You are on page 1of 9

Unit 2 UNit 5:

A statistical hypothesis is a statement about a population parameter


that is tested using statistical methods. There are two types of
statistical hypotheses: null hypotheses and alternative hypotheses.
The null hypothesis (H0) is a statement that assumes that there is
no significant difference between the observed data and the
expected data. This hypothesis is used as the starting point in
statistical testing, and it is assumed to be true until it is proven
otherwise.
The alternative hypothesis (Ha) is the opposite of the null
hypothesis, and it states that there is a significant difference
between the observed data and the expected data. This hypothesis
is tested against the null hypothesis to determine if it can be
rejected or accepted.
For example, in a study to determine if there is a relationship
between education level and income, the null hypothesis might be
that there is no significant relationship between education level and
income. The alternative hypothesis would be that there is a
significant relationship between education level and income. The
statistical test would be used to determine if the null hypothesis can
be rejected or accepted based on the data collected.
Errors in hypothiesis testing :
Sometimes, our conclusions about an experiment may not match
reality. There can be errors in analysis. The two types of errors in
hypothesis testing are: Type I Error and Type II Error.
Type I Error
A Type I error is when the null hypothesis is true, but we reject it
in favor of the alternate. It is sometimes known as a false positive.
This means we falsely believe that differences before and after an
experiment are due to the treatment when that is not the case.
Type II Error
Type II error- when the null hypothesis is false, but we do not
reject it. It is sometimes known as a false negative. This means we
believe that differences before and after an experiment are only due
to chance and not treatment. This results in wastage of
experimental resources though the results have been achieved.

Level of significance :
The level of significance is defined as the fixed probability of
wrong elimination of null hypothesis when in fact, it is true. The
level of significance is stated to be the probability of type I error
and is preset by the researcher with the outcomes of error. The
level of significance is the measurement of the statistical
significance. It defines whether the null hypothesis is assumed to
be accepted or rejected. It is expected to identify if the result is
statistically significant for the null hypothesis to be false or
rejected.
Level of Significance Symbol
The level of significance is denoted by the Greek symbol α (alpha).
Therefore, the level of significance is defined as follows:
Significance Level = p (type I error) = α
The values or the observations are less likely when they are farther
than the mean. The results are written as “significant at x%”.
Example: The value significant at 5% refers to p-value is less than
0.05 or p < 0.05. Similarly, significant at the 1% means that the p-
value is less than 0.01.
The level of significance is taken at 0.05 or 5%. When the p-value
is low, it means that the recognised values are significantly
different from the population value that was hypothesised in the
beginning. The p-value is said to be more significant if it is as low
as possible. Also, the result would be highly significant if the p-
value is very less. But, most generally, p-values smaller than 0.05
are known as significant, since getting a p-value less than 0.05 is
quite a less practice.

Steps use in testing of hypotheisis :


There are several steps that are typically followed when testing a
hypothesis:
-Clearly define the hypothesis that you want to test. The hypothesis
should be specific and testable.
-Determine the appropriate statistical test to use. There are many
different statistical tests that can be used to test a hypothesis, and
the appropriate test depends on the specific characteristics of the
data and the nature of the hypothesis being tested.
-Collect and analyze the data. This typically involves collecting a
sample of data and using statistical techniques to analyze it.
-Calculate the p-value. The p-value is a measure of the probability
that the results of the statistical test occurred by chance. If the p-
value is below a certain threshold (usually 0.05), the results are
considered statistically significant and the hypothesis is accepted. If
the p-value is above the threshold, the hypothesis is rejected.
-Interpret the results. If the hypothesis is accepted, this suggests
that there is a statistically significant relationship between the
variables being tested. If the hypothesis is rejected, this suggests
that there is not a statistically significant relationship between the
variables.
-Report the results. It is important to report the results of the
hypothesis test in a clear and concise manner, including a
description of the statistical test used, the sample size, the p-value,
and a summary of the results.
Unit 6
Unit 1 : What do you mean by sampling distribution?Why this Explain how sample size affects the margin of error with
Central Limit Theorem : The central limit theorem states that study is important in statistics? A sampling distribution is examples : The margin of error is a measure of the
whenever a random sample of size n is taken from any distribution the distribution of a statistic (such as the mean) computed precision of an estimate. It is the amount by which an
with mean and variance, then the sample mean will be approximately from a sample of a population. It describes how the statistic is estimate may differ from the true value of the parameter
normally distributed with mean and variance. The larger the value of likely to vary across different samples from the same being estimated. The margin of error is usually calculated as
the sample size, the better the approximation to the normal. population. a function of the sample size, the standard deviation of the
Consider x1, x2, x3,……,xn are independent and Studying the sampling distribution of a statistic sample, and the desired level of confidence.
identically distributed with mean μ and finite variance σ2, then any
is important in statistics because it allows us to understand As the sample size increases, the margin of
random variable Zn as,
the behavior of the statistic and to make inferences about the error decreases. This is because a larger sample size
population based on samples from the population. For provides more information about the population, which
example, if we know the sampling distribution of the mean of leads to a more precise estimate of the population parameter.
a certain population, we can use this information to construct For example, consider a survey that aims to estimate the
confidence intervals around the sample mean, which can be proportion of a population that supports a particular
used to make inferences about the population mean. candidate in an election. The survey is conducted by
Then the distribution function of Zn converges to the standard normal
randomly sampling 1000 people from the population. The
distribution function as n increases without any bound.
The sampling distribution of a statistic is often approximated sample shows that 55% of the people support the candidate.
using the central limit theorem, which states that, given a The margin of error for this estimate can be calculated as
sufficiently large sample size, the distribution of the statistic follows:
will be approximately normal, regardless of the shape of the Margin of error = 1.96 * √(0.55 * 0.45 / 1000) = 0.04
distribution of the population. This allows us to use normal This means that we can be 95% confident that
distribution-based methods to make inferences about the the true proportion of people in the population who support
population based on samples from the population. the candidate is somewhere between 51% and 59%.
