You are on page 1of 4

The relationship between sales (units) and price (monetary units)

Quantity sold(y) Price (x1)


102 100
100 100
120 85
77 150
46 180
90 120
26 200
69 230
65 300
85 250

Simple Regression - Col_1 vs. Col_2


Dependent variable: Col_1
Independent variable: Col_2
Linear model: Y = a + b*X

Coefficients
Least Squares Standard T
Parameter Estimate Error Statistic P-Value
Intercept 117.177 20.0618 5.84081 0.0004
Slope -0.22844 0.108494 -2.10556 0.0683

Analysis of Variance
Source Sum of Squares Df Mean Square F-Ratio P-Value
Model 2494.56 1 2494.56 4.43 0.0683
Residual 4501.44 8 562.679
Total (Corr.) 6996.0 9

Correlation Coefficient = -0.597135


R-squared = 35.657 percent

Significance of the model

 Has the price an effect on the quantity sold?


H0: B1=0
H1: B10

So we have a two-tailed test. The value of t statistic is t=-2.105. We have to compare this value with the
critical values, which, in this case are ±tα/2;(n-2)= ±t0.025;8=±2.306.
Since the value of t-statistic is between the critical values then we cannot reject the null hypothesis (SEE
THE GRAPHS BELOW), meaning that the price do not have an effect on the quantity sold (which is odd
from the economic point of view).
 In general, if the price increases then the quantity sold will decrease. It’s that so?
H0: B1=0
H1: B1<0

Now, in this case we have an one-tailed test to the left. The critical value is -tα;(n-2)=-t0.05;8=-1.860.

Since the value of the t-statistic is less than the critical one, we can reject the null hypothesis (SEE THE
GRAPHS BELOW), so the price has a negative impact on the quantity sold, with 95% confidence.

So, here is the strange thing. We don’t have a relationship (see the two tailed test), but we have also a
negative impact of the price on the quantity (see the one tailed test). So where is the mistake?

In fact, the mistake is in the way of thinking, in the way we state the hypothesis, and this way has to be
connected with the economic.

Now let’s see the p-value approach

Acceptance and rejection zone for the null hypothesis


Two-tailed test

Rejection zone Rejection zone

Acceptance
zone

-t-statistic +t-statistic
One-tailed test (right)

Rejection zone

Acceptance
zone

+t-statistic

One tailed test (left)

Rejection zone

Acceptance
zone

-t-statistic

As we know, p-value is the smallest significance level at which the null hypothesis is rejected and it is not
connected with the level of significance in any way (do not depend on alpha which is always arbitrarily
chosen).

 For two-tailed test: p=P(t<- t-statistic OR t>+t-statistic)


 For one-tailed test(right): p=P(t>+t-statistic)
 For one tailed test (left): p=P(t<-t-statistic)

In our case t-statistic is t=-2.10556.

Using the statistical-calculator provided by www.danielsoper.com we have


So, in the case of two-tailed test p=0.0683, and in the case of one-tailed test
p=0.0341, so as you can see
pone tailed-test=ptwo-tailed test/2

You might also like