FA24 Psy 202 Sections 2 & 7 Conceptual Assignment 5

1. Why do we need inferen al sta s cs?

We need inferen al sta s cs because we can take informa on from the sample to es mate the
popula on standard devia on and mean. We take a sample of informa on to es mate the
popula on informa on.
How does probability theory help us to make inferences?
The probability theory helps us make inferences by using informa on from a sample to make
predic ons about a popula on.
How does probability di er from sta s cs?
When using probability, you start by knowing about the popula on and predict what you will get
in the sample. You know what the popula on looks like. You are given a model, such as a fair
coin, and use it to predict data.
In sta s cs, you start by knowing the data and determining what model to use.

2. Explain the frequen st view of probability.

In the frequen st view, for example, if you repeatedly ip a coin over and over, the outcome will
be half of the ips are heads and the other half is tails. It looks at the rela ve frequency of an
event over an in nite series of trials. It is objec ve.
How does the Bayesian view of probability di er from the frequen st view?
The Bayesian view is not objec ve, it is subjec ve. Because we cannot en rely comprehend
in nity which is what the frequen st view of probability uses, the Bayesian view goes o of what
is en rely subjec ve. For example, an elec on is usually only run once, it looks at subjec vity. It
is a li le more complicated, the subjec ve probabili es are based on your prior probabili es.
In the frequen st view, you are star ng o asking about the probability of these events, in the
Bayesian view, it ques ons the probability of a model.
On which of these two views are most of our analyses based?
If you have a reason to have strong con dence in the priors, Bayesian is be er. But, because in
this class we don’t have strong con dence in our priors, most of our analyses are based on the
frequen sts’ view.

3. De ne an elementary event.
In an elementary event, you only have one outcome that sa s es the condi on.
How do sample spaces relate to elementary events?
Sample spaces list all possible outcomes, therefore it lists all of the possible elementary events
you could get.
How do non-elementary events relate to elementary events?
fi
fi
ti
tt
ti
ti
ti
ti
ti
ti
fl
ti
ti
fi
ti
ti
ti
ti
ti
ti
ff
ti
ti
ti
ti
ti
ti
fi
ti
ti
ti
ti
ti
fi
ti
ff
ti
ti
ti
ff
ti
ti
fl
ti
ti
fi
ti
ti
ti
ti
ti
ti
tt
ti
ti
ti
ff
ti
ti
 In a non-elementary event, you have 2 or more outcomes that sa sfy the condi on. Non-
elementary events are made of a bunch of elementary events.

4. If an elementary event has a probability of 0, what does that mean?

If an elementary event has a probability of 0, that means it is an event that never happens.
If an elementary event has a probability of 1, what does that mean?
If an elementary event has a probability of 1, that means the event is certain to occur.
According to the law of total probability, what should the probabili es in a probability
distribu on sum to?
The probabili es in a probability distribu on sum should always be 1.

5. When calcula ng probabili es for non-elementary events, what is P(A∪B)?

This means that we are looking for something that sa s es condi on A or condi on B.
What is P(A∩B)?
P(A∩B) means that we are looking for outcomes that sa sfy both condi on A AND condi on B.

6. Why do we need to subtract P(A∩B) when calcula ng P(A∪B)?

You subtract P(A∩B) so that you are le with only the informa on from condi on A and
condi on B because when calcula ng the sum of P(A∪B), condi ons A & B are counted twice, so
you need to subtract P(A∩B) from P(A∪B).
When do we not need to perform that subtrac on?
You don’t need to perform that subtrac on when you are adding the probability of A and the
probability of B when they are not combined/mutually exclusive.
In the formula for P(A∩B), what is P(A|B)?
This is asking what the probability of A given that B has already occurred.
When can we reduce P(A|B) to just P(A)?
We can make this reduc on if they, meaning A and B, are independent. The two events are
mutually exclusive.