Studying the sampling distribution of a statistic Now consider the same survey, but with a
is also important in hypothesis testing, where we use the sample size of 100 people instead of 1000. The sample
sampling distribution to determine the likelihood of shows that 52% of the people support the candidate. The
observing a particular sample statistic if a certain hypothesis margin of error for this estimate is now:
Since xi are random independent variables, so Ui are also independent. about the population is true. Margin of error = 1.96 * √(0.52 * 0.48 / 100) = 0.14
This implies, This means that we can be 95% confident that
The Sampling Distribution of the Mean is the mean of the the true proportion of people in the population who support
population from where the items are sampled. If the the candidate is somewhere between 38% and 66%.
population distribution is normal, then the sampling As you can see, the margin of error is larger
distribution of the mean is likely to be normal for the samples when the sample size is smaller. This means that the
of all sizes.Following are the main properties of the sampling estimate is less precise when the sample size is smaller.
distribution of the mean: 1.Its mean is
Explain different criteria of good estimator :There are
equal to the population mean, thus,
several criteria that can be used to evaluate the quality of an
(?X͞ =sample mean and ?p Population
estimator, including:
mean) 2.The population standard deviation divided
Unbiasedness: An unbiased estimator is one that has a mean
by the square root of the sample size is equal to the standard
that is equal to the true value of the parameter being
deviation of the sampling distribution of the mean, thus:
ln(mu(t))=n ln (1 + x)=n(x –x22 + x33− ……..) estimated. An estimator that is biased consistently over- or
Multiply each term by n, and as n → ∞, all terms but the first go to under-estimates the true value of the parameter.
zero. Efficiency: An efficient estimator is one that has a small
(σ = variance, meaning it is less prone to producing extreme or
population standard deviation, n = sample size) 3.The improbable estimates.
sampling distribution of the mean is normally distributed. Consistency: A consistent estimator is one that becomes
This means, the distribution of sample means for a large increasingly accurate as the sample size increases.
sample size is normally distributed irrespective of the shape Sufficiency: A sufficient estimator is one that makes use of
Which is the moment generating function for a standard normal of the universe, but provided the population standard
random variable. all of the information in the sample in order to produce the
deviation (σ) is finite. Generally, the sample size 30 or more estimate.
Example: The record of weights of the female population follows a
is considered large for the statistical purposes. If the Robustness: A robust estimator is one that is not sensitive to
normal distribution. Its mean and standard deviation are 65 kg and 14
population is normal, then the distribution of sample means small deviations from the assumptions that were made in
kg, respectively. If a researcher considers the records of 50 females,
will be normal, irrespective of the sample size. deriving the estimator.
then what would be the standard deviation of the chosen sample?
Solution: Range: The range of an estimator is the set of values that
The Sampling Distribution of Proportion measures the the estimator can take on. An estimator with a narrow range
Mean of the population μ = 65 kg proportion of success, i.e. a chance of occurrence of certain
Standard deviation of the population = 14 kg is preferred because it is less likely to produce extreme or
events, by dividing the number of successes i.e. chances by
sample size n = 50 the sample size ’n’. Thus, the sample proportion is defined implausible estimates.
Standard deviation is given by as p = x/n. Interval estimator: An interval estimator is one that
The sampling distribution of proportion obeys the binomial provides a range of values within which the true value of the
probability law if the random sample of ‘n’ is obtained with parameter is likely to lie. An interval estimator that is
= 14/√50 replacement. Such as, if the population is infinite and the narrow and accurate is preferred.
=14/7.071 probability of occurrence of an event is ‘π’, then the Point estimator: A point estimator is one that provides a
= 1.97 probability of non-occurrence of the event is (1-π). Now single estimate of the value of the parameter. A point
consider all the possible sample size ‘n’ drawn from the estimator that is accurate and has a small variance is
Explain the statement “ The central limit theorem forms the basis population and estimate the proportion ‘p’ of success for preferred.
of inferential statistics ” : The central limit theorem states that, each. Then the mean (?p) and the standard deviation (σp) of
given a sufficiently large sample size from a population with a finite the sampling distribution of proportion can be obtained as: Unit 3:
level of variance, the mean of all samples from the same population
will be approximately normally distributed. This result holds
regardless of the shape of the distribution of the population from Parametric Tests Non-Parametric Tests
which the samples are drawn. ?p = mean of proportion Lower statistical power. Greater statistical power.
In inferential statistics, we are often interested in estimating the π = population proportion which is defined as π = X/N, where
value of a population parameter (such as the mean) based on a sample X is the number of elements that possess a certain Applied to normal or Applied to categorical
from the population. The central limit theorem tells us that, as long as characteristic and N is the total number of items in the interval variables. variables.
the sample size is large enough, the distribution of the sample mean population. Used for large samples. Used for small samples.
will be approximately normal, which means we can use normal σp = standard error of proportion that measures the success
distribution-based methods to make inferences about the population (chance) variations of sample proportions from sample to Its data distribution is The form of data
parameter. sample normal. distribution is not known.
For example, if we want to estimate the mean income of a certain n= sample size, If the sample size is large (n≥30), then the Don’t make many
population, we can take a sample of incomes from that population, sampling distribution of proportion is likely to be normally Make a lot of assumptions.
assumptions.
calculate the mean of the sample, and use the central limit theorem to distributed.
construct a confidence interval around that mean. This interval will The following formula is used when population is finite, and Demand a higher condition Require a lower validity
give us a range of values that we can be confident contains the true the sampling is made without the replacement: of validity. condition.
mean of the population, based on the normal distribution of the sample
mean. Less chance of errors. Higher probability of errors.
So the central limit theorem forms the basis of inferential statistics
because it provides a way to use the sample mean to make inferences The calculation is The calculation is less
about the population mean, by assuming that the sample mean is complicated. complicated.
approximately normally distributed. Hypotheses are based on
The hypotheses are based
ranges, median, and data
on numerical data.
frequency.
