Models: Lecture quiz (fick 14/15 rätt) Flashcards
What are possible reasons why researchers/scientists use representations instead of investigating targets directly? Mark all correct options.
A)
Some targets are not physically accessible; they are, for example, too far away.
B)
Some targets are not scientifically interesting; we already understand them.
C)
Some targets are too complex to investigate directly; we cannot represent all aspects of the target.
D)
Some targets are easily understood; we often want to see something in a more complex way.
A)
Some targets are not physically accessible; they are, for example, too far away.
C)
Some targets are too complex to investigate directly; we cannot represent all aspects of the target.
If we find out that a theory is false, then we cannot use the theory as a model. Is this true or false? Mark the correct option.
A) True
B) False
B) False
Consider the following scenario. A person wishes to model the flow of water through a floodgate. She can either choose to model the flow of water using a simplified computer simulation, or she can construct a scale model of the floodgate in question. Which of the following claims are true? Mark all correct options.
A)
She can only use the scale model, because simplified simulations are not sufficently customized for real world targets.
B)
She can use either of the two models. Which one she chooses should primarily depend on the transparency of the model.
C)
She can only use the computer simulation, because constructions are inherently unreliable.
D)
She can use either of the two models. Which one she chooses should primarily depend on the purpose of the model.
D)
She can use either of the two models. Which one she chooses should primarily depend on the purpose of the model.
Assume that you are using a rat as a model for a human - but rats and humans are different in many ways, and similar in others. Familiarize yourself with the terms negative, neutral and positive analogies by matching each description with the correct term.
A)
Properties we have identified in the rat, but that we don’t know about in humans (yet)
B)
A drug reduces obesity in rats, but has not been tested on humans.
C)
Hormones are the same
D)
Lifespans are different
E)
General organ physiology is similar
F)
Brain size being different
A)
Properties we have identified in the rat, but that we don’t know about in humans (yet)
- Neutral analogy
B)
A drug reduces obesity in rats, but has not been tested on humans.
- Neutral analogy
C)
Hormones are the same
- Positive analogy
D)
Lifespans are different
- Negative analogy
E)
General organ physiology is similar
- Positive analogy
F)
Brain size being different
- Negative analogy
Decide which of these alternatives are similarities and which are differences between models and experiments. Match the alternatives with the correct description.
A)
The source of the errors is a…
B)
Being able to manipulate factors is a…
C)
That we observe results in both models and experiments is a…
D)
How certain we are that the result stems from our assumptions and not something else is a…
E)
The need for justifying the assumed similarity between what you are studying and what you want to find out about is a…
A)
The source of the errors is a…
- Difference
B)
Being able to manipulate factors is a…
- Similarity
C)
That we observe results in both models and experiments is a…
- Similarity
D)
How certain we are that the result stems from our assumptions and not something else is a…
- Difference
E)
The need for justifying the assumed similarity between what you are studying and what you want to find out about is a…
- Difference
Is similarity itself a recipe for good representations? Mark all correct options.
A)
Yes. Necessarily, a model with a higher degree of similiarity is a better representation than a model with a lower degree of similarity.
B)
No. Similarity in itself is not important, only the purpose of the model is.
C)
No. Similarity is only important with regard to purpose-relevant properties.
D)
Yes. All models must be similar in all properties.
C)
No. Similarity is only important with regard to purpose-relevant properties.
Imagine that you are to model the growth and decline of a population over time using a faucet to fill a transparent bucket of water with marked lines for the volume of liquid in it. Would this model be similar enough to serve its purpose? Mark all correct options.
A)
No. This model is not similar to the target in any respect.
B)
No. This does not allow for populations decreasing, perhaps a plugged hole should be added to the bucket.
C)
Yes. Growth is about the addition of people which the added water can represent fully.
B)
No. This does not allow for populations decreasing, perhaps a plugged hole should be added to the bucket.
Assume that a model M1’s parameters are implied by model M2’s parameters, Which model is more precise? Mark the correct option.
A) M1 is more precise.
B) M2 is more precise.
B) M2 is more precise.
Assume that the purpose is to use two globes as representations of the relative size of the moon and the sun, which of the following models are the simplest ones? Mark all correct options.
A)
A scale model of the moon and a scale model of the sun, both made of paper.
B)
A scale model of the moon and a scale model of the sun, that both take into account their position in the planetary system and their respective orbits.
C)
A scale model of the moon and a scale model of the sun, both made of plastic.
D)
A scale model of the moon and a scale model that introduces the span of solar flares.
A)
A scale model of the moon and a scale model of the sun, both made of paper.
C)
A scale model of the moon and a scale model of the sun, both made of plastic.
Robustness is a desirable virtue of models. Which of the models described below are robust? Mark all correct options.
A)
A model whose second (but not first) condition is changed and yields a different result.
B)
A model which has some of its conditions changed but yields the same result.
C)
A model which can have none of its parameters changed.
D)
A model which replaces none of its parameters but yields different results when applied to the same data.
B)
A model which has some of its conditions changed but yields the same result.
Match each model example (A-E) with its most prominent model virtue (1-5). Note that some of these models might have other virtues as well. We are simply asking what virtue would seem most prominent based on the short description of the model (and potentially other information).
A)
In economics, an individual’s demand for public goods U can be modelled according to a utility function F(X, G) where X is the individual’s private goods and G is the available public goods. The relation between X and G can be described via relatively simple mathematical equations (which most people can follow and understand).
