LUNA COLLEGES

Tayug,Pangasinan
FINAL EXAM IN EDF 202-ADVANCED EDUCATIONAL STATISTICS
2nd SEMESTER, 2023-2024
l. Discuss clearly and comprehensively the following. Be able to cite their relevance in
the field education
1. Descriptive and inferential statistics
Descriptive statistics is a method of data analysis. The objective here is to compute
quantities that characterize our data. The average or center of the data is indicated by a
number called the mean, median, or mode. The standard deviation and range are used to
express how dispersed the data is. More complex methods, such regression and correlation,
are used to characterize paired data.
Inferential statistics are used when starting with a sample and attempting to draw
conclusions about the population. The subject of hypothesis testing comes up while working
with this branch of statistics. Here, the scientific aspect of statistics is demonstrated by the
way we formulate a hypothesis and then utilize statistical methods and our sample to assess
whether or not to reject it. Actually, this explanation only scratches the surface of this
extremely helpful statistical component. Compile and arrange a data set's attributes. A group
of observations or answers from a sample or the complete population is called a data set.
In quantitative research, describing response characteristics—such as the average of one
variable, like age, or the relationship between two variables, like age and creativity—is the
first stage in statistical analysis after data collection. For practically each research activity,
descriptive analyses are essential.

You aim to draw inferences from inferential statistics that go beyond the immediate facts. To
try and determine what the public would believe based on sample data, for example, we
employ inferential statistics. Alternatively, we assess the likelihood that an observed
difference between groups in this study is reliable or if it could have occurred by chance
using inferential statistics. Therefore, whereas descriptive statistics are used to explain what
is happening in our data, inferential statistics are used to draw conclusions from our data to
more general situations.

2. Parametric and Non- Parametric Tests

Tests that use parametric assumptions presume that the sample data originates from a
population with a fixed set of parameters and a probability distribution similar to the normal
distribution.
Typical parametric tests concentrate on examining and contrasting data's variance or mean.
The most widely used central tendency measure to characterize data is the mean, however
outliers have a significant influence on it as well. It is crucial to examine your data and decide
if using the mean to represent it is the best course of action. If so, using parametric tests is the
best course of action! In the event that the median more accurately depicts your data,
nonparametric testing may be a preferable choice.
Nonparametric tests rely on fewer assumptions—for example, that the outcome is roughly
regularly distributed—they are also referred to as distribution-free tests. Certain probability
distributions, such as the normal distribution, are used in parametric tests, which entail
estimating the distribution's essential parameters—such as the mean or difference in means—
from the sample data. Less assumptions comes at the expense of nonparametric tests being
generally less powerful than their parametric equivalents (i.e., they may be less likely to
reject H0 when the alternative is true). Determining if a continuous outcome has a normal
distribution and, consequently, whether a parametric or nonparametric test is appropriate can
be challenging at times. In particular, if the data do not, in fact, follow a normal distribution,
the tests may fail to reject H0: Data follow a normal distribution. When the sample size is
tiny, as is regrettably frequently the case when we want to use these tests, low power is a
significant problem. The most useful method for determining normalcy is to use a histogram
to examine the distributional form of the outcome in the sample and, if possible, supplement
it with information from other studies that might suggest the outcome's likely distribution in
the population.
3. Null and Alternative Hypothesis
The absence of any apparent effect in our experiment is reflected in the null hypothesis. The
null hypothesis is usually expressed mathematically with an equal sign. H0 represents this
hypothesis.
In our hypothesis test, we look for evidence that refute the null hypothesis. Our goal is to get
a
p-value that is tiny enough to fall below our alpha threshold of significance, which would
allow us to properly reject the null hypothesis. We are unable to reject the null hypothesis if
our p-value is higher than alpha.

It is reflected in the alternative or experimental hypothesis that our experiment will yield
an observed outcome. The alternative hypothesis will usually be expressed mathematically as
an inequality, or not equal to symbol. Either Ha or H1 are used to represent this hypothesis.

