MR. ASGAR M.

ANDA
SE 413 Advanced Educational Statistics
Take Home Exam

I. Conceptual

1) Why is it very important to know the nature of the data the researcher
collected?

Data is any set of characters that is gathered and translated for some purpose,

usually analysis. In research, it is any information that has been collected, observed,

generated or created to validate original research findings. Data is essential for

verification, validation, and assessment of models for predicting the long-term structural

durability and performance of materials in extreme environments. Many models would

be useless if there was insufficient data to validate and assess them.

It is important to understand nature of data because data may be collected in a

variety of ways. We may gather information through an interview, a questionnaire, an

observation, an experiment, or existing data. The level of data measurement is also

different. It might be in the form of a nominal, ordinal, interval, or ratio. Each level has its

own set of features that influence the sort of analysis that may be done. Furthermore, if

we understand the nature of the data, we may obtain reliable results. Maintaining the

integrity of research requires accurate data. If the data is obtained incorrectly, it might

lead to researchers following futile paths of exploration and distorted findings resulting

in wasted resources.

Data is used as the raw material in research to reach conclusion about a topic.

It is necessary to understand the nature of data since it is a type of information that is

crucial for making constructive decisions. Similarly, researchers have depended mostly

on data to help them make decision about topic. Today, collecting data has become a

most priority for researchers in order to better understand their research papers.
2) For what purpose the assumptions are made or imposed for some statistical
tests such as the parametric?

All parametric tests in statistical analysis presume that the data has some

certain characteristics. A set of assumptions underpins the majority of the statistical

tests we perform. The outcome of the study and the interpretation of the findings are

altered if certain assumptions are violated. Those parametric assumptions are important

for a variety of reasons. They place constraints on how to interpret the results of the

data. For instance, If we have normality and homoscedasticity and get a significant

result, the only logical interpretation of a null hypothesis rejection is that the population

means vary. In addition, the assumptions include that we will make conclusions from the

samples based on the characteristics of the populations from which we sample. We

know a lot about our sampled populations since we assume normalcy and

homoscedasticity, and we may draw conclusions based on what we know.

It is important to consider testing the assumptions that go along with our

analysis as we prepare to do our statistics. The assumption testing of our selected

analysis helps us to see if we can accurately draw inferences from our findings.

Furthermore, we might consider assumptions to be the conditions that must be met

before we can proceed with our investigation. Testing assumptions for the application of

parametric tests may appear time-consuming, yet it is a necessary step in data analysis.

As a result, any study, whether for a journal article, thesis, or dissertation, must adhere

to these assumptions for appropriate interpretation. The assumptions change based on

the parametric analysis.

3) For what purpose the nonparametric statistics are being developed?

Nonparametric statistics are sometimes referred to as distribution-free statistics

because they are not reliant on a specific sort of population distribution, such as a

normal distribution. The nonparametric statistics are widely considered to involve fewer

or less severe assumptions about the nature of the population distribution being

analyzed. It is commonly used for researching populations that take on a ranked order,

and it requires due diligence on its limitations, strengths, and potential pitfalls.

Nonparametric statistics were developed to assist researchers in finding

alternate data analysis techniques in many situations when the parametric statistics are

not applicable. Nonetheless, because of the considerably low degree of confidence

acquired from the preceding statistics, we do not always employ nonparametric

statistics as a substitute technique if parametric assumptions are not satisfied. When

the data distribution is not normal, nonparametric statistics are used. Its usage is also

appropriate when we do not know or have trouble determining the distribution of the

population. Without understanding the underlying distribution, we may confidently

extrapolate from the data.

Nonparametric statistics have several advantages, including the ability to be

applied to a wide range of circumstances, the ability to work with lower sample sizes,

and the ability to work with different types of data. Nonparametric statistics have been

developed as an alternative to parametric tests such as the T-test or ANOVA, which can

only be used provided the underlying data meets particular requirements and

assumptions. Nonparametric statistics are popular because they are simple to apply.

The data may be used to a wider range of tests since the parameters are not required,
4) What are the effects of sample sizes on the analysis of the data?

The number of people included in a research study to reflect a population is

referred to as sample size in research. The sample size is an important factor to

consider when conducting research. For a variety of reasons, obtaining a sample size

that is acceptable in both respects is crucial. A high sample size is also more typical of

the population, reducing the impact of outliers or extreme data. In order to get findings

among variables that are significantly different, a sufficiently large sample size is also

required. Larger sample sizes provide more accurate values.

The data analysis may be influenced by sample sizes. A study with an

insufficient sample size may generate inconclusive results and may also be unethical,

because exposing human subjects to the possible risks of research is only justified if

there is a reasonable possibility that the study will yield meaningful information.

Similarly, a study with an excessively large sample size will waste resources and

expose more people than necessary to any associated risk. As a result, determining the

right sample size for a study is an important part of the research design process.

Sample sizes should not be excessively large or small, as both have drawbacks that

could compromise the conclusions of the study. A limited sample size may hinder

extrapolation of findings, whereas a large sample size may increase the detection of

differences, emphasizing statistical differences that are not clinically relevant. Our

estimate is more precise since we have more data and consequently more information.

As our sample size increases, we gain more confidence in our estimate, our uncertainty

decreases, and we gain more precision.

5) For what purpose the correction of continuity being impose on some
nonparametric statistics?

When a continuous distribution approaches a discrete distribution, it is

necessary to provide a correction for continuity. When a normal distribution approaches

the binomial distribution, it's called a binomial distribution approximation. It was

introduced to increase the accuracy of some nonparametric statistics. Because the

normal distribution accepts all real numbers and the binomial distribution only accepts

integers, a continuity correction factor is required to improve accuracy. The continuity

correction is used to a discrete distribution in order to approximate the probability of a

certain value with a normal distribution. As a result, for small sample sizes in some

nonparametric statistics, the continuity correction factor provides a better answer. The

chi square test continuity correction, often known as Yates' correction for continuity, is

one of the continuity corrections used in nonparametric statistics. Yate’s Correction was

created to be used in a specific case while testing for independence in a contingency

table. It has the effect of preventing overestimation of statistical significance for small

data sets. Generally, we use continuity correction while estimating the binomial using

the normal distribution. While the original distribution was discrete, the normal

distribution is continuous. Continuity correction has a higher approximation in some

nonparametric statistics.
6) Compare and contrast parametric and nonparametric statistics.

The major distinction is that parametric statistics are based on assumptions

about the population distribution from which the sample was taken, but nonparametric

statistics are not. This means that data can be acquired from a sample that does not

follow a specified distribution. Parametric statistics refers to a statistics and statistical

methods that deal with data with a known probability distribution or that have the

parameters of a distribution stated. Nonparametric statistics, on the other hand, is a

statistical method in which the data is not expected to come from predetermined models

with a small number of parameters.

The information about the population distribution is known and is based on a

defined set of parameters in parametric statistics. In nonparametric statistics, the

information about a population distribution is uncertain and the parameters are not

defined. We should examine various criteria concerning the sample data and

assumptions when deciding whether to employ parametric or nonparametric statistics,

and carefully analyze the validity of those assumptions. Parametric tests rely on

statistical distributions in data, but nonparametric tests don't. Nonparametric analysis

makes no assumptions and uses the median value to measure the central tendency. In

conclusion, parametric and nonparametric tests are both important components of data

analysis. A parametric or nonparametric test is applied depending on whether the data

is regularly distributed or not.