What do you mean by non parametric test? write down What is binomial test? why is it used? Describe the function and procedure of chi square test for
advantages of non parametric test over parametric test: The binomial test is a statistical test used to determine goodness of fit. The chi-square test for goodness of fit is a
Nonparametric tests are statistical tests that do not assume a particular whether the probability of success in a binary outcome (such statistical test used to determine whether a sample comes from
shape or distribution for the data. They are generally less powerful as a "success" or "failure") is different from a hypothesized a population with a specific distribution. It is often used to test
than parametric tests, which do make assumptions about the data, but value. It is based on the binomial distribution, which whether a sample is consistent with a hypothesis about the
describes the probability of a specific number of successes in population distribution.
they are more flexible and can be applied to a wider range of data sets.
a fixed number of trials, given a probability of success on To perform the chi-square test for goodness of fit, follow these
There are several advantages of nonparametric tests over
each trial. steps:
parametric tests: 1.Determine the sample size and the number of categories or
The binomial test is often used in situations where there are
1.Nonparametric tests do not require assumptions about the underlying two possible outcomes (e.g., "success" or "failure") and the bins in the data.
distribution of the data, which makes them more robust and less probability of success is known or can be estimated. For 2.Calculate the expected frequency for each category or bin
sensitive to deviations from these assumptions. example, the binomial test might be used to determine based on the hypothesized population distribution.
2.Nonparametric tests are generally easier to understand and interpret, whether the probability of winning a game is different from 3.Calculate the observed frequency for each category or bin
as they do not rely on complex statistical models or assumptions. 50%, or to determine whether the probability of a product based on the sample data.
3.Nonparametric tests can be applied to data that are not normally defect is different from a certain value. 4.Calculate the chi-square statistic using the formula:
distributed, or to data that are ordinal (ranked) rather than continuous. chi-square = sum((observed frequency - expected frequency)^2
4.Nonparametric tests can be used when the sample size is small, as Method of binomial test : / expected frequency)
they do not require large sample sizes to be effective. To conduct a binomial test, the following steps are typically 5.Use a table or computer software to determine the p-value for
5.Nonparametric tests are less sensitive to the presence of outliers or followed: the chi-square statistic.
Define the research question and null hypothesis: As with 6.Compare the p-value to the alpha level (typically 0.05) to
extreme values in the data.
any statistical test, it is important to start by defining the determine whether the sample comes from a population with
6.Nonparametric tests can be used when the data are collected in a the hypothesized distribution.
research question and the corresponding null hypothesis that
way that does not allow for calculation of means and standard you are testing. It is important to note that the chi-square test for goodness of fit
deviations (e.g., if the data are categorical rather than continuous). Determine the hypothesized probability of success: assumes that the sample is a random and representative sample
Determine the hypothesized probability of success that is from the population. If the sample is not representative, the test
Discuss assumptions of non parametric test : being tested against. may not be accurate. Additionally, the chi-square test is
There are generally fewer assumptions associated with nonparametric Collect the data: Collect data on the number of successes sensitive to sample size, and may not be powerful when the
tests than with parametric tests. However, there are still some and failures in a fixed number of trials. sample size is small. In these cases, it may be necessary to use a
assumptions that are commonly made in nonparametric tests: Calculate the test statistic: Use the formula for the binomial different statistical test or to increase the sample size.
Independence: The observations in the sample should be independent test to calculate the test statistic.
of one another. Determine the p-value: Use the test statistic to determine the Discuss the rationale and method of chi square test for
Random sampling: The sample should be drawn randomly from the p-value, which is the probability of observing a result as independence of attributes.
population. extreme or more extreme than the one observed, given that The chi-square test for independence of attributes is a statistical
Equal sample sizes: Nonparametric tests often assume that the sample the null hypothesis is true. test used to determine whether there is a relationship between
sizes in each group being compared are equal. When the sample sizes Interpret the results: Compare the p-value to the chosen two categorical variables. It is often used to test whether there
are not equal, the results of the test may not be reliable. significance level (usually 0.05). If the p-value is less than is an association between the two variables, or whether one
Normality: Some nonparametric tests assume that the underlying the significance level, the null hypothesis can be rejected and variable is independent of the other.
distribution of the data is normal, even though they do not make this the alternative hypothesis can be accepted. If the p-value is To perform the chi-square test for independence of attributes,
assumption explicitly. This is because they are based on the normal greater than the significance level, the null hypothesis cannot follow these steps:
distribution in some way (e.g., the Wilcoxon rank sum test is based on be rejected. 1.Determine the sample size and the number of categories or
the normal distribution of the ranks of the observations). bins in each variable.
Symmetry: Some nonparametric tests assume that the distribution of Discuss process of kolmogorov smirnov process. 2.Construct a contingency table that shows the frequency of
the data is symmetrical. This assumption may be violated if the data The Kolmogorov-Smirnov (K-S) test is a nonparametric each combination of categories or bins in the two variables.
are heavily skewed. statistical test used to determine whether two samples come 3.Calculate the expected frequency for each combination of
Equal variances: Some nonparametric tests assume that the variances from the same population. It is often used when the categories or bins based on the assumption of independence
of the groups being compared are equal. When the variances are not assumptions for parametric tests are not met, such as when between the two variables.
equal, the results of the test may not be reliable. the data is not normally distributed. 4.Calculate the observed frequency for each combination of
It is important to carefully consider these assumptions when choosing To perform the K-S test, follow these steps: categories or bins based on the sample data.
and applying a nonparametric test, as violations of these assumptions 1.Calculate the sample size for each group. 5.Calculate the chi-square statistic using the formula:
can affect the reliability of the test results. 2.Calculate the cumulative distribution functions (CDFs) for chi-square = sum((observed frequency - expected frequency)^2
each sample. The CDF is a function that describes the / expected frequency)
Write down steps of non parametric test : probability that a random 3.variable will be less than or equal 6.Use a table or computer software to determine the p-value for
The steps of a nonparametric test are generally similar to those of a to a certain value. the chi-square statistic.
parametric test, with some key differences: 4.Calculate the maximum absolute difference between the 7.Compare the p-value to the alpha level (typically 0.05) to
Define the research question and null hypothesis: As with any two CDFs. This is known as the K-S statistic. determine whether there is a relationship between the two
statistical test, it is important to start by defining the research question 5.Use a table or computer software to determine the p-value variables.
