The weak positive correlation between satisfaction and salary (r = 0.1564) suggests that higher salaries might contribute to greater satisfaction with the program . This relationship could imply that financial rewards enhance perceived program value or just personal contentment with career outcomes. However, the weak correlation also underscores that satisfaction is likely influenced by other factors beyond salary, such as personal growth, learning experience, and networking opportunities. Therefore, while salary contributes to some extent to satisfaction, a holistic view should be taken to understand what drives satisfaction among graduates .

The weak negative correlation between work experience and GMAT scores (r = -0.1737) suggests that students with more workplace experience tend to have slightly lower scores on the GMAT . This might be due to experienced individuals focusing less on test preparation due to time constraints or prioritizing practical experience over standardized testing skills. It may also indicate that professionals might rely on their experience to compensate for lower GMAT scores during MBA applications . Such a trend reflects a trade-off between work experience and standardized test preparation.

The correlation analysis reveals a weak positive correlation between work experience and Spring GPA (r = 0.1591) and a weak negative correlation between work experience and GMAT scores (r = -0.1737). This suggests that individuals with more work experience might perform slightly better academically during their program, indicated by the positive correlation with Spring GPA. On the other hand, the negative correlation with GMAT scores implies that experienced students possibly score lower on standardized tests but might leverage their experience for better academic performance later in the program. This infers a nuanced view of how prior professional experience affects educational outcomes .

Using linear regression in this context has limitations, as indicated by the poor predictive power of the model. The key limitations include an Adjusted R² of -0.0014, signaling that the model explains virtually none of the variance in salary . Additionally, p-values for GPA, gender, and work experience are all greater than 0.05, showing that these variables do not have a statistically significant impact on salary. This suggests that linear regression with these predictors may introduce unnecessary complexity without meaningful insight, likely due to the omission of other influential factors such as industry trends, negotiation skills, job location, or networking .

The data cleaning process involves excluding rows where selected numerical columns contain the values 999 or 998 . This exclusion is critical as these values likely represent placeholders for errors or missing data, which could skew the analysis if not removed. By ensuring only valid data points are included, the analysis can be more accurately interpreted without the distortion from erroneous entries .

The explored regression model does not directly address multicollinearity, but signs of its absence can be inferred from the poor predictive power and insignificant p-values associated with the predictors (GPA, gender, and work experience). Multicollinearity often affects the model's ability to assign appropriate significance to predictors, but in this case, the Adjusted R² is negative, suggesting that none of the variables contribute meaningfully to the explanation of salary variance in isolation or in combination. A further investigation using statistical diagnostics like the Variance Inflation Factor (VIF) would be necessary to conclusively identify multicollinearity .

The findings suggest that GPA, gender, and work experience might not predict starting salaries significantly due to various external factors that play a more dominant role. The regression model highlights an Adjusted R² of -0.0014 and non-significant p-values for these variables , indicating poor predictive value. This could be because salary outcomes are more influenced by market conditions, demand in specific industries, negotiation skills, and networking capabilities rather than just academic or demographic backgrounds. The unseen variables likely mask the impact of traditional metrics, highlighting a disconnect between academic performance and actual financial outcomes in the job market .

The exclusion technique improves the robustness of data analysis by removing rows with placeholder or erroneous values (999, 998) that don't represent real data. By eliminating these anomalies, the dataset becomes cleaner, and analyses such as correlation or regression are more reliable, as they are not impacted by outliers or incorrect entries that could distort results . This cleaning ensures that conclusions drawn are based on accurate reflections of the data, thereby enhancing the reliability of any insights related to salary expectations .

The absence of a significant correlation between academic metrics like GMAT and GPA with salary suggests that traditional assumptions in MBA education—such as the belief that higher academic scores directly result in higher pay—may not hold true . This revelation indicates that salary is likely influenced by other factors outside academia, undermining the predictive power of such metrics. This challenges educators to reconsider the weight placed on standardized testing and grades when defining what prepares students effectively for the workforce, acknowledging that skills like networking, personal initiatives, and industry-specific knowledge may play more pivotal roles .

The correlation analysis challenges the traditional assumptions by showing that academic metrics like GMAT scores, GPA, and work experience have minimal impact on starting salaries . Notably, there is no significant correlation between GMAT scores or GPA with salary, which contradicts the expectation that higher academic achievements lead to better pay. Furthermore, while there is a weak positive correlation between satisfaction and salary, it suggests satisfaction is somewhat influenced by salary levels . This indicates that other non-academic factors might be more critical for salary outcomes.