Scribd
Upload
8 views

Unit 522 Understanding and Visualizing Linear Equations Without Answers

Uploaded by

z13612909240
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
8 views

Unit 522 Understanding and Visualizing Linear Equations Without Answers

Uploaded by

z13612909240
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 8

Assignment for unit 522

Visualizing bivariate and simple trivariate hypotheses using linear


equations
Henk van der Kolk

16 October, 2023

Goals of the practice questions and this assignment


The goal of this assignment is to work with linear equations. In this assignment we use
linear equations as summaries of hypotheses. Later we ‘find’ these relationships and create
confidence intervals and do ‘tests’ about these linear equations by using data. This allows
us to say something about these relationships in the population.
Hypotheses can be communicated as:
• a ‘story’ (in words and sentences)
• a ‘causal diagram’ (with boxes/variables, arrows and connected pluses and
minuses)
• an ‘equation’ (with b’s or beta’s, and an intercept)
• a ‘line graph’ (with an x and a y axis and a linear or non linear ‘line’ showing the
relationship between the two)
After finishing this and some other units, I expect students to be able to translate one way
of communicating a hypothesis into the other. When using line graphs, we restrict attention
to bivariate hypotheses (we focus on hypotheses in which at least one variable is a scale
variable (interval or ratio) and trivariate hypotheses (we focus on hypotheses in which one
variable is either a dummy or a nominal variable and the others are scale variables). When
communicating multivariate models (more than 2 independent variables and one
dependent variable), I expect students to be able to use stories, causal diagrams and
equations.
In this assignment we restrict attention to additive linear models. This means that the
effects of the independent variables are independent: both contribute to the dependent
variable. In a later assignment you will also practice using interaction effects, in which
relationships depend on one another.

Understanding a bivariate linear relationship


Have a look at the figure below and at the equations that follow.

1
y ¿ β 0 + β 1 × x i.
y ¿ 3+0.5 × x i .

1. Why is a β used in the equation and not a b?


**
2. What does the value 3 mean in this equation and how is this value called?
**
3. What does the value 0.5 mean in this equation and how is this value called?
**
4. Suppose that a researcher hypothesizes that some variable (x) affects another
variable (y), without indicating whether this expected effect is positive or negative.
What would be the correct set of equations to describe this hypothesis?
a . ; H 0 : β1 =0 ; H 1 : β 1 ≠ 0
b . ; H 1 : β 1=0 ; H 0 : β 0 ≠ 0
c . ; H 0 : β 0=0 ; H 1 : β 1 ≧ 0
d . ; H 1 : β 1=0 ; H 1 : β1 ≧ 0

**

Describing some bivariate relationships


We now move to two other examples.
5. The following picture describes a linear relationship between two variables. Which
equation fits best? Is it positive or negative?

2
6. The following picture describes a linear relationship between two variables. Which
equation fits best? Is it positive or negative?

Trivariate linear relationships: addition with a dummy


Have a look at the figure below and at the equations that follow. We assume x and y are
both scale variables and the third variable is a dummy coded 0 (straight line) and 1 (dotted
line). The answer is given below this graph.

3
Y is your grade
X1 (study time) is a numerical variable

x2 can be activated = 1 or not 0. x2 is a dummy variable


Straight line = 0 (did not sleep)
Doted line = 1 (did sleep)

Y = b0 + b1x1 + b2x2
1st : Find the values for the reference category (coded 0)
- Intercept: when x=0, y = 3
- Slope: 0.5. with 2 points A(2;4) and B(4;5) -> (5-4)/(4-2) = ½ = 0.5
We can already fill in part of it: Y = 3+0.5x1 -1x2

y ¿ β 0 + β 1 × x 1i .
=β 2+ β3 × x 2i . ¿ this simplifies to: ¿ y ¿=β 2 + β 1 × x 1i . + β 3 × x 2i . ¿ y ¿=3+ ( 0.5 ) × x 1i . + (−1 ) × x 2i .
¿ ¿
7. Do the next one yourself. Which linear equation would summarize the expectation
in this graph, assuming that the second variable is a variable coded zero (straight)
and one (dotted)?

4
Codes: 0 for straight and 1 for doted -> My reference is straight line
- 1st: start with the reference category:
o Intercept = 2 and Slope = 1 -> Y = 2 + 1x + 1x2

- 2nd : then check the differences between lines

**
8. Which linear equation would summarize the expectation in this graph, assuming
again that the second variable is a variable coded zero (straight) and one (dotted)?

5
**
9. Which linear equation would summarize the expectation in this graph, yet again
assuming that the second variable is a variable coded zero (straight) and one
(dotted)?

6
**
These are all basic ‘additive’ models in which one of the two independent variables is a
dummy, coded 0 and 1.

Trivariate linear relationships (additive with a nominal variable)


The ‘third’ variable is a NOMINAL variable. In linear equations this implies we ‘dummify’ a
nominal variable first and add most dummies to linear equation. You also select a
‘reference category’. Explain in your own words how you will do that.
**
10. Which linear equation would summarize the expectation in this graph, again
assuming that the reference category of nominal variable is shown by a straight line,
group d2 is dashed and d3 is dotted?

11. Which linear equation would summarize the expectation in this graph, again
assuming that the reference category of nominal variable is shown by a straight line,
group d2 is dashed and d3 is dotted?
For reference: Intercept = 3; slope = -1.
For group d2: +(it’s above reference) and difference is +1
For group d3: - (it’’s below) and difference is -1
Y = 3 –1x +1x2 –1x3

7
What is the statistic associated with the group d2:

Let’s verify: I am from d2 and I studied (x=2 hours), what is y(my grade) ?
- Y = 3 –1x +1x2 –1x3
- Y = 3 –1(2) +1(1) –1(0)
- Y=3–2+1=2

**

What will we do after this?


Make sure you are able to create and interpret these equations correctly, for they form the
‘descriptive’ basis of linear models. You will later learn how to interpret more complex
models, with more than two variables and equations in which the lines are not parallel for
example.
One of the main ideas of ‘inferential statistics’ is testing whether the stated NULL
hypotheses (“there is NO relationship”) should be rejected in favor of the stated H1/HA
hypothesis. For this we need random samples. Testing in the context of statistics means
using data from a sample to check whether we can or cannot reject the null-hypothesis.
Confidence intervals in this context refers to the range in which the slope most probably is.
If that range includes zero, we cannot reject the null-hypothesis.
<< END OF THE ASSIGNMENT>>

You might also like