and the corresponding null hypothesis that you are testing. for the K-S statistic. The chi-square test for independence of attributes is a powerful
Select the appropriate nonparametric test: Based on the research 6.Compare the p-value to the alpha level (typically 0.05) to tool for testing relationships between categorical variables, but
question and the nature of the data, choose the appropriate determine whether the samples come from the same it is sensitive to sample size. When the sample size is small, the
nonparametric test. population. test may not have enough power to detect a relationship
Check the assumptions of the nonparametric test: Make sure that the The K-S test is a powerful tool for comparing two samples, between the variables. In this case, it may be necessary to use a
assumptions of the nonparametric test are satisfied by the data. This but it is sensitive to sample size. When the sample sizes are different statistical test or to increase the sample size. It is also
may include checking for independence, random sampling, equal small, the test may not have enough power to detect a important to note that the chi-square test assumes that the
sample sizes, normality, symmetry, and equal variances, depending on difference between the samples. In this case, it may be sample is a random and representative sample from the
the specific test being used. necessary to use a different statistical test or to increase the population. If the sample is not representative, the test may not
Calculate the test statistic: Use the formula for the chosen sample size. be accurate.
nonparametric test to calculate the test statistic.
Determine the p-value: Use the test statistic to determine the p-value, What do you mean by median test ? Describe procedure Differentiate between chi-square test and KS test for
which is the probability of observing a result as extreme or more of median test. uniformity. The Chi Square test is used to test whether the
extreme than the one observed, given that the null hypothesis is true. The median test is a nonparametric statistical test used to distribution of nominal variables is same or not as well as for other
Interpret the results: Compare the p-value to the chosen significance determine whether two samples come from populations with distribution matches and on the other hand the Kolmogorov Smirnov
level (usually 0.05). If the p-value is less than the significance level, a common median. It is often used when the assumptions for (K-S) test is only used to test to the goodness of fit for a continuous
the null hypothesis can be rejected and the alternative hypothesis can parametric tests are not met, such as when the data is not data. Kolmogorov Smirnov (K-
be accepted. If the p-value is greater than the significance level, the normally distributed. S) test compares the continuous cdf, F(X), of the uniform
null hypothesis cannot be rejected. To perform the median test, follow these steps: distribution to the empirical cdf, SN(x), of the sample of N
1.Determine the sample size for each group.
Advantages and disadvantages of non parametric test : 2.Rank the combined data from lowest to highest value.
Nonparametric tests are statistical tests that do not assume a specific observations. By definitionv,
3.Determine the median value for each group by selecting the
distribution for the data. Here are some advantages and disadvantages If the sample from the random-
value at the middle position of the ranked data for that group.
of nonparametric tests: number generator is R1, R2,......,RN, then the empirical cdf, SN(X),
4.Calculate the difference between the two medians.
Advantages: is defined by
5.Use a table or computer software to determine the p-value
1.Nonparametric tests do not require assumptions about the underlying for the difference between the medians.
distribution of the data, so they can be used with a variety of data As N
6.Compare the p-value to the alpha level (typically 0.05) to becomes larger, SN(X) should become a better approximation to
types. determine whether the difference between the medians is
2.They are often more robust than parametric tests, meaning they can F(X), provided that the null hypothesis is true. The Kolmogorov-
statistically significant. Smirnov test is based on the largest absolute deviation or difference
still be reliable even if the assumptions of the parametric test are not It is important to note that the median test can only be used to
met. between F(x) and SN(X) over the range of the random variable. I.e.
compare the medians of two groups, not the means. If you
Disadvantages: want to compare the means of two groups, you will need to
1.Nonparametric tests may have lower statistical power than use a different statistical test. it is based on the statistic
parametric tests, meaning they may be less able to detect differences The chi-square test uses the
between groups.
2.Nonparametric tests usually have a fewer number of statistical tests
available, so there may be less flexibility in analyzing the data.
3.The interpretation of nonparametric tests can be more difficult than sample statistic
parametric tests because the results are not presented in terms of the Where Oi is the observed number in
familiar normal distribution. the ith class, Ei is the expected number in the ith class, and n is the
number of classes. For the uniform distribution, Ei ist the expected
number in each class is given by: Ei = N/n, N is the total number of
observation.
Describe the function and procedure of mann whitney u test: Discuss the function and procedure for Kruskals Wallis H What do you mean by multiple correlation? Write down the
The Mann-Whitney U test, also known as the Mann-Whitney- test. The Kruskal-Wallis H test is a nonparametric statistical relationship between multiple correlation coefficient and
Wilcoxon (MWW) test or the Wilcoxon rank-sum test, is a hypothesis test used to compare the median of three or more simple correlation coefficient.
nonparametric statistical test used to determine whether two independent samples. It is similar to the one-way analysis of Multiple correlation refers to the correlation between multiple
independent samples come from the same population. It is often used variance (ANOVA), but is more robust and easier to compute. predictor variables and a single outcome variable. It can be
when the assumptions for parametric tests are not met, such as when The test is based on the ranks of the observations within each represented by the multiple correlation coefficient, which is a
the data is not normally distributed or the variances of the two groups group, rather than the observations themselves. measure of the strength and direction of the relationship between
are not equal. The Kruskal-Wallis H test is used to test the null hypothesis the predictor variables and the outcome variable.
To perform the Mann-Whitney U test, follow these steps: that the medians of all groups are equal. If the p-value is less The multiple correlation coefficient is similar to the simple
1.Determine the sample size for each group. than the chosen level of significance (usually 0.05), then the correlation coefficient, which measures the correlation between
2.Combine the data from both groups and rank the values from lowest null hypothesis can be rejected and it can be concluded that two variables. However, the multiple correlation coefficient
to highest. there is a significant difference between at least two of the takes into account the relationships between multiple predictor
3.Calculate the sum of the ranks for each group. These values are groups. variables and the outcome variable, whereas the simple
known as the U statistics. To conduct the Kruskal-Wallis H test, the following steps are correlation coefficient only measures the relationship between
4.Use a table or computer software to determine the p-value for the U typically followed: two variables.
statistics. 1.Rank all of the observations from all of the groups together. In general, the multiple correlation coefficient will be a stronger
5.Compare the p-value to the alpha level (typically 0.05) to determine 2.Calculate the sum of the ranks for each group (Ri). measure of the relationship between the predictor variables and
whether the samples come from the same population. 3.Calculate the test statistic, which is the sum of the squared the outcome variable than any of the individual simple
The Mann-Whitney U test is a powerful tool for comparing two ranks for each group divided by the number of observations in correlation coefficients between the predictor variables and the
samples, but it is sensitive to sample size. When the sample sizes are the group (Hi). outcome variable. This is because the multiple correlation
small, the test may not have enough power to detect a difference 4.Calculate the p-value using the test statistic and the number coefficient takes into account the relationships between all of the
between the samples. In this case, it may be necessary to use a of groups and observations. predictor variables and the outcome variable, whereas the simple
different statistical test or to increase the sample size. 5.Compare the p-value to the chosen level of significance. correlation coefficient only considers the relationship between
two variables.
