Lesson 12 Hypothesis Testing and Interpretation
Lesson 12 Hypothesis Testing and Interpretation
Learning Competencies
A hypothesis functions as an answer to the research question and guides data collection and
interpretation. A hypothesis enables researchers not only to discover a relationship between
variables, but also to predict a relationship based on theoretical guidelines and/or empirical
evidence. To give meaning to the hypothesis, there is a need to test it using the gathered data. In
this lesson, you are going to learn how to test the hypothesis using a statistical software (e.g.,
EXCEL or SPSS) or manually. As you undergo the rudiments of hypothesis testing, you will
appreciate the logic of inferential analysis which is essential in the correct interpretation of the
results. Samples are provided for you to easily pick up the process.
Inferential questions in your research demand the use of inferential statistics to analyze the
gathered data. Remember that in the previous lesson, descriptive statistics describes data (for
example, a chart or graph) and inferential statistics allows you to make predictions (“inferences”)
from that data. With inferential statistics, you take data from samples and make generalizations
about a population.
For example, you might stand in Novo Mall entrance and ask a sample of 100 people if they
like Chinese products. You could make a bar chart of yes or no answers (that would be descriptive
statistics) or you could use your research data (and inferential statistics) to reason that around 75-
80% of the population like shopping at Novo Mall although you interviewed only 100 samples.
1. Estimating parameters. This means taking a statistic from your sample data (for example the
sample mean) and using it to say something about a population parameter (i.e. the
population mean).
Probable Pairs of 2 samples: 1, 2:
1,3; 1,4; 1,5; 2,3; 2,4; 2,5; 3,4; 3,5;
N= 5 n=2
1, 2, 3, 4, 5
& 4,5
Sample (n)
Population (N)
2. Hypothesis tests. This is where you can use sample data to answer research questions. For
example, you might be interested in knowing if exposure to mass media will increase the
productive vocabulary of the high school students. Or if the number of hours spent in
reviewing the lessons increase the retention of the key concepts learned by the learners.
To explain the steps of hypothesis testing, let us compare the mean reading
comprehension scores of two groups (Male and Female) of students.
n=30
N = 150 Male = 13;
Female =17
Using this research scenario, let us learn how to conduct hypothesis testing. Testing
the hypothesis involves several steps:
1. Specify the null hypothesis.
As you learned in the previous lessons, the null hypothesis states the proposition that
two or more groups do not differ or variables X and Y are not related.
In the situation given, the null hypothesis could be: There is no difference in the
reading comprehension scores of the students grouped by sex.
This symbol is read as Mean Female minus Mean Female is equal to zero.
This symbol is read as Mean Female minus Mean Female is not equal to zero. The
difference is significantly greater than zero.
Number of
Hypothesis Variable Measurement Groups Statistical Tool
Scale Involved
There is no Reading scores Interval
difference in the Two (Male vs t-test for
reading scores of Sex Nominal Female) independent
students grouped groups
by sex
There is no Pretest scores Interval
difference in the One (Grade 9 t-test for
pretest and post- Post-test scores Interval students) dependent or
test scores of the paired groups
Grade 9 students
There is no
difference in the Communication Interval More than two One-way
communication anxiety scores groups ANOVA
anxiety scores of (Education,
the third-year College Affiliation Nominal Agriculture,
students grouped CICS, CHIM)
by college.
There is no
relationship Number of hours Ratio One group Pearson product-
between the spent reviewing (Grade 7 moment
number of hours lessons students) correlation
spent reviewing
lessons and the Number of Interval
number of concepts
concepts understood
understood by the
Grade 7 students
There is no
relationship Mobile phone Nominal One group Chi-square
between the Ownership (Grade 8 analysis
ownership of a students)
mobile phone and Place of residence Nominal
the place of
residence of the
Grade 8 students.
The term significance level means denoted by the Greek letter sigma alpha (α) is
generally set at 0.05 in educational researches. This means that there is a 5% chance that
you will accept your alternative hypothesis when your null hypothesis is actually true. If
you repeat the same study 100 times, chances are that 95 times the null hypothesis is
TRUE; only 5 times that is it NOT TRUE, meaning the alternative hypothesis is
CORRECT.
The smaller the significance level, the greater the burden of proof needed to
reject the null hypothesis, or in other words, to support the alternative hypothesis. If
you set your alpha or significance level at 0.01, you are hypothesizing that if the study is
repeated 100 times, 99 times, the null hypothesis is TRUE; only 1 time it is NOT TRUE.
It means therefore that if you set a smaller significance level, the chance of rejecting or
not accepting the null hypothesis is very small or slim. Thus, if you are rejecting the null
hypothesis at .05 level, you have 95% confidence that your finding is correct – there is a
SIGNIFICANT difference or relationship. If the null hypothesis is rejected at 0.01 level,
your level of confidence that you are correct in making a decision is 99%.
P-value = 0.01 This will happen 1 in 100 times by pure chance if your null
hypothesis is true. Not likely to happen strictly by chance. If test statistic has a p-
value of .000 to .01, you have to REJECT the null hypothesis and ACCEPT the
alternative hypothesis to be true. Then, you claim in your study that there is a
significant difference or relationship, whatsoever is the hypothesis.
P-value = 0.75 This will happen 75 in 100 times by pure chance if your null
hypothesis is true. It is very likely to occur strictly by chance. If your test statistic has
a p-value of 0.051 and higher, you have to ACCEPT that your null hypothesis is
TRUE.
NOTE: There are two ways to check if the computed test statistic is asking you to accept
or reject the null hypothesis.