Through the use of our hypothesis test, we are seeking to obliquely illustrate the alternative
hypothesis. The alternative hypothesis is accepted if the null hypothesis is disproved. The
alternative hypothesis is not accepted if the null hypothesis is not disproved. The alternative
hypothesis, returning to the mean human body temperature example from above, is, "The
average adult human body temperature is not 98.6 degrees Fahrenheit." The other theory, if
we are researching a novel treatment, is that our intervention actually modifies our subjects in
a significant and quantifiable way.
4. P-value and alpha value
The test's p-value indicates the likelihood of finding an effect in the sample data that is at
least as large as the one we actually saw. Let's say, for instance, that the p-value for the
hypothesis test is 0.02. The following is how to interpret this p-value: 2% of the time we run
this hypothesis test, we would get the effect seen in the sample data, or larger, owing to
random sample error, if there was really no difference between the new pill and the regular
pill.
The threshold we apply in a hypothesis test to assess whether or not our p-value is low
enough to reject the null hypothesis is called the alpha level. Usually set at 0.05, it can
occasionally be as low as 0.01 or as high as 0.10.It's crucial to remember that the likelihood
of mistakenly rejecting a true null hypothesis is likewise defined by the alpha level. When the
alpha level of a hypothesis test is set at 0.05, this implies that approximately 5% of the tests
would correctly reject the null hypothesis if the procedure of conducting the test were to be
done numerous times.
5. Measures of central Tendency
The middle, or average, of a dataset can be found with the aid of measures of central
tendency. The mode, median, and mean are the three most widely used metrics to assess
central tendency.
The most frequent value is the mode.
The median, in an ordered dataset, is the midway value.
The mean is the total number of values divided by the sum of all values.
Descriptive statistics require an understanding of your dataset's variability and distribution in
addition to its central tendency.
6. Measures of dispersion
Dispersion metrics are useful for characterizing data variability. The degree of data scattering
can be expressed statistically using the term dispersion. In order to quantify the dispersion of
data, specific sorts of measures are utilized, and these are known as measures of dispersion.
Positive real values that indicate how homogenous or heterogeneous the provided data is are
known as measures of dispersion. If all of the data points in a set are the same, then the
measure of dispersion has a value of 0. On the other hand, the measures of dispersion
likewise rise in value as the data become more variable.
7. T-test
The t test is a statistical tool used to compare two groups' means. It is frequently employed in
hypothesis testing to ascertain whether two groups are distinct from one another or whether a
procedure or treatment genuinely affects the population of interest.
The t test, like other parametric tests, is a parametric test of difference, which means it makes
the same assumptions about your data. Your data is assumed by the t test. possess
independence
are distributed (roughly) routinely possess homogeneity of variance, or an equivalent level of
variance within each group under comparison.
8.Anova
ANOVA is a statistical technique that's employed to examine variations between two or more
means. As you will see, the name "Analysis of Variance" is appropriate for the approach
because it is used to infer means, even though it may seem strange to call it "Analysis of
Means."
is used to examine overall differences in means as opposed to particular ones. It is best to see
this through example. In the case study "Smiles and Leniency," it was examined how
different grins affected how leniently someone was seen. We looked into four different kinds
of smiles: felt, false, neutral, and unpleasant.
9.Point Biserial, Spearman rho and Pearson’s r Correlations
The Point-Biserial Correlation Coefficient, like all correlation coefficients (such as Pearson's
r and Spearman's rho), measures the degree of association between two variables in a single
measure that ranges from -1 to +1, where +1 denotes a perfect positive association, 0 denotes
no association at all, and -1 indicates a perfect negative association. Since they are all
interdependency measures, correlation coefficients do not express a