What is willcoxon matched pair signed rank test ? why is it Differentiate between Kruskals Wallis one way ANOVA Write down the properties of multiple correlation coefficient
superior to sign test? The Wilcoxon matched-pairs signed-rank test is test and Friedman two way ANOVA test. Multiple correlation coefficient is a statistical measure that
a nonparametric statistical hypothesis test used to compare two related The Kruskal-Wallis one-way analysis of variance (ANOVA) quantifies the strength of the relationship between multiple
samples. It is similar to the paired t-test, but is more robust and easier test and the Friedman two-way ANOVA test are both predictor variables and a single criterion variable. Here are some
to compute. The test is based on the ranks of the differences between nonparametric statistical tests that are used to compare the properties of multiple correlation coefficient:
the pairs of observations, rather than the differences themselves. mean ranks of multiple groups on a single dependent variable. Range: The multiple correlation coefficient has a range of -1 to
The Wilcoxon test is superior to the sign test in several ways. First, the However, there are a few key differences between these two 1, where -1 indicates a perfect negative correlation, 0 indicates
Wilcoxon test takes into account the magnitude of the differences tests: no correlation, and 1 indicates a perfect positive correlation.
between the pairs, while the sign test only looks at whether the The number of groups: The Kruskal-Wallis test is used to Interpretation: A multiple correlation coefficient of 0.7, for
differences are positive or negative. This means that the Wilcoxon test compare the mean ranks of two or more groups, while the example, indicates that the predictor variables are able to explain
is more sensitive to differences between the pairs and is less likely to Friedman test is used to compare the mean ranks of three or 70% of the variance in the criterion variable.
produce a Type II error (failing to reject the null hypothesis when it is more groups. Statistical significance: The multiple correlation coefficient can
actually false). The type of independent variable: The Kruskal-Wallis test is be tested for statistical significance to determine whether the
Second, the Wilcoxon test has more statistical power than the sign used when the independent variable is a between-subjects relationship between the predictor variables and the criterion
test, which means that it is more likely to detect a difference between factor (i.e., each subject is only in one group), while the variable is likely to be due to chance or whether it is a real
the pairs when one exists. This is especially important when the Friedman test is used when the independent variable is a relationship.
sample size is small. within-subjects (repeated measures) factor (i.e., each subject is Sensitivity: The multiple correlation coefficient is sensitive to
Finally, the Wilcoxon test is more robust to departures from the in multiple groups). the presence of outliers in the data.
assumptions of the paired t-test (such as normality of the differences), The type of dependent variable: Both the Kruskal-Wallis test Multicollinearity: The multiple correlation coefficient can be
making it a more reliable test in practical situations. and the Friedman test can be used with continuous or ordinal affected by multicollinearity, which is the presence of strong
dependent variables. correlations between predictor variables. In such cases, it may be
Discuss the rationale and method of willcoxon matched pair Post-hoc tests: If the results of the Kruskal-Wallis test are difficult to interpret the individual contributions of the predictor
signed rank test. The Wilcoxon matched-pairs signed-rank test is significant, you can use the Mann-Whitney U test to determine variables to the criterion variable.
used to compare two related samples, where each individual in one which pairs of groups are significantly different from each Linearity: The multiple correlation coefficient assumes a linear
sample is matched with an individual in the other sample. The test is other. If the results of the Friedman test are significant, you relationship between the predictor variables and the criterion
based on the ranks of the differences between the pairs of can use the Nemenyi test or the Conover test to determine variable. If the relationship is non-linear, the multiple correlation
observations, rather than the differences themselves. which pairs of groups are significantly different. coefficient may not accurately represent the strength of the
The rationale for using the Wilcoxon test is that it is a nonparametric Unit 4 : relationship.
test, which means that it makes no assumptions about the underlying What do mean by partial correlation? Write down the Differentiate between multiple and partial correlation
distribution of the data. This is especially useful when the data are not relationship between partial and simple correlation coefficient.
normally distributed or when the sample size is small. coefficients. Multiple correlation coefficient is a measure of the strength and
To conduct the Wilcoxon test, the following steps are typically Partial correlation is a statistical measure that quantifies the direction of the relationship between multiple predictor variables
followed: strength of the relationship between two variables while and a single outcome variable. It takes into account the
1.Calculate the difference between each pair of observations. controlling for (or partialing out) the effects of one or more relationships between all of the predictor variables and the
2.Determine the sign (positive or negative) of each difference. other variables. It can be used to determine whether the outcome variable.
