2) Explain why using the smaller of $n_{1}-1$ or $n_{2}-1$ degrees of freedom to determine the critical $t$ instead of $d f=\frac{\left(\frac{s_{1}^{2}} {n_{1}}+\frac{s_{2}}{n_{2}} ight)^{2}}{\left(\frac{\frac{3_{2}}{n_{1}}} {n_{1}-1} ight)^{2}\left(\frac{s_{2}}{n_{2}} ight)^{2}}{n_{2}-1}$ is more cautious (in terms of making a Type I Error). SP.JG. 188