- CFA Exams
- CFA Level I Exam
- Study Session 2. Quantitative Methods (1)
- Reading 4. Introduction to Linear Regression
- Subject 5. Testing the significance of a regression coefficient

###
**CFA Practice Question**

In comparing the relation of company sales with GDP, your regression software provides the following output:

Is it possible that the regression coefficient is insignificant at the 95% level of confidence?

A. No since the t-statistic of 4.86 greatly exceeds the critical values of the insignificance acceptance range.

B. No, since the t-statistic is within the critical values of the insignificance acceptance range.

C. Yes, since the t-statistic is not large enough to be classified as insignificant.

**Explanation:**t-start = 0.85/0.175 = 4.86.

t-critical at df = 89 and α = 0.025 is 2.0.

Note that with a large sample size (over 30), we can begin to rely on the more familiar z-statistics.

###
**User Contributed Comments**
5

User |
Comment |
---|---|

zwer |
Are we supposed to derive critical values ourselves during the exam? |

volkovv |
Unlikely (but we may). I think it is worthwhile to remember main critical values for 90%, 95%, 99% and the trends in values, such as critical values decrease when degrees of freedom increase |

mcspaddj |
The critical value for T in this case is the (coefficient - zero)/ standard error. My problem with this question is that there is always a theoretical possiblity of the variable being insignificant. Even if we had a t value approaching infinity, there will still be a possibility (albeit extremely small) chance of an error in rejecting/failing to reject the null hypothesis. |

dblueroom |
mcspaddj, we don't have to worry about that, leave that to statisticians. |

ramdabom |
I believe we could have used F-statistic as well |