3.Rank the differences, with ties ranked in the middle. relationship between two variables is independent of the On the other hand, partial correlation coefficient is a measure of
4.Calculate the sum of the ranks for the positive differences (W+) and relationships between each of those variables and one or more the strength and direction of the relationship between two
the sum of the ranks for the negative differences (W-). other variables. variables, while controlling for the effects of one or more other
5.Calculate the test statistic, which is the smaller of W+ and W-. Simple correlation, on the other hand, is a measure of the variables. It is used to identify the unique contribution of each
6.Compare the test statistic to the critical value from a table of the strength of the relationship between two variables without predictor variable to the prediction of the outcome variable,
Wilcoxon signed-rank test, or use a software package to calculate the controlling for any other variables. while controlling for the effects of the other predictor variables.
p-value. The relationship between partial and simple correlation For example, consider three predictor variables X1, X2, and X3,
7.If the p-value is less than the chosen level of significance (usually coefficients is that partial correlation allows you to control for and an outcome variable Y. The multiple correlation coefficient
0.05), then the null hypothesis can be rejected and it can be concluded the effects of other variables on the relationship between two would measure the relationship between X1, X2, and X3, and Y,
that there is a significant difference between the two samples. variables, while simple correlation does not. This can be useful taking into account the relationships between all of the predictor
when you want to understand the unique relationship between variables and the outcome variable. The partial correlation
What do you mean by Friedman two way ANOVA test?Describe two variables, rather than the relationship that is influenced by coefficient between X1 and Y, controlling for X2 and X3, would
process of the test. The Friedman two-way analysis of variance other variables. measure the relationship between X1 and Y while controlling for
(ANOVA) is a nonparametric statistical test that is used to compare the effects of X2 and X3 on the relationship. This allows us to
the mean ranks of three or more groups on a single dependent variable. What do you mean by multiple correlation? Write down identify the unique contribution of X1 to the prediction of Y,
It is an extension of the Friedman test, which is used to compare the the relationship between multiple correlation coefficient while controlling for the effects of X2 and X3.
mean ranks of two groups. and simple correlation coefficient. What is multiple regression? Write down the method of
The Friedman two-way ANOVA test is used when the independent Multiple correlation refers to the correlation between multiple obtaining regression line
variable is a within-subjects (repeated measures) factor and the predictor variables and a single outcome variable. It can be Multiple regression is a statistical technique used to predict the
dependent variable is a continuous or ordinal variable. It is used to represented by the multiple correlation coefficient, which is a value of a criterion or dependent variable based on the values
determine whether there are significant differences between the groups measure of the strength and direction of the relationship of two or more predictor or independent variables. It is used to
in the dependent variable. between the predictor variables and the outcome variable. model the relationship between multiple predictor variables
To conduct the Friedman two-way ANOVA test, you would first need The multiple correlation coefficient is similar to the simple and a single criterion variable.
to collect data from a sample of subjects for each group. The data correlation coefficient, which measures the correlation
Multiple regression analysis involves the following steps:
should be collected in a way that allows you to determine the mean between two variables. However, the multiple correlation
ranks for each group. 1.Collect data on the predictor variables and the criterion
coefficient takes into account the relationships between
Next, you would need to calculate the mean ranks for each group and multiple predictor variables and the outcome variable, whereas variable.
compare them using a statistical test. This can be done using a chi- the simple correlation coefficient only measures the 2.Determine the regression model that best fits the data.
square test or a Kruskal-Wallis test. relationship between two variables. 3.Estimate the regression coefficients for the predictor
If the results of the statistical test are significant, you would then need In general, the multiple correlation coefficient will be a variables.
to conduct post-hoc tests to determine which pairs of groups are stronger measure of the relationship between the predictor 4.Test the significance of the regression coefficients.
significantly different from each other. This can be done using the variables and the outcome variable than any of the individual 5.Use the regression equation to predict the value of the
Nemenyi test or the Conover test. simple correlation coefficients between the predictor variables criterion variable for given values of the predictor variables.
It's important to note that the Friedman two-way ANOVA test should and the outcome variable. This is because the multiple The regression equation for a multiple regression model is:
only be used when the assumptions of the test are met. These correlation coefficient takes into account the relationships y = b0 + b1x1 + b2x2 + ... + bn*xn
assumptions include normality of the residuals and homogeneity of between all of the predictor variables and the outcome where y is the criterion variable, b0 is the intercept term, and
variance. If these assumptions are not met, it may be necessary to use variable, whereas the simple correlation coefficient only b1, b2, ..., bn are the regression coefficients for the predictor
a different statistical test. considers the relationship between two variables. variables x1, x2, ..., xn, respectively.
Describe the function and procedure of mann whitney u test: Discuss the function and procedure for Kruskals Wallis H What do you mean by multiple correlation? Write down the
The Mann-Whitney U test, also known as the Mann-Whitney- test. The Kruskal-Wallis H test is a nonparametric statistical relationship between multiple correlation coefficient and
Wilcoxon (MWW) test or the Wilcoxon rank-sum test, is a hypothesis test used to compare the median of three or more simple correlation coefficient.
nonparametric statistical test used to determine whether two independent samples. It is similar to the one-way analysis of Multiple correlation refers to the correlation between multiple
independent samples come from the same population. It is often used variance (ANOVA), but is more robust and easier to compute. predictor variables and a single outcome variable. It can be
when the assumptions for parametric tests are not met, such as when The test is based on the ranks of the observations within each represented by the multiple correlation coefficient, which is a
the data is not normally distributed or the variances of the two groups group, rather than the observations themselves. measure of the strength and direction of the relationship
are not equal. The Kruskal-Wallis H test is used to test the null hypothesis between the predictor variables and the outcome variable.
To perform the Mann-Whitney U test, follow these steps: that the medians of all groups are equal. If the p-value is less The multiple correlation coefficient is similar to the simple
1.Determine the sample size for each group. than the chosen level of significance (usually 0.05), then the correlation coefficient, which measures the correlation between
2.Combine the data from both groups and rank the values from lowest null hypothesis can be rejected and it can be concluded that two variables. However, the multiple correlation coefficient
to highest. there is a significant difference between at least two of the takes into account the relationships between multiple predictor
3.Calculate the sum of the ranks for each group. These values are groups. variables and the outcome variable, whereas the simple
known as the U statistics. To conduct the Kruskal-Wallis H test, the following steps are correlation coefficient only measures the relationship between
4.Use a table or computer software to determine the p-value for the U typically followed: two variables.
statistics. 1.Rank all of the observations from all of the groups together. In general, the multiple correlation coefficient will be a stronger
5.Compare the p-value to the alpha level (typically 0.05) to determine 2.Calculate the sum of the ranks for each group (Ri). measure of the relationship between the predictor variables and
whether the samples come from the same population. 3.Calculate the test statistic, which is the sum of the squared the outcome variable than any of the individual simple
The Mann-Whitney U test is a powerful tool for comparing two ranks for each group divided by the number of observations in correlation coefficients between the predictor variables and the
samples, but it is sensitive to sample size. When the sample sizes are the group (Hi). outcome variable. This is because the multiple correlation
small, the test may not have enough power to detect a difference 4.Calculate the p-value using the test statistic and the number coefficient takes into account the relationships between all of
between the samples. In this case, it may be necessary to use a of groups and observations. the predictor variables and the outcome variable, whereas the
different statistical test or to increase the sample size. 5.Compare the p-value to the chosen level of significance. simple correlation coefficient only considers the relationship
between two variables.