Since the dichotomous variable of the two variables is either 0 or 1, it is also known as the
binary variable. The Point-Biserial Correlation Coefficient is computed mathematically in the
same way as Pearson's Bivariate Correlation Coefficient. We do need to meet the same
assumptions, which are homoscedasticity and the continuous variable's normal distribution,
because we are using the same mathematical idea causal relationship. Since the dichotomous
variable of the two variables is either 0 or 1, it is also known as the binary variable. The
Point-Biserial Correlation Coefficient is computed mathematically in the same way as
Pearson's Bivariate Correlation Coefficient. We do need to meet the same assumptions,
which are homoscedasticity and the continuous variable's normal distribution, because we are
using the same mathematical idea.
The Point-Biserial Correlation is not the ideal method for examining data from experiments
because all correlation analyses require the variables to be randomly independent. The
optimum option in these situations is a Linear Regression Analysis with Dummy Variables.
Additionally, a t-test for independent samples or other dependency tests (e.g., Mann-
Whitney-U, Kruskal-Wallis-H, and Chi-Square) can be used to answer many of the issues
that are generally answered with a Point-Biserial Correlation Analysis. Some of these tests
are not only reliable when it comes to requiring regularly distributed variables, but they also
examine the causal or dependent link that exists between the dependent and independent
variables under investigation.

10. Logistic and Linear Regressions

In general, the term "regression" refers to a system's decline toward a state that is less
complex than it was previously. This provides us with a basic understanding of regression,
which defines it as the process of breaking down a system's complexity into a simpler form.
Examining the history of the term—or more accurately, its initial application in statistical
analysis—may yield a second insight. Galton observed that rather than being taller, the
nephews of those who were exceptionally tall tended to be of average height while
researching the height of those families. This gave rise to the hypothesis that, given enough
time, factors like height had a tendency to regress towards the average.

Regression analysis is now considered a legitimate area of statistical analysis. The study of
models that derive simplified relationships from distribution sets is the focus of the discipline
The associations between an outcome variable and at least one explanatory variable are
modeled by linear regression. With the help of this adaptable approach, you may isolate the
significance of each variable and disentangle the implications of intricate research issues.
Linear models can also fit interaction effects and curvature.

In linear regression, the explanatory variables are referred to as independent variables (IV)
and the results as dependent variables (DV) by statisticians. Simple linear regression is the
process used when a linear model has a single IV. several regression is the term statisticians
use to describe situations where there are several IVs. According to these models, the
independent variables' linear functions determine the dependent variable's average value.

ll - Kindly write also your most significant learning/s in our EDF 202 (Advanced
Educational Statistics).
Understanding statistics enables use to gather data in the right ways, do accurate analyses,
and communicate the findings clearly. The process of making scientific discoveries, data-
driven judgments, and predictions relies heavily on statistics. We can have a far deeper
understanding of a subject by using statistics. First of all, statisticians are mentors that help
you learn from data and solve typical problems that may mislead you. Second, it's vital that
you have the ability to evaluate critically the caliber of analyses that others provide to us,
given the increasing significance of data-driven judgments and opinions. In my opinion,
statistics is a fascinating field that involves the excitement of learning, questioning
preconceptions, and making discoveries. It is made easier to create fresh knowledge using
statistics. We gradually advance the boundaries of our knowledge. Read about my
experiences and difficulties early in my career in scientific research to gain further insight
into my passion for statistics as a seasoned statistician. But now my perspective is completely
different. I can sum up my change of heart in two words: big data. I have been amazed over
the past ten years at how society has evolved, both positively and negatively, as a result of the
information we have been unintentionally gathering. I understood that I had to get involved
since this bright new world holds so much promise, power, and danger for everyone. That
was the start of my new life in statistics. We would be lost without these techniques. Without
it, we wouldn't be able to determine what information to gather, what it can tell us, and when
it can't provide any information at all. However, the underlying philosophical underpinnings
of its approaches to world representation need thought, even from us statistic haters. This is
due to the fact that comprehending the crucial difference between the world our data is
supposed to represent and the data we gather depends directly on the efficacy of all those
algorithms that are seeping into every part of our lives. I can now, at last, appreciate the vast
breadth and elegance of the concepts behind statistics. On average, it's better late than never.

Prepared By: Checked By:

MIRASOL R. MALING (Sgd.) WALTER C. NOCASA, EdD