving forward for this experiment would mostly consist of organizing our response variables in such a way that is relevant and makes sense. Our overall sound level data is fine; however, our second response variable may need more of a challenge to distinguish. The original plan was to measure the prolonged time that the cereal stayed above a specified decibel value. This may be unreasonable given that the Rice Krispies were many minutes longer than the other two, so there would be much inconsistency in the data collection. Therefore, our response variable might be changed to the calculated decibel deceleration (rate at which the decibel levels decreased in each cereal). This will work better because we can keep the time at which the run is taken is consistent for each trial.

Also, as stated previously, future work of this experiment may involve testing more cereals. This will reveal a wider range of conclusions of what cereals are loudest compared to others. As far as costs, the change in response variable mentioned above would not require additional costs, whereas testing more cereals would.

TEAM MEMBER CONTRIBUTIONS

Team member contributions were set equal between our two members. One person was involved in gathering the food and container supplies, measuring the data, and transferring the data to the website. The other person was then responsible for setting up each run/trial, compiling/organizing the data, and interpreting the data. The first person contributed about 3.5 hours to this process and the second person contributed about 4 hours. This was a total of 7.5 hours, resulting in labor costs of about $150.

LINK TO FULL DATA SET

Click here to see raw data