2.1.982 Part 1 Section 18.104.22.168, LINEST
a. The standard states that x and y are sample means.
Office treats x and y as the sample values, and x-bar and y-bar as the sample means of known-xs and known-ys, respectively.
b. The standard does not place any restrictions on dimensions of known-ys or known-xs.
In Office, if known-ys and known-xs are both specified but don't match in size (or number of variables for the multivariate case), #REF! is returned.
c. The standard describes statistic seb as the standard error value for the constant b.
In Office, if b is forced to be zero (const-flag=FALSE), then it has no standard error and #N/A is returned.
d. The standard does not mention collinearity cases for this function.
In Excel, vectors in known-xs that are collinear (have no additional predictive value because they can be expressed as sum of multiples of existing vectors) are omitted from the regression calculations. The vectors are given a coefficient of 0 and a standard error of 0 (m = 0; se=0). If vectors are omitted, df is affected because df depends on the number of values used for predictive purposes.
e. The standard states the syntax is: LINEST ( known-ys [ , [ known-xs ] [ , [ const-flag ] [ , stats-flag ] ] ).
Office uses the following syntax: LINEST ( known-ys [ , [ known-xs ] [ , [ const-flag ] [ , [ stats-flag ] ] ] ] ).