How to use quantitative research – terms
When you commission market research for your business, it’s likely you’ll call in the services of a professional market research company. While you do not need to be a technical expert in statistics and questionnaire design, it is useful to learn some of the key terms that you’ll hear and some of the key questions you should ask to make sure the research is well planned and run to a high standard. This will give you more confidence when it comes to how to use quantitative research.
When the answers are known, and the respondent has to choose from a list of options, these types of question are called “closed” questions. You can spot these types of questions, because they tend to use “what”, “which” or “if” in the way they are phrased.
So from the example above, the actual research question might look like this. “If you bought product X in the past, which of these three options – price, quality, convenience – was the main influencer of your purchase?”.
Closed questions are best used when you want to ask a large number of questions. Closed questions are easier and faster for respondents to answer. They require less thinking time (to come up with an answer) and only require the time to ‘pick’ the most suitable answer. You can get through more questions in less time.
They are also best used when you need to ask the questions to a large number of people. So that you have a statistically representative sample of the total target audience.
Statistically representative sample
As mentioned at the start of the guide, census level surveys are expensive and time consuming. Quantitative surveys are less expensive and faster.
But it’s important to understand how the statistical connection between the representative sample and the total population is derived.
So here are some terms that you should understand and look for when your market research company talks about sample size.
This is the sum of the values of a question divided by the total number in the population.
Let’s say we ask three people what they would pay for a product, and they respond $10, $12 and $16. The mean is therefore the sum of the values ($38) divided by the total number (3 people) or $38/3 = $12.67.
This is usually called the ‘average’ by most people, but for some reason statistical experts call it the ‘mean’.
Perhaps they do that to be mean, we don’t know.
The variance is the sum of the squared deviations about the mean divided by the number in the population.
This is where most non-statistical people start to get confused, so let’s go slowly using the same three $10, $12 and $16 answers above.
The deviations are the difference between the actual values and the mean.
In this case then, the deviations would be -$2.67 ($10 – $12.67), -$0.67 ($12 – $12.67) and $3.33 ($16 – $12.67). So the sum of the ‘squared’ deviations would be $-2.672 + $-0.672 + $3.332. Or 7.13 + 0.45 + 11.09. Which equals a variance of 18.67.
The standard deviation
In reality, variance is rarely referred to except in market research. But the square root of the variance is referred to often as this number is called the standard deviation.
In the above case, the square root of $18.67 is 4.32 so our standard deviation is 4.32.
When a sample is taken of the total population the results for that sample can be compared to the total population.
However, every sample taken will be slightly different. Imagine, you have to pick 100 people out of 1,000, there are multiple variations on which 100 people you pick out.
What confidence intervals do, based on calculations around the standard deviation, is calculate the likelihood of the sample you picked out falling within the standard deviation range you calculated.
You would expect this to be at least 80% and in actual fact, 95% confidence level is the norm in the market research industry. This means that in 95% of cases, the results from your sample would fall within the standard deviation range of the total population.
How these confidence intervals are actually calculated requires a more advanced understanding of statistics. This includes variables such as margin of error and knowledge of different distribution curves.
In general, you don’t need to know the detail of this. But if it is of interest, you can read more on minimum sample sizes and confidence intervals here.
Sample size recommendation
In practical terms, what all this statistical calculation leads to is a sample size recommendation from the market research agency. How many people they need to interview to be (95%) confident the results represent the total population.