The p-values are indicated in all software statistical computations, like EXCEL
Data analysis, SPSS, etc. If the significance level you set in Step 3 is 0.05 and the
corresponding p-value of the computed test statistic is lower, you need to
REJECT the null hypothesis. For example, if the p-values are .04,.039, .025,
.019, it means that the difference between or among groups are significantly
different, so REJECT the null hypothesis. In correlation analysis, it would mean
X is significantly related with Y. Any p-value greater than 0.05 (e.g., .051, .06,
.073, etc.) indicates that you have to ACCEPT the null hypothesis, “There is no
significant difference between or among groups” or “There is no significant
relationship between X and Y.”
b. Using the computed test statistic and the tabular/critical value
If you are MANUALLY computing for the test statistic, you can make a
statistical decision by comparing the computed value of the statistic with the
tabular value (found at the back of any statistics book or in Google) at the
designated degrees of freedom (df). See Annexes A to E for the manual
computation of the different test statistics.
If the computed value is greater than the tabular/critical value, your decision is
to REJECT the null hypothesis. If your computed value is smaller than the
tabular value, your decision is to ACCEPT the null hypothesis.
Examine the Excel t-test for independent groups results. Both the p-value of the
statistic t and the critical/tabular value are presented. The same rejection of the
null hypothesis is correct.
This is the results of the t-test for independent groups using Excel Data Analysis;
t-Test: Two-Sample Assuming Unequal Variances
Male Female
Mean 14.38461538 16.82352941
Variance 3.756410256 6.529411765
Observations 13 17
Hypothesized Mean Difference 0
df 28
t Stat -2.972876048
P(T<=t) one-tail 0.003003304
t Critical one-tail 1.701130934
P(T<=t) two-tail 0.006006609
t Critical two-tail 2.048407142
The p-value of the test statistic t = 2.973 (rounded off) is p = 0.006. This value
is lower than .05 set, so you have to REJECT the null hypothesis “There is no
difference between the reading scores of students grouped by sex.” Look also at
the comparison between the computed versus the critical value 2.973 vs. 2.048.
The computed t-value is higher or greater than 2.048, so the same decision of
rejecting the null hypothesis is correct. SO, whether you use the p-values and
critical values as point of comparison to make a statistical inference, the same
results would be given. In t-test, ignore the sign of the computed value; get the
absolute value always. The sign (+ or -) depends on which group you use first
and last. If the first is bigger than the second, the t-value is positive; if reversed,
the t-value is negative.
Note: Even if you are manually computing the test statistic, you can Google the
associated probability by typing Probability Calculator associated with _____
(give the name of the tests, t-test, Pearson r, chi-square). Once there, it asks you
the computed test statistic and the degrees of freedom. Once done, in a click,
the p-value is given.
After you make a decision to reject or accept the null hypothesis, you have to
interpret the results. If t-test gives a significant result (the null hypothesis is
rejected), identify which group has a higher mean than the other. Then, explain the
probable factors/reasons that made the difference significant. Use your review of
related literature to support your finding and argument. From your arguments, you
can come out with a valid and reliable conclusion.
Sample Interpretation of the Findings Using the Example Above
As shown in the table, female students with a mean score of 16.82 outscored
the males who obtained a mean score of 14.38. It indicates that female students
understand better what they read than their counterparts. A study conducted by
Logan and Johnston (2009:200) about the relationship between reading
comprehension and gender comes up with the finding that: Girls are better in
reading comprehension than boys; Girls read more frequently than boys do; and
Girls have more positive attitude to reading. In another study, females are better
than males in foreign language comprehension, “in terms of language
comprehension, several studies have demonstrated female superiority.”(Saidi,
2012:231).
Computed
Group Mean Variance t-value P - value
Female 18.823 6.529
Male 14.384 3.756 2.973** 0.006
** = significant at 0.01
Single asterisk (*) means significant at 0.05 level. Just state “There is a significant
difference . . . .
Double asterisks (**) means highly significant at 0.01 level. Just state “There is a
highly significant difference . . . .
Activity 1. In your own words, explain the major steps in hypothesis testing.
Activity 2. In your own words, explain the two ways to make a statistical decision (rejecting or
accepting the null hypothesis).
Activity 4. With your chosen research topic/title and inferential research question, complete
the following table by supplying the needed information. Note , for every variable, identify the
measurement scale and statistical tool to be used.
QUIZ
1. This function of statistical inference allows the research to make statements about the population
when only a portion of it was observed.
2. This value is the one computed from the sample data which is the estimate of the population
information.
3. This value, found at the annexes of a statistics books, is the reference value to determine if the
computed value permits the rejection or acceptance of the null hypothesis.
4. This value tells you the initial amount to set to determine when to reject or acct the null hypothesis.
5. This is the decision to take if the computed value is greater than the critical value.
Directions: Write Yes if the researcher did the correct action; No if it is wrong.
6. Researcher Joan rejected the null hypothesis when the computed value has a p-value of 0.07.
7. Researcher Mark used chi-square to analyze his data to test the hypothesis “There is no difference in
the average grades of students from central and barangay high schools.”
8. Researcher Jayden accepted the null hypothesis when the computed value is 4.15 but the
critical/tabular value is 8.11.
9. Researcher Marionette used t-test in testing the hypothesis that “There is no association between
average grades and number of hours to study assignments.”
10. In a correlation analysis, Researcher Toni found a positive coefficient (r = 0.789; p = .001) for
number of hours spent to text message and number of productive vocabulary used by the students.
He concludes that the longer a student sends text messages, the more vocabulary words he uses.