#### Let's Take A Closer Look

Explaining complicated subject matter simply since 1986

Agatha Christie was a British novelist who wrote the Miss Marple and Hercule Poirot whodunits in the 1930s and 1940s. This was the golden era of British murder mysteries, with Arthur Conan Doyle’s Sherlock Holmes and Doctor Watson the most memorable of all. Fascinated with murder mysteries, Anthony and Elva Pratt invented a board game they called Murder! The U.S. rights were sold to Parker Brothers, who gave it the name Americans know today.

#### Clue is a whodunit game

Players try to solve a murder mystery by determining who did the foul deed, where the crime took place, and which weapon the murderer used.

#### The game board is made up to look like rooms in the murder victim’s mansion

Players move around the board, collecting clues along the way as they compete to be the first to solve the murder. The victim is the owner of the mansion where the murder took place, Mr. Boddy (get it?). Every player is a suspect.

#### There are six suspects

Miss Scarlet, Mrs. White, Mrs. Peacock, Mr. Green, Colonel Mustard, and Professor Plum. Their names match the colors of the tokens players use to mark their positions as they move around the board.

#### There are six weapons

A candlestick, a knife, a lead pipe, a revolver, a rope, and a wrench.

#### The mansion has nine rooms

The Ballroom, Billiard room, Dining room, Lounge, Library, Study, Conservatory, Kitchen, and Hall.

#### Players roll a single die and move about the board in search of clues

The first to properly identify the killer, the weapon, and the room where the murder took place wins. Winners use strategy, logic, and the process of elimination. The penalty for guessing wrong is immediate elimination from the game. Guessing is not a strategy, which is why winners don’t bother. Winners patiently collect enough clues and then correctly interpret them.

#### How many possible combinations of murderer, weapon, and location are there?

Multiply 6 possible suspects times 6 possible weapons times 9 rooms and you come up with 324 possible combinations of suspect, weapon, and room.

#### 323 of them are wrong

If you guess, you have only one chance in 324 of being right.

#### Think about our 324 possible solutions

Each of them is made up of three separate components. In the game of Clue, they are the murderer, the weapon, and the room. Knowing only which weapon or room or person doesn’t get us to the answer and neither does knowing only two out of three. To solve the murder mystery, we need to figure out all the components of the solution, not just some of them.

#### How I teach researchers about sampling

The best teachers connect new things to things students already know. In training sessions with MBAs, I use the six murderers, six weapons and nine locations of the Clue game to illustrate what goes into choosing a survey sample that has the necessary good cross-section of respondents.

Teams are told their goal is to conduct a study of Consumer Attitudes in Europe. Study after study show that age, income and geographic location are three important variables, so let’s start with them.

#### Instead of six murderers, think of six different age groups:

If we only sample any single age bracket, we have eliminated five of six possibilities before we have even begun. If we sample three of the six age groups, our sample will still never be representative. For our sample to adequately stand in for the population at large, we need to choose enough from each age group.

##### Instead of six Clue weapons, let’s think about six income brackets:

When we collect data from each of six income groups within each of six age groups, we have “solved” only 36 of 324 possible solutions, so let’s move on to our sample’s geographic composition.

#### Let’s think about the nine rooms in the mansion as nine European countries

The continent has more than four dozen countries and we don’t have the time and money for all of them, so for illustrative purposes, let’s choose Denmark, England, France, Germany, Italy, Netherlands, Portugal, Spain, and Turkey as our “rooms.” Most companies selling products and services in Europe would be satisfied with sampling these nine countries because they are well-dispersed across Europe and more than half of the continent’s population lives in them.

Did you notice the first letters of those countries also accidentally anagram nicely into GIFT SPEND?

#### Choosing countries is just Step One

Take England, for example. Three out of four studies go only to London and insist their conclusions apply to all of England. This is a typical corner-cutting strategy that makes the ridiculous assumption that London = England. Anyone who tells you this is not to be trusted.

As London holds only about one-eighth of the population of England, we cannot draw reliable conclusions about the country without including in our sample the attitudes of people from the North East, North West, South East, South West, East Midlands, West Midlands, East, and Yorkshire regions, where seven out of eight English people live – you know, the other 60 million English consumers your competitors are ignoring.

#### Now we’re getting somewhere

To cover all the possibilities and not arbitrarily exclude any group from our study of Consumer Attitudes in Europe, we need to sample each of 36 age and income groups within each of our nine countries. We will then be able to compare the attitudes of people in all six age groups in all six income categories across across all nine countries, 324 segments in all.

#### Companies that commission surveys like to see sophisticated data analytics

Most don’t know surveys can only deliver meaningful analyses when each cell in the sample matrix contains a minimum of 40 cases. Conclusions and inferences that are based on samples with fewer than 40 cases per cell are known to numbers experts to be unstable, unreliable, and untrustworthy because they violate the mathematical laws that are built into complex formulas. With a 40-case-per-cell minimum and 324 cells, we need to collect data from 12,960 respondents. Anything less and we’ve handicapped our chances of solving our business problems.

#### Solve your marketing problems and figure out what new products to bring to market by first determining which categories are important to investigate

You will find in almost every instance, for every product and service, your research will need to effectively sample different age brackets, income brackets, and geographies. How well you do that is up to you.

#### Those who don’t give enough thought to their study samples are wasting money, and worse than that, opportunity

Don’t be one of them.