What is willcoxon matched pair signed rank test ? why is it Differentiate between Kruskals Wallis one way ANOVA Write down the properties of multiple correlation
superior to sign test? The Wilcoxon matched-pairs signed-rank test is test and Friedman two way ANOVA test. coefficient
a nonparametric statistical hypothesis test used to compare two related The Kruskal-Wallis one-way analysis of variance (ANOVA) Multiple correlation coefficient is a statistical measure that
samples. It is similar to the paired t-test, but is more robust and easier test and the Friedman two-way ANOVA test are both quantifies the strength of the relationship between multiple
to compute. The test is based on the ranks of the differences between nonparametric statistical tests that are used to compare the predictor variables and a single criterion variable. Here are
the pairs of observations, rather than the differences themselves. mean ranks of multiple groups on a single dependent variable. some properties of multiple correlation coefficient:
The Wilcoxon test is superior to the sign test in several ways. First, the However, there are a few key differences between these two Range: The multiple correlation coefficient has a range of -1 to
Wilcoxon test takes into account the magnitude of the differences tests: 1, where -1 indicates a perfect negative correlation, 0 indicates
between the pairs, while the sign test only looks at whether the The number of groups: The Kruskal-Wallis test is used to no correlation, and 1 indicates a perfect positive correlation.
differences are positive or negative. This means that the Wilcoxon test compare the mean ranks of two or more groups, while the Interpretation: A multiple correlation coefficient of 0.7, for
is more sensitive to differences between the pairs and is less likely to Friedman test is used to compare the mean ranks of three or example, indicates that the predictor variables are able to
produce a Type II error (failing to reject the null hypothesis when it is more groups. explain 70% of the variance in the criterion variable.
actually false). The type of independent variable: The Kruskal-Wallis test is Statistical significance: The multiple correlation coefficient can
Second, the Wilcoxon test has more statistical power than the sign used when the independent variable is a between-subjects be tested for statistical significance to determine whether the
test, which means that it is more likely to detect a difference between factor (i.e., each subject is only in one group), while the relationship between the predictor variables and the criterion
the pairs when one exists. This is especially important when the Friedman test is used when the independent variable is a variable is likely to be due to chance or whether it is a real
sample size is small. within-subjects (repeated measures) factor (i.e., each subject relationship.
Finally, the Wilcoxon test is more robust to departures from the is in multiple groups). Sensitivity: The multiple correlation coefficient is sensitive to
assumptions of the paired t-test (such as normality of the differences), The type of dependent variable: Both the Kruskal-Wallis test the presence of outliers in the data.
making it a more reliable test in practical situations. and the Friedman test can be used with continuous or ordinal Multicollinearity: The multiple correlation coefficient can be
dependent variables. affected by multicollinearity, which is the presence of strong
Discuss the rationale and method of willcoxon matched pair Post-hoc tests: If the results of the Kruskal-Wallis test are correlations between predictor variables. In such cases, it may
signed rank test. The Wilcoxon matched-pairs signed-rank test is significant, you can use the Mann-Whitney U test to be difficult to interpret the individual contributions of the
used to compare two related samples, where each individual in one determine which pairs of groups are significantly different predictor variables to the criterion variable.
sample is matched with an individual in the other sample. The test is from each other. If the results of the Friedman test are Linearity: The multiple correlation coefficient assumes a linear
based on the ranks of the differences between the pairs of significant, you can use the Nemenyi test or the Conover test relationship between the predictor variables and the criterion
observations, rather than the differences themselves. to determine which pairs of groups are significantly different. variable. If the relationship is non-linear, the multiple
The rationale for using the Wilcoxon test is that it is a nonparametric Unit 4 : correlation coefficient may not accurately represent the strength
test, which means that it makes no assumptions about the underlying What do mean by partial correlation? Write down the of the relationship.
distribution of the data. This is especially useful when the data are not relationship between partial and simple correlation Differentiate between multiple and partial correlation
normally distributed or when the sample size is small. coefficients. coefficient.
To conduct the Wilcoxon test, the following steps are typically Partial correlation is a statistical measure that quantifies the Multiple correlation coefficient is a measure of the strength and
followed: strength of the relationship between two variables while direction of the relationship between multiple predictor
1.Calculate the difference between each pair of observations. controlling for (or partialing out) the effects of one or more variables and a single outcome variable. It takes into account
2.Determine the sign (positive or negative) of each difference. other variables. It can be used to determine whether the the relationships between all of the predictor variables and the
3.Rank the differences, with ties ranked in the middle. relationship between two variables is independent of the outcome variable.
4.Calculate the sum of the ranks for the positive differences (W+) and relationships between each of those variables and one or more On the other hand, partial correlation coefficient is a measure of
the sum of the ranks for the negative differences (W-). other variables. the strength and direction of the relationship between two
5.Calculate the test statistic, which is the smaller of W+ and W-. Simple correlation, on the other hand, is a measure of the variables, while controlling for the effects of one or more other
6.Compare the test statistic to the critical value from a table of the strength of the relationship between two variables without variables. It is used to identify the unique contribution of each
Wilcoxon signed-rank test, or use a software package to calculate the controlling for any other variables. predictor variable to the prediction of the outcome variable,
p-value. The relationship between partial and simple correlation while controlling for the effects of the other predictor variables.