7. De ne a random variable.
A random variable is any variable whose value is determined completely by probability. There
are two types of random variables: discrete and con nuous. Discrete random variables have a
limited number of possible outcomes. Con nuous random variables can be frac onal amounts,
meaning an in nite number of possible amounts.

8. What are the ve characteris cs of Bernoulli trials (A.K.A. binomial trials)?

1. Only two possible outcomes (mutually exclusive)
fi
ti
ti
ti
fi
ti
fi
ti
ti
ti
ti
ft
ti
ti
ti
ti
ti
ti
fi
ti
ti
ti
ti
ti
ti
ti
ti
ti
ti
ti
ti
ti
 2. Constant theta— meaning there is the same probability of success on a single trial completed
3. There must be a speci ed number of trials
4. The trials must be independent
5. Random variable (x): the number of successes in n trials
In a binomial distribu on, what does the size parameter (N) refer to?
The size parameter (N) refers to the number of trials taking place.
In a binomial distribu on, what does the success probability (ϴ) refer to?
The success probability (ϴ) refers to the probability of a single trial resul ng in a successful
outcome within a binomial experiment.

9. What does it mean for scores to be normally distributed?

If scores are normally distributed, the data is symmetrically distributed around a central region.
These distribu ons are always bell-shaped and have an in nite number of possible values but
never touch the x-axis. They are all perfectly symmetrical and have a skewness & kurtosis of
zero. Addi onally, the mean = median = mode— all measures of central tendency are the exact
same number. The points of In ec on at -1a and +1a.
Which two descrip ve sta s cs dis nguish one normal distribu on from another?
If you know the mean and know the standard devia on, you can predict the probability of every
possible value you could get on the x-axis— which are the two descrip ve sta s cs that
dis nguish normal distribu ons.
How does the normal distribu on di er from the binomial distribu on?
Binomial distribu ons are discrete, while normal distribu ons are con nuous.

10. In a normal distribu on, what percentage of the area under the curve falls within one
standard devia on of the mean?
~68%
What percentage falls within two standard devia ons of the mean?
~95%
What percentage falls within three standard devia ons of the mean?
~99.7%
What percentage falls below the mean?
~50

11. When would you use a t-distribu on instead of a normal distribu on?
You would use a t-distribu on when you have a small sample size and the popula on standard
devia on is unknown.
ti
ti
ti
ti
ti
ti
ti
ti
ti
fi
ti
ti
ti
ti
ti
fl
ti
ti
ti
ti
ff
ti
ti
ti
ti
fi
ti
ti
ti
ti
ti
ti
ti
ti
ti
 How does the shape of the t-distribu on di er from that of the normal distribu on?
In a t-distribu on, it appears taller than a normal distribu on because “extreme” scores are
more likely so they have fa er tails.

12. For which scale of measurement would we use a χ2 distribu on?

We would use it for categorical data, so when we are dealing with nominal scores, that’s when
we use chi-squares.
How does the χ2 distribu on relate to the normal distribu on?
If you take a certain number of scores from a normal standard curve, square them, and add
them up, that will give you the chi-square distribu on.
What is the shape of the χ2 distribu on?
It is always right-skewed in shape.

13. How does the F-distribu on relate to the χ2 distribu on?

The F-distribu on is the ra o of two independent chi-square distribu ons. Each chi-square is
divided by its respec ve degrees of freedom.
What do the shapes of the F-distribu on and the χ2 distribu on have in common?
Both of these distribu ons are always posi vely skewed and only take posi ve values.

14. Of the ve probability distribu ons discussed in this chapter, which is considered the
most important?
The Normal distribu on is considered the most important distribu on.
Why is that probability distribu on considered the most important?
It is considered the most important because it is symmetrical and bell-shaped, so, even if the
values change, the distribu on remains normal. We can use normal distribu ons to draw
conclusions about other data as it accurately describes the distribu on of values.
fi
ti
ti
ti
ti
ti
ti
ti
tt
ti
ti
ti
ti
ti
ti
ti
ti
ff
ti
ti
ti
ti
ti
ti
ti
ti
ti
ti
ti
ti