B)
Statistical outliers are often problematic when doing a regression analysis whose underlying assumptions are false, in a given case. However, certain regression-analysis models are designed to be able to handle such outliers when the underlying assumptions (e.g. about a variable’s normal distribution) indeed are false.
C)
In order to get a better understanding of the danger in real car crashes, researchers have modelled car crashes with real cars and crash dummies. The dummies’ properties (e.g. their spines) are supposed to be, as much as possible, identical to the same properties in real humans.
D)
For any attempt at modelling a language that is complete and sound (with sufficient expressive power), it will be possible to deductively show that there will exist unprovable sentences in the language. This result is known as Gödel’s (first) incompleteness theorem.
E)
Viewed as a model, Newton’s laws of motion could – at the time – explain every observed object with fewer principles, as compared to the older Aristotelian view of motion and force.
1) Simplicity
2) Analytical tractability
3) Transparency
4) Robustness
5) Similarity
A)
In economics, an individual’s demand for public goods U can be modelled according to a utility function F(X, G) where X is the individual’s private goods and G is the available public goods. The relation between X and G can be described via relatively simple mathematical equations (which most people can follow and understand).
- Transparency
B)
Statistical outliers are often problematic when doing a regression analysis whose underlying assumptions are false, in a given case. However, certain regression-analysis models are designed to be able to handle such outliers when the underlying assumptions (e.g. about a variable’s normal distribution) indeed are false.
- Robustness
C)
In order to get a better understanding of the danger in real car crashes, researchers have modelled car crashes with real cars and crash dummies. The dummies’ properties (e.g. their spines) are supposed to be, as much as possible, identical to the same properties in real humans.
- Similarity
D)
For any attempt at modelling a language that is complete and sound (with sufficient expressive power), it will be possible to deductively show that there will exist unprovable sentences in the language. This result is known as Gödel’s (first) incompleteness theorem.
- Analytical tractability
E)
Viewed as a model, Newton’s laws of motion could – at the time – explain every observed object with fewer principles, as compared to the older Aristotelian view of motion and force.
- Simplicity
What is the reasoning behind using mirror-modelling rather than other modelling strategies? Mark all correct options.
A)
Since the model has a high degree of similarity in every relevant aspect, we can assume that the results we get would hold for the real life target as well.
B)
Mirror-models are relatively easy and time efficient to construct.
C)
We sometimes cannot experiment on the target, for instance if the target is an inhabited building and the experiment could risk harming inhabitants. But if the mirror-model is similar to the target in every aspect that matters for our purpose, we can still find out important things.
D)
By isolating features and idealizing them the mirror-model gives a clear answer, even if the mirror-model and the target are quite different.
A)
Since the model has a high degree of similarity in every relevant aspect, we can assume that the results we get would hold for the real life target as well.
C)
We sometimes cannot experiment on the target, for instance if the target is an inhabited building and the experiment could risk harming inhabitants. But if the mirror-model is similar to the target in every aspect that matters for our purpose, we can still find out important things.
Assume someone says: “When modelling the train car for the purpose of studying its behaviour when colliding with a car at an intersection, I am looking for a very life-like model where some minor change(s) to the car’s weight distribution would not dramatically alter the result.” Mark all correct options. You suggest that he/she should create a computer mirror-model because…
A)
A mirror model often has the similarity that is relevant for the purpose. That is, the mirror model will, if created with great care, be realistic to real collisions between cars and trains.
B)
A mirror model is often precise. That is, the specifications of a mirror model, which are highly accurate to real collisions between trains and cars, will imply the specifications of a model with less accurate specifications.
C)
A mirror model is often simple. That is, it will be relatively few properties to take into account when creating this model in order to make it mirror real collision between cars and trains.
D)
A mirror model is often transparent. That is, (assuming it’s a good mirror model) due to the model’s very high similarity, it will be possible for someone using the mirror model to grasp why a particular result is given as an output.
A)
A mirror model often has the similarity that is relevant for the purpose. That is, the mirror model will, if created with great care, be realistic to real collisions between cars and trains.
B)
A mirror model is often precise. That is, the specifications of a mirror model, which are highly accurate to real collisions between trains and cars, will imply the specifications of a model with less accurate specifications.
What is a requirement when using isolating models? Mark all correct options. (Fick fel, rätta gärna)
A)
That what you are studying does not consist of too many parts.
B)
That the model does not represent the real world target.
C)
That what you are studying can be divided into different parts.
D)
That no other kind of model is a practical alternative.
A)
That what you are studying does not consist of too many parts.
What is the relationship between the modelling virtues “similarity” and “simplicity”? Mark all correct options.
A)
A model may be both simpler and more similar to the target in comparison with another model.
B)
When evaluating two models, their level of similarity is assessed by comparing them to their target and their simplicity is evaluated by comparing them to each other.
C)
One can determine if a model is simple just by counting the number of model assumptions or model components.
D)
Simplicity and similarity are opposites. That is, when a model’s similarity goes up, its simplicity goes down.
E)
No model is simple or similar just in itself; they always need to be compared to something else.
A)
A model may be both simpler and more similar to the target in comparison with another model.
B)
When evaluating two models, their level of similarity is assessed by comparing them to their target and their simplicity is evaluated by comparing them to each other.
E)
No model is simple or similar just in itself; they always need to be compared to something else.