7.If the p-value is less than the chosen level of significance (usually coefficients is that partial correlation allows you to control for For example, consider three predictor variables X1, X2, and X3,
0.05), then the null hypothesis can be rejected and it can be concluded the effects of other variables on the relationship between two and an outcome variable Y. The multiple correlation coefficient
that there is a significant difference between the two samples. variables, while simple correlation does not. This can be would measure the relationship between X1, X2, and X3, and
useful when you want to understand the unique relationship Y, taking into account the relationships between all of the
What do you mean by Friedman two way ANOVA test?Describe between two variables, rather than the relationship that is predictor variables and the outcome variable. The partial
process of the test. The Friedman two-way analysis of variance influenced by other variables. correlation coefficient between X1 and Y, controlling for X2
(ANOVA) is a nonparametric statistical test that is used to compare and X3, would measure the relationship between X1 and Y
the mean ranks of three or more groups on a single dependent variable. What do you mean by multiple correlation? Write down while controlling for the effects of X2 and X3 on the
It is an extension of the Friedman test, which is used to compare the the relationship between multiple correlation coefficient relationship. This allows us to identify the unique contribution
mean ranks of two groups. and simple correlation coefficient. of X1 to the prediction of Y, while controlling for the effects of
The Friedman two-way ANOVA test is used when the independent Multiple correlation refers to the correlation between multiple X2 and X3.
variable is a within-subjects (repeated measures) factor and the predictor variables and a single outcome variable. It can be What is multiple regression? Write down the method of
dependent variable is a continuous or ordinal variable. It is used to represented by the multiple correlation coefficient, which is a obtaining regression line
determine whether there are significant differences between the groups measure of the strength and direction of the relationship Multiple regression is a statistical technique used to predict
in the dependent variable. between the predictor variables and the outcome variable. the value of a criterion or dependent variable based on the
To conduct the Friedman two-way ANOVA test, you would first need The multiple correlation coefficient is similar to the simple values of two or more predictor or independent variables. It is
to collect data from a sample of subjects for each group. The data correlation coefficient, which measures the correlation used to model the relationship between multiple predictor
should be collected in a way that allows you to determine the mean between two variables. However, the multiple correlation variables and a single criterion variable.
ranks for each group. coefficient takes into account the relationships between
Multiple regression analysis involves the following steps:
Next, you would need to calculate the mean ranks for each group and multiple predictor variables and the outcome variable,
1.Collect data on the predictor variables and the criterion
compare them using a statistical test. This can be done using a chi- whereas the simple correlation coefficient only measures the
square test or a Kruskal-Wallis test. relationship between two variables. variable.
If the results of the statistical test are significant, you would then need In general, the multiple correlation coefficient will be a 2.Determine the regression model that best fits the data.
to conduct post-hoc tests to determine which pairs of groups are stronger measure of the relationship between the predictor 3.Estimate the regression coefficients for the predictor
significantly different from each other. This can be done using the variables and the outcome variable than any of the individual variables.
Nemenyi test or the Conover test. simple correlation coefficients between the predictor variables 4.Test the significance of the regression coefficients.
It's important to note that the Friedman two-way ANOVA test should and the outcome variable. This is because the multiple 5.Use the regression equation to predict the value of the
only be used when the assumptions of the test are met. These correlation coefficient takes into account the relationships criterion variable for given values of the predictor variables.
assumptions include normality of the residuals and homogeneity of between all of the predictor variables and the outcome The regression equation for a multiple regression model is:
variance. If these assumptions are not met, it may be necessary to use variable, whereas the simple correlation coefficient only y = b0 + b1x1 + b2x2 + ... + bn*xn
a different statistical test. considers the relationship between two variables. where y is the criterion variable, b0 is the intercept term, and
b1, b2, ..., bn are the regression coefficients for the predictor
variables x1, x2, ..., xn, respectively.
What are underlying assumptions of linear regression model?
Linear regression is a statistical method used to model the linear
relationship between a dependent variable and one or more
independent variables. There are several underlying assumptions of
the linear regression model:
Linearity: There is a linear relationship between the dependent
variable and the independent variables.
Independence of errors: The errors (residuals) of the model are
independent of each other.
Homoscedasticity: The errors have constant variance
(homoscedasticity).
Normality of errors: The errors are normally distributed.
Absence of multicollinearity: There is no multicollinearity among the
independent variables.
It is important to check whether these assumptions are met before
fitting a linear regression model to the data, as violating these
assumptions can lead to incorrect model estimates and predictions.

What do you mean by standard error of estimate? Write down the


role of it in regression analysis.
The standard error of estimate is a measure of the variability of the
predicted values around the true value of the criterion variable. It is an
estimate of the standard deviation of the error or difference between
the observed values of the criterion variable and the predicted values.
The standard error of estimate plays an important role in regression
analysis because it helps to determine the accuracy of the predictions
made by the regression model. A low standard error of estimate
indicates that the predicted values are close to the true values, while a
high standard error of estimate indicates that the predicted values are
far from the true values.
The standard error of estimate is also used to construct confidence
intervals around the predicted values. This allows researchers to
estimate the range of values within which the true value of the
criterion variable is likely to fall with a certain level of confidence.
The standard error of estimate can be used to compare the
performance of different regression models and to choose the model
that provides the best prediction accuracy. It can also be used to test
the statistical significance of the regression coefficients and to
determine whether the predictor variables have a significant effect on
the criterion variable.

What do you mean by coefficient of determination? How is it


different from correlation coefficient?
The coefficient of determination, also known as the R-squared or the
squared correlation coefficient, is a measure of the strength of the
relationship between a predictor variable and a criterion variable. It is
a statistical measure that indicates the proportion of the variance in the
criterion variable that is explained by the predictor variable.
The coefficient of determination is calculated as the square of the
Pearson correlation coefficient, which is a measure of the strength and
direction of the linear relationship between two variables. The Pearson
correlation coefficient can range from -1 to 1, where -1 indicates a
strong negative relationship, 0 indicates no relationship, and 1
indicates a strong positive relationship.
The coefficient of determination can range from 0 to 1, where 0
indicates that the predictor variable does not explain any of the
variance in the criterion variable, and 1 indicates that the predictor
variable explains all of the variance in the criterion variable.
The coefficient of determination is different from the Pearson
correlation coefficient in that it represents the proportion of variance
explained, rather than the strength and direction of the relationship. It
is a useful measure of the fit of the regression model because it
provides a single value that summarizes the overall goodness of fit of
the model.

You might also like