AI代写|COMM1110 – Assessment 2b Student Exemplar


This section will explore the statistical and economic significance of the link between hcost and income. To do this, the hcost/lowinc relationship {Rl) will be compared with the hcost/lowinc-/owSEIFA (R2) and hcost/lowinc-/owSEIFA-age-comtime (R3) relationships.

This report’s analysis may be subject to confoundment where a third, uncaptured variable influences the dependent and independent variable(s). Consequently, causality (reverse or forward causality) cannot be assumed. We also cannot assume analysis lacks spurious relationships, or if the direction of causality inferred is 100% correct. Other assumptions are that the sample is normally distributed {Central Limit Theorem); sample estimators are unbiased; and the sample is representative of the population and not overly skewed by outliers.

Lowinc is not the best variable to analyse households’ income levels. Firstly, it does not characterise the sample well enough. A variable called /owQuint which returns 1 to 5 (hence categorising households’ specific income quintiles) is more useful than /owinc’s binary restrictions. Secondly, /owinc assumes individuals characterised as ‘low-income’ must be in the bottom two income quintiles when ‘low-income’ individuals (in the context of mortgages)could actually be in the bottomfourquintiles.

To analyse the statistical significance of Rl (hcost/lowinc), assume: the null hypothesis Ho:

B1=0; rejection criteria as H1: j31;c0; and a<0.05 as definitive of our rejection zone.

As Exhibit 1 shows, Rl returns a p-value of 4.812e11 -71<0.05 (the chance of our null hypothesis being satisfied is <0.05%). This is supported by the fact the 95% confidence interval (0.1066, 0.1316) for lowinc does not contain the value O (i.e. 95% chance lo wine does NOT have ‘zero’ impact on hcost). Hence, rejecting Ho in favour of H1, it is posited that the link between hcost and lowinc is statistically significant.

Taking similar approaches for R2 (hcost/lowinc-/owSEIFA) and R3 (hcost/lowinc-/owSEIFA-age-comtime), this section will now compare the regressions. Summary tables, however, shall be used hereon for clarity (see Appendix for relevant regressions).

Rl is the most statistically significant due to having the smallest P-value (Table 1) when com pared to R2 and R3. Indeed, the overall statistical significance of R3 is questionable since its p-values and confidence intervals support the null hypothesis. For example, age and comtime return p-values>0.05 (Table 1) and include ‘O’ in their confidence intervals (Table 3),meaning the nulls cannot be rejected with 95% confidence. Consequently, age and comtime are likely statistically insignificant links to hcost.