1

HOMEWORK 7
Homework 7

The chi-square test is a statistical test commonly used to compare observed data with

data we would expect to obtain according to a specific hypothesis (Franke et al., 2012). It is

used to determine if any difference between the observed and expected results is statistically

significant or could reasonably be explained by chance alone.

The chi-square test is typically used when we want to compare observed frequencies

or counts across different categorical variables. For instance, one may use a chi-square test to

analyze survey results to see if the proportion of different responses varies significantly

between men and women (Franke et al., 2012). One could also use it to compare the

frequency of different medical diagnoses across age groups to see if age is a relevant factor.

To perform the chi-square test, we first develop a hypothesis about what the expected

frequencies would be under the condition that the variables are independent (Franke et al.,

2012). For instance, if we thought gender didn't affect survey responses, we might expect the

proportions of each response to be the same between men and women. We then collect

observational data and calculate a chi-square statistic based on the differences between the

expected and observed frequencies across each category.

The chi-square value is then compared to critical values from the chi-square

distribution based on the desired significance level and degrees of freedom. If the chi-square

value exceeds the critical value, we can reject the hypothesis that any differences are due to

chance and determine there is a statistically significant association between the categorical

variables being studied (Franke et al., 2012). This helps assess whether patterns in frequency

data are meaningful or could plausibly be attributed to random variation.

The purpose of linear regression analysis is to model the relationship between two or

more variables so that we can predict the value of a dependent variable based on the value of

an independent variable.
2
HOMEWORK 7
Specifically, linear regression finds the best-fitting straight line through data points in

a scatterplot to demonstrate how the dependent variable changes with the independent

variable. The slope of the line indicates how much the dependent variable changes when the

independent variable is increased by one unit, on average.

Linear regression can be used to address a wide variety of research questions where

we want to understand how changes in one variable relate to or predict changes in another.

For example, a researcher could use linear regression to investigate the relationship between

average class size (independent variable) and standardized test scores (dependent variable)

across different school districts.

The research question could be: Does average class size significantly predict

standardized test scores, when controlling for other school characteristics? The researcher

would collect data on average class sizes and test scores for multiple districts, as well as

control variables like funding levels, demographic factors, etc. Then a linear regression model

could evaluate whether smaller class sizes have a statistically significant relationship with

higher test scores.

The results could help determine if class size reduction should be recommended as a

policy intervention to improve educational outcomes. So in summary, linear regression

allows researchers to quantify relationships between variables and draw conclusions about

prediction and causality.

Linear regression has some key assumptions that must be met to ensure the validity of

results:

Linear relationship assumption - There is a linear (straight-line) relationship between the

independent and dependent variables (Osborne & Waters, 2019). To check this, researchers

can examine scatter plots and ensure the data lies in a reasonably straight line.
3
HOMEWORK 7
Independence of errors assumption - The errors or residuals (differences between observed

and predicted values) must be independent of each other (Osborne & Waters, 2019). Violating

this assumption can result in an overestimation of significance. Researchers can check for

autocorrelation of errors.

Homoscedasticity assumption - The variance of errors is consistent across all levels of the

independent variable. Heteroscedasticity violates this assumption and impacts significance

tests (Flatt & Jacobs, 2019). Researchers can check residual plots.

Normality assumption - The errors are normally distributed about the mean point on the

regression line. Non-normal errors can skew significance tests (Flatt & Jacobs, 2019).

Researchers can check Q-Q plots or skewness/kurtosis of residuals.

No perfect multicollinearity - The independent variables are not perfectly linearly related to

each other. Otherwise, it is difficult to separate their contribution to prediction. Researchers

can check variance inflation factor (VIF) values.

An example of reporting regression results from the article would be as follows:

"Multiple regression was conducted to predict perceived customer relationship quality from

electronic direct mail, interpersonal communication, preferential treatment, and perceived

reward. A significant regression equation was found (F(4,114) = 60.538, p < 0.001) with an

R2 of 0.680. Perceived reward (β = 0.460, p < 0.001) and preferential treatment (β = 0.447, p

< 0.001) were significant predictors of perceived customer relationship quality (Lam et al.,

2013)." This reports the key output values of the regression model including the F-statistic,

R-squared value, individual predictor beta coefficients, and their significance levels.
4
HOMEWORK 7
References

Flatt, C., & Jacobs, R. L. (2019). Principle assumptions of regression analysis: Testing,

techniques, and statistical reporting of imperfect data sets. Advances in Developing

Human Resources, 21(4), 484-502.

Franke, T. M., Ho, T., & Christie, C. A. (2012). The chi-square test: Often used and more

often misinterpreted. American journal of evaluation, 33(3), 448-458.

Lam, A. Y., Cheung, R., & Lau, M. M. (2013). The influence of customer relationship

management on customer loyalty. Contemporary management research, 9(4).

Osborne, J. W., & Waters, E. (2019). Four assumptions of multiple regression that researchers

should always test. Practical assessment, research, and evaluation, 8(1), 2.