Testing hypothesis of a single population
We often use inferential statistics to make decisions or judgments about the value of a
parameter mean. One of the most commonly used methods for making such decisions or
judgments is to perform a hypothesis test.
A hypothesis is simply a statement that something is true. ex: “ the mean weight of all bags
of pretzels being packaged differs from the advertised weight of 454 grams “ .
Hypothesis testing: is a statistical procedure which evaluate whether a statement about a
parameter from one or more populations is supported by data on sample drawn at random
from those populations. This statement compares the population mean to a given value.
Compares the population distribution in with respect to means in case of Quantitative
variables and proportions in case of dichotomous.
Testing hypothesis procedure depends on:
Nature of variable: Quantitative continuous or qualitative.
Distribution of Data: whether distribution is normal or not.
Sample size: Large (n > 30) or small (n < 30).
Standard deviation of population: whether known or unknown.
Example:
An investigator wishes to know if the distribution of normal children in Lebanon w.r.t IGM is
different compared to that outside. Mean of IGM for western children through estimation to
be M = 1.00 mg/ml. The investigator collected the data on lebanese children and the result :
x = 1.21 mg/ml
s = 0.54 mg/ml
n = 56
Question: Is L = 1. could this result be from a of population = 1. Is the mean of IGM of
normal children in Lebanon is different than that for western children.
Testing procedure traditional method:
A. Testing hypothesis of when n 30 or is Known: z- distribution
To perform a hypothesis test or test of significance, we take the following steps:
1. State the hypotheses:
A hypothesis test involves two hypotheses. The null and alternative hypotheses.
Null hypothesis: referred to as hypothesis of no difference. A claim that there is no
difference between the population mean and the hypothesized value 0. We use the symbol
H0 to stand for the null hypothesis.
H0: = 0. In our case H0: L = 1.00 mg/ml .
Alternative hypothesis: Research hypothesis. An alternate to the null hypothesis. We use the
symbol HA to stand for the alternate hypothesis. Three choices are possible for the
alternative
hypothesis:
a. If our primary concern is deciding whether a population mean, , is different from
a specified value 0 then the alternative hypothesis is: H A: 0. Hypothesis test
is called a two-tailed test
b. If our primary concern is deciding whether a population mean, , is less than a
specified value 0 then the alternative hypothesis is: H A: < 0. Hypothesis test is
called a one-tailed test, in more specific a left-tailed test.
c. If our primary concern is deciding whether a population mean, , is greater than a
specified value 0 then the alternative hypothesis is: H A: >0. Hypothesis test is
called a right-tailed test.
In our case HA: L 1.00 mg/ml .
2. Assumptions:
Sample is random and independent
n 30 , apply central limit theorem and thus shape of sampling distribution of x is
approximately normal .
3. Assume H0 is true. If H0 is true then L = 1.00mg/ml. If we draw the sample distribution of
56 children it will be a normal distribution. some values are likely to be in sampling
distribution and some unlikely. by convenience define a significant level.
Significance level: Symbolized by , is the probability value that forms the boundary
between rejecting and not rejecting the null hypothesis. It is the level of significance with
which we would be willing to reject the true null hypothesis. If for example a 5%
significance level is chosen, then there are 5 chances in 100 that we would reject H 0. when a
test statistic lies in this area, the result is referred to as significant at the level.
Define the critical region: The region on the far end of the distribution. If one end of the
distribution is involved one-tailed test. If both ends are involved two-tailed test. The
critical region is called the rejection region. Is the range of values of the test value that
indicates that there is a significant difference and that the null hypothesis should be
rejected.
When computed Z falls in the critical region we reject the null hypothesis. The probability
that a test statistic falls in the critical region is denoted by . ( note: The nonrejection
region is denoted by (1 - ).
The critical value separates the critical region from non-critical region.
In our case from the Z distribution table we find that the critical value Z is 1.96 for a
significance level of 5%.
a.
b. left-tailed test
c.
