You are on page 1of 1

Search on Cross Validated… 1

SPONSORED BY

Home
What is the difference between 𝛽1 and 𝛽 1̂ ? Ask Question

Questions
Asked 4 years, 8 months ago Active 3 months ago Viewed 42k times
Tags

Users Suppose I have a random sample {𝑥𝑛 , 𝑦𝑛 }𝑁


𝑛=1 . Hot Meta Posts
Unanswered
11 Suppose
12 Handling of cross-posted questions when
flagged
𝑦𝑛 = 𝛽 0 + 𝛽 1 𝑥𝑛 + 𝜀 𝑛

and Related
6
𝑦𝑛̂ = 𝛽0̂ + 𝛽1̂ 𝑥𝑛 (𝑀) (𝑁)
𝛽𝑖̂ ∼ 𝛽𝑖̂ for linear regression?
0

What is the difference between 𝛽1 and 𝛽1̂ ? 7 What are the consequences of “copying” a
data set for OLS?

2
regression 3 how to prove that 𝜎̂ is a consistent for
𝜎2 ?

Share Cite Edit Follow edited Oct 2 '16 at 21:42 asked May 2 '16 at 23:12 0 Conditional Variance of Linear Regression
whuber ♦ Stan Shunpike Coefficients 𝐶𝑜𝑣(𝛽0̂ , 𝛽1̂ |𝑊 ∗ )
254k 40 561 988 3,341 2 24 33
5 For simple linear regression, is 𝛽1 linear in
𝑦𝑖 ?
6 What is a random variable and what isn't in
6 𝛽 is your actual coefficient and 𝛽̂ is your estimator of 𝛽. – ARAT May 2 '16 at 23:25 regression models

1 Isn't this a duplicate of an earlier post? I would be surprised... – Richard Hardy May 3 '16 at 6:44
Hot Network Questions
add a comment
Dif-in-Dif aggregating or not?

Is there a webpage that shows the night sky but


3 Answers Active Oldest Votes
can filter out dim stars?

How does one take advantage of unencrypted


𝛽1 is an idea - it doesn't really exist in practice. But if the Gauss-Markov assumption hold, 𝛽1 traffic?

would give you that optimal slope with values above and below it on a vertical "slice" vertical to the Are "three way" trades allowed in Catan?

10 dependent variable forming a nice normal Gaussian distribution of residuals. 𝛽1̂ is the estimate of In Europe, can I refuse to use Gsuite / Office365 at
𝛽1 based on the sample. work?

What are the earliest inventions to store and


The idea is that you are working with a sample from a population. Your sample forms a data cloud, release energy (e.g. fly wheels)?

if you will. One of the dimensions corresponds to the dependent variable, and you try to fit the line Realistic task for teaching bit operations
that minimizes the error terms - in OLS, this is the projection of the dependent variable on the Numerate duplicate of every value
vector subspace formed by the column space of the model matrix. These estimates of the
population parameters are denoted with the 𝛽 ̂ symbol. The more data points you have the more
How Functional Programming achieves "No
runtime exceptions"
accurate the estimated coefficients, 𝛽𝑖̂ are, and the better the estimation of these idealized Excess income after fully funding all retirement
population coefficients, 𝛽𝑖 . accounts. Now what?

Why is there no spring based energy storage?


Here is the difference in slopes (𝛽 versus 𝛽 )̂ between the "population" in blue, and the sample in Why can't politicians comment on individual
isolated black dots: cases?

How do airplanes maintain separation over large


bodies of water?

Don't count your eggs before they hatch

Reason to use tridents over other weapons?

What does it mean for a word or phrase to be a


"game term"?

Can an Airline board you at departure but refuse


boarding for a connecting flight with the same
airline and on the same ticket?

Is my triangle on the lattice?

How to put aside personal grievances during


performance reviews

Why does diode keep conducting even after


voltage across it is negative

Can an electron and a proton be artificially or


naturally merged to form a neutron?

Paid off $5,000 credit card 7 weeks ago but the


money never came out of my checking account

When selling stocks at a stop-loss price during a


crash, who do I actually sell them to?

What does "rücken zusammen" mean in this


case?

Question feed

The regression line is dotted and in black, whereas the synthetically perfect "population" line is in
solid blue. The abundance of points provides a tactile sense of the normality of the residuals
distribution.

