# Learn How to Detect and Handle with Multicollinearity in SPSS

 We provide SPSS Help for Students, at any level! Get professional graphs, tables, syntax, and fully completed SPSS projects, with meaningful interpretations and write up, in APA or any format you prefer. Whether it is for a Statistics class, Business Stats class, a Thesis or Dissertation, you'll find what you are looking for with us Our service is convenient and confidential. You will get excellent quality SPSS help for you. Our rate starts at $35/hour. Free quote in hours. Quick turnaround! RETURN SALES MARGIN DTOC • BANKS: 1 if the sector is “Banks”, 0 elsewhere. • COMPUTER: 1 if the sector is “Computers”, 0 elsewhere. • CONSTRUCT: 1 if the sector is “Construction”, 0 elsewhere. $\begin{array}{cc} & RETURN={{\beta }_{0}}+{{\beta }_{1}}SALES+{{\beta }_{2}}MARGIN+{{\beta }_{3}}DTOC+{{\beta }_{4}}BANK \\ & \text{ }+{{\beta }_{5}}COMPUTER+{{\beta }_{6}}CONSTRUCT+\varepsilon \\ \end{array}$ • We first observe that there are no signs of serious multicollinearity ($$VIF’s < 4$$). Also, we see that the overall model is significant ($$p = 0.000$$). • We also observe that all the variables are significant, except for BANKS and SALES. • Also, for the different categories, the higher the$\beta \$coefficient, the higher the impact on the RETURN variable. The effects are measured with respect to the “Energy” category. Using that criteria, we have that the descending ranking of the sectors would be
1. Computers
2. Construction
3. Banks
4. Energy
$\text{ }{{H}_{0}} :{{\mu }_{BANK}}={{\mu }_{COMPUTERS}}={{\mu }_{CONSTRUCTION}}={{\mu }_{ENERGY}}$

### Results from Levene's Test

• The Levene Test shows that we reject the null hypothesis of equal variances.

$\begin{array}{cc} & {{H}_{0}}:{{\beta }_{4}}={{\beta }_{5}}={{\beta }_{6}}=0 \\ & {{H}_{A}}:\text{ at least on of those }\beta 's\ \text{is not zero} \\ \end{array}$
• Banks:

• Computers:

• Construction:

• Energy:

### Part 2: How to Deal with An Anomalous Data Set in SPSS

• Just by seeing the graph we notice that there’s a very clear linear correlation between the two independent variables. This indicates that most likely we’ll find multicollinearity problems.

• At first sight it looks like a significant model, with a very high R-square, but there’s a clear multicollinearity problem (VIF’s = 185.529). This means that essentially $${{X}_{1}}$$ is a linear function of $${{X}_{2}}$$.

• The correlation coefficient is not significantly different from zero, and therefore there is not enough evidence of linear association

• The correlation coefficient is not significantly different from zero either, and therefore there is not enough evidence of linear association.