If your dependent variable is sales channel selection and your independent variable is channel attribute such as time, ease of use, price, etc., then you can use a linear regression model to analyze the relationship between the independent variable (channel attribute) and the dependent variable (sales channel selection).

For ease of use as an independent variable, you may use Likert scale data as input, with a linear regression, which is a common approach. Likert scales are ordinal data, which means that the responses are ranked in order of magnitude, but the difference between the ranks is not necessarily equal. To handle ordinal data, Linear regression can be used, but it is important to make sure that the assumptions of linear regression are met: linearity, homoscedasticity, normality, and independence.

In interpreting the results of your linear regression model, you can examine the coefficients of the independent variable(s) to determine the strength and direction of the relationship between the independent variable(s) and the dependent variable. A positive coefficient indicates a positive correlation between the independent variable and the dependent variable, while a negative coefficient indicates a negative correlation. The magnitude of the coefficient will indicate the strength of the relationship.

If you find that the assumptions of linear regression are not met, you can consider using an ordinal logistic regression model, which is a variation of logistic regression model specifically designed for ordinal data.

In addition, you could use visualization tools such as histograms, box plots, scatter plots, etc. to explore the data further, and to ensure that the assumptions of linear regression are met.

Hi Shezan,

Thanks for the elaborate and prompt response.

Following up on this, I have two more questions,

1. Is it required to recode the likert scale values to numerical value (-2 to +2) before applying regression?

What would be the difference in interpretation between the two?

2. how to interpret the below,

Estimate Std. Error t value Pr(>|t|)

(Intercept) 0.37931 0.08327 4.555 1.18e-05 ***

EaseLow priority 0.62069 0.15391 4.033 9.27e-05 ***

EaseModerate priority 0.34618 0.10429 3.320 0.00117 **

EaseNeutral 0.28736 0.11188 2.568 0.01133 *

EaseSomewhat a priority 0.39847 0.17109 2.329 0.02138 *

According to this the customers have ranked Ease of use as one of the factors that influence channel selection. As per above result, EaseLow Priority and Ease Moderate priority have a significant impact. But, it is a bit confusing to make sense of this interpretation.

It is not strictly required to recode the Likert scale values to numerical values (-2 to 2) before applying regression, but it can be useful in some cases. For example, if you want to use a linear regression model, it may be more appropriate to use numerical values rather than ordinal categories. The interpretation of the results will also be different. When using ordinal categories, the coefficients represent the change in the dependent variable associated with a one-unit change in the independent variable, whereas when using numerical values, the coefficients represent the change in the dependent variable associated with a one-unit change in the independent variable.

The table you provided shows the results of a linear regression analysis where ease of use is the independent variable and channel selection is the dependent variable. The dependent variable is modeled as a linear function of the independent variable, with an intercept and a coefficient for each level of the independent variable.

The intercept (0.37931) is the expected value of the dependent variable when all independent variables are equal to zero.

The coefficients for EaseLow priority, EaseModerate priority, EaseNeutral, and EaseSomewhat a priority represent the change in the dependent variable associated with a one-unit change in the independent variable, holding all other independent variables constant.

The t-value represents the ratio of the coefficient to its standard error. The larger the t-value, the more significant the coefficient. The p-value is used to determine the significance of the t-value. A small p-value (typically less than 0.05) indicates that the coefficient is significantly different from zero and that the independent variable has a significant effect on the dependent variable.

In this case, all independent variables have a positive coefficient and a p-value less than 0.05, which suggests that ease of use is positively associated with channel selection. Additionally, the coefficients for EaseLow priority and EaseModerate priority are larger than the others, which suggests that these levels of ease of use have a stronger positive effect on channel selection than the other levels.