Share Cite Edit Follow edited May 3 '16 at 9:22 answered May 2 '16 at 23:40
Antoni Parellada
21.6k 14 84 176

add a comment

The "hat" symbol generally denotes an estimate, as opposed to the "true" value. Therefore 𝛽 ̂ is an
estimate of 𝛽. A few symbols have their own conventions: the sample variance, for example, is
10 2
often written as 𝑠2 , not 𝜎̂ , though some people use both to distinguish between biased and
unbiased estimates.

In your specific case, the 𝛽 ̂ values are parameter estimates for a linear model. The linear model
supposes that the outcome variable 𝑦 is generated by a linear combination of the data values 𝑥𝑖 s,
each weighted by the corresponding 𝛽𝑖 value (plus some error 𝜖 )

𝑦 = 𝛽 0 + 𝛽 1 𝑥1 + 𝛽 2 𝑥2 + ⋯ + 𝛽 𝑛 𝑥𝑛 + 𝜖

In practice, of course, the "true" 𝛽 values are usually unknown and may not even exist (perhaps the
data is not generated by a linear model). Nevertheless, we can estimate values from the data that
approximate 𝑦 and these estimates are denoted as 𝛽 .̂

Share Cite Edit Follow edited Oct 8 '20 at 14:05 answered May 2 '16 at 23:41
Matt Krause
18k 2 57 98

add a comment

The equation

4 𝑦𝑖 = 𝛽 0 + 𝛽 1 𝑥𝑖 + 𝜖 𝑖

is what is termed as the true model. This equation says that the relation between the variable 𝑥
and the variable 𝑦 can be explained by a line 𝑦 = 𝛽0 + 𝛽1 𝑥. However, since observed values are
never going to follow that exact equation (due to errors), an additional 𝜖𝑖 error term is added to
indicate errors. The errors can be interpreted as natural deviations away from the relationship of 𝑥
and 𝑦. Below I show two pairs of 𝑥 and 𝑦 (the black dots are data). In general one can see that as
𝑥 increases 𝑦 increases. For both of the pairs, the true equation is

𝑦𝑖 = 4 + 3𝑥𝑖 + 𝜖𝑖

but the two plots have different errors. The plot on the left has large errors and the plot on the right
small errors(because the points are tighter). (I know the true equation because I generated the data
on my own. In general, you never know the true equation)

Lets look at the plot on the left. The true 𝛽0


= 4 and the true 𝛽1 = 3. But in practice when given
data, we don't know the truth. So we estimate the truth. We estimate 𝛽0 with 𝛽0̂ and 𝛽1 with 𝛽1̂ .
Depending on which statistical methods are used, the estimates can be very different. In the
regression setting, the estimates are obtained via a method called Ordinary Least Squares. This is
also know as the method of line of best fit. Basically, you need to draw the line that best fits the
data. I am not discussing formulas here, but using the formula for OLS, you get

𝛽0̂ = 4.809 and 𝛽1̂ = 2.889

and the resulting line of best fit is,

A simple example would be the relationship between heights of mothers and daughters. Let 𝑥 =
height of mothers and 𝑦 = heights of daughters. Naturally, one would expect taller mothers to have
taller daughters (due to genetic similarity). However, do you think one equation can summarize
exactly the height of a mother and a daughter, so that if I know the height of the mother I will be
able to predict the exact height of the daughter? No. On the other hand, one might be able to
summarize the relationship with the help of an on an average statement.

TL DR: 𝛽 is the population truth. It represents the unknown relationship between 𝑦 and 𝑥. Since we
cannot always get all possible values of 𝑦 and 𝑥, we collect a sample from the population, and try
and estimate 𝛽 using the data. 𝛽 ̂ is our estimate. It is a function of the data. 𝛽 is not a function of
the data, but the truth.

Share Cite Edit Follow answered May 2 '16 at 23:57


Greenparker
13.3k 3 34 73

add a comment

Your Answer

Links Images Styling/Headers Lists Blockquotes Code HTML Tables Advanced help

Post Your Answer

Not the answer you're looking for? Browse other questions tagged regression or ask your own
question.

CROSS VALIDATED COMPANY STACK EXCHANGE Blog Facebook Twitter LinkedIn Instagram
NETWORK
Tour Stack Overflow
Technology
Help For Teams
Life / Arts
Chat Advertise With Us
Culture / Recreation
Contact Hire a Developer
Science
Feedback Developer Jobs
Other
Mobile About
Disable Responsiveness Press
Legal
Privacy Policy
site design / logo © 2021 Stack Exchange Inc; user contributions licensed
Terms of Service under cc by-sa. rev 2021.1.12.38303

You might also like