4. Compute the value of the test statistic: Uses the data obtained from a sample to make a
decision about whether or not the null hypothesis should be rejected.
x - 0 1.21 - 1.00
Z0 = ----------- = ------------------ = 2.92
/n 0.54/ 56
5. Compare test statistic to critical value and state your decision: If the value of the test
statistic falls in the rejection region, then reject H0; otherwise, accept H0.
Z0 > 1.96 so we reject H0 .
6. Conclusion: The population mean of IGM of Lebanese normal children is different than
1.00mg/ml.
P-value: It is the area in the tail or tails of a distribution beyond the value of test statistic.
The probability that the value of the calculated test statistic, or a more extreme one,
occurred by chance alone is denoted by P.
B. Testing hypothesis of when n ˂ 30 & is unknown: t-distribution
Steps of testing hypothesis:
1. State the hypotheses
a. H0: = hypothesized value HA: hypothesized value
b. H0: ≥ hypothesized value HA: < hypothesized value
c. H0: ≤ hypothesized value HA: > hypothesized value
2. Assumptions
n<30 & is unknown thus use t-distribution
-Assume population is normally distributed.
- Assume random sample
Assume H0 is true
3. Define significant value, critical value, rejection region and draw the graph.
4. Compute the value of the test statistic
5. Compare test statistic to critical value
6. State decision
7. State conclusion
Example:
A smog alert is issued when the amount of a particular pollutant in the air is found to be
greater than 7 ppm. Samples collected from 16 stations give an x of 7.84 with an s of 2.01.
Do these findings indicate that the smog alert criterion has been exceeded, or can the results
be explained by chance? Since is not known and n is small, we rely on the t test.
1. H0: 7.0 ppm
HA: > 7.0 ppm
2. Assumptions
n<30 & is unknown thus use t-distribution
-Assume population is normally distributed.
- Assume random sample
Assume H0 is true
3. = 0.05 ; Critical region: Since the H A: > 7.0 ppm indicates a one-tailed test, we place
all of = 0.05 on the positive side. From T-table we find that for t15,0.05 = 1.753.
4. Test statistic:
x- 7.84-7.0 0.84
t = ------- = --------------- = ---------- = 1.68
s/n 2.01/16 0.50
5. Compare test statistic to critical value: 1.68 < 1.753 . t = 1.68 does not fall in the critical
region.
6. Fail to reject H0 .
7. Conclusion: The critical air pollution level of 7 ppm was not exceeded, or smog alert
criterion has not been exceeded.
C. Testing hypothesis of p: z-distribution
Steps of testing hypothesis:
1. State the hypotheses
a. H0: p = hypothesized value HA: p hypothesized value
b. H0: p ≥ hypothesized value HA: p < hypothesized value
c. H0: p ≤ hypothesized value HA: p > hypothesized value
2. Assumptions and conditions
np ≥5 & nq ≥5
apply central limit theorem thus shape of sampling distribution of ^p is approximately normal
thus use z distribution
Assume H0 is true
3. Define significant value, critical value, rejection region and draw the graph.
4. Compute the value of the test statistic
5. Compare test statistic to critical value
6. State decision
7. State conclusion
Example:
A statistian read that at least 77% of the population oppose replacing $1 bills with $1 coins.
To see if this claim is valid, the statistian selected a sample of 80 people and found that 55
were opposed to replacing the $1 bills. At α = 0.01, test the claim that at least 77% of the
population are opposed to the change.
Solution:
1. State the hypotheses
H0: p ≥ 0.77 HA: p < 0.77
2. Assumptions and conditions
np = 80 x 0.77= 61.6 ≥5 & nq = 80x (1-0.77)= 18.4 ≥5. Apply central limit theorem thus
shape of sampling distribution of ^p is approximately normal
thus use z distribution
3. Define significant value, critical value, rejection region and draw the graph.
4. Compute the value of the test statistic:
5. Compare test statistic to critical value : -1.75 > -2.33
6. State decision : Do not reject H0
7. State conclusion: There is not enough evidence to reject his claim that at least 77% of the
population oppose replacing $1 bills with $1 coins.
