Statistical functions in Grid by Zaweba, explained

Averages are the easy part of statistics. The harder questions come straight after: how spread out are these figures, is that difference real, where is this trend heading. Grid by Zaweba has 106 statistical functions for those questions, enough for a school project, a quality log at work or a look through a year of sales.

The family is large, so it helps to think of it in a few groups.

Describing a set of numbers

Start with the averages and counts, then the measures of spread. Standard deviation and variance tell you how far values typically sit from the middle. Percentiles and quartiles tell you where a value falls in the pack, and the ranking functions give each value its place in the order. Correlation says how closely two columns move together.

A good pattern is a small summary block beside your data: count, average, median, standard deviation, minimum and maximum, each in its own labelled cell. When new rows arrive the block updates, and you can spot at a glance when something has shifted.

Distributions and significance tests

Grid includes the standard probability distributions and their inverses: the normal distribution, Student's t, chi squared, F, gamma, binomial, Poisson and others, including the one usually used to model proportions and rates that sit between nought and one. Each lets you ask for the probability of a value, the cumulative probability up to it, or, through the inverse, the value that matches a given probability.

The significance tests sit alongside them. With these you can check whether two groups' results differ by more than chance would explain, or whether observed counts fit what you expected. As with any statistics tool, the function does the arithmetic; deciding which test suits your data is still your call, so read each description in Insert Function before you rely on it.

Trends and forecasts

LINEST fits a straight line through your data and returns its slope, intercept and related statistics. LOGEST does the same for growth that compounds. TREND and FORECAST use a fitted line to predict values for new points, which is the quick way to project next quarter's figures from the last few.

LINEST returns more than one value, and Grid's dynamic arrays let it spill those values into the cells next to the formula. Type it once, leave space to the right and below, and the results appear. Label them, because a bare block of numbers is easy to misread a week later.

Treat a forecast as a guide rather than a promise. A straight line through twelve months of sales says nothing about a new competitor or a hot summer. Pair the numbers with a scatter or line chart so you can see whether a straight line is a fair description of the data in the first place.

Older names still work

Spreadsheets have renamed many statistical functions over the years, and older workbooks are full of the original names. Grid keeps 39 of them as compatibility functions, including STDEV, NORMDIST, TDIST, PERCENTILE and RANK. A workbook built years ago opens and calculates without you rewriting its formulas.

For new work, prefer the modern names. They say clearly which version of a calculation you mean, a sample or a whole population, one tail or two, and anyone reading your sheet later will understand it faster.

Avoiding the usual slips

Check whether your data is a sample or the whole population before choosing between the two kinds of standard deviation. Watch for text that looks like numbers, such as figures pasted in from a website, because statistical functions skip text and a count that comes up short is often the first clue. And remember that a blank cell and a zero are different things to an average.

With those in hand, the statistical functions in Grid by Zaweba turn a column of figures into something you can explain to someone else, which is the real point of doing the sums.