Quota sampling: all the keys to using it in your surveys

Quota sampling is one of the most common sampling methods in market research. Mastering it will help you to easily implement surveys without help. Today we tell you everything you need to know about quota sampling to use it safely and confidently.

What is quota sampling?

Quota sampling is a type of non-probability sampling in which researchers select participants in a sample based on certain specific characteristics, usually demographic data (such as age, gender, socio-economic status, etc.) in predetermined proportions. These are generally used so that the quotas correspond to the same proportions in the general population. For example, if it is known that 60% of a population is female and 40% is male, the researcher will actively seek to fill that ‘quota’ when selecting participants.

When quotas have been defined in a market survey and the desired number of a quota is reached, no more data is collected for that segment. The quota is filled and the survey continues in the field only for the quotas that have not yet been filled.

Characteristics of quota sampling

Quota sampling is widespread in market research because it allows for results with a high degree of reliability and efficiency.

• It is non-random sampling. This means that not all members of the survey population have the same probability of being selected. This is very common in market research using consumer panels, which is the vast majority. For example, if you want to conduct a survey on the population of a country, not all citizens participate in the panel and therefore it cannot be considered random sampling. In this case, using quotas is very beneficial to make the sample as close as possible to the national population.
• It is quick and inexpensive. The reason why market researchers use non-random samples is that they are much more efficient in terms of speed and cost. There are no directories with personal data on the entire population that can be used to select participants completely at random. It would be possible to randomly pick physical addresses from the street directory and go and interview the people who physically live at those addresses, but the cost would be prohibitive. For this reason, random samples are rarely used.
• It is common. For the above reasons, quota sampling is very suitable for market research and is therefore widely used.

When to use quota sampling?

Quota sampling is highly recommended when conducting online surveys using consumer panels. Setting quotas allows the sample to keep proportions equivalent to those of the survey population and thus reliable results can be obtained quickly and efficiently.

In contrast, quota sampling will not be the best option when:

• Access to the entire survey universe is available. For example, if a company wants to interview its employees, it will be able to contact all of them through its internal communication channels. Or a retailer might propose to interview all the people who have bought from the shop. In both cases, the researcher has access to the entire universe and therefore the sampling will be random.
• The characteristics of the survey population are not known. For example, a car company may want to survey people who are considering buying a car in the next six months. Unless you have data from a previous survey, you will not know the characteristics of the study population and it will be impossible to set quotas. If you have information on the survey population, but suspect that it is not very reliable or is outdated, we do not recommend using quotas either, as their use could introduce bias into the survey.

Representative population quotas

Representative quotas of a country’s population are specific proportions of individuals that accurately reflect the actual distribution of certain demographic variables in the country, such as gender, age, region, etc. The design of these quotas is based on a country’s population. The design of these quotas is based on official data, such as periodically updated censuses. For example, data on the Spanish population can be obtained from the website of the National Institute of Statistics.

By using quotas that correspond to the population of a country, you can obtain data that represent the population. For example, the gender of people can influence their opinion on an issue. By introducing quotas in your sampling you achieve the same proportion of men and women as in the country’s population and avoid the bias that could occur if you accidentally end up interviewing many more women than men or vice versa.

Non-proportional quotas

When setting quotas you can also opt to use quotas that are not proportional to the population.

Non-proportional quotas (or disproportionate quotas) are quotas where the sizes of subgroups in the sample do not accurately reflect their actual proportion of the population. Instead of selecting participants according to the actual weight of each group in the population, a different number is deliberately chosen for practical or analytical reasons.

Suppose you want to know precisely what the general population thinks in a survey, and you are particularly interested in what Generation Z thinks. The weight of Generation Z in the overall population is small, so depending on the size of the survey sample, you may not have enough cases of the generation to get an accurate picture. In this case, you can do quota sampling and set age quotas that match generations and give a more than proportional weight to the Generation Z segment.

But be careful, when you define non-proportional quotas, you have to take into account that the total sample is biased. If you have a higher share of Generation Z than their weight in the population, their opinions will be overrepresented in the total column, which will lose representativeness with respect to the population as a whole. In order to use the total column in your analysis, you will need to ‘weight’ the results of the quotas so that the total set is once again representative of the population. Without this adjustment, the total sample is not representative of the population.

Cross-quotas

Cross-quotas are a more advanced way of setting quotas in a sample, as they combine two or more variables at the same time. A classic case is to set quotas by gender and age bracket. Using them in your quota sampling, instead of setting a quota for ‘men’ and another quota for the age range ‘18-35 years’, we set a quota for ‘men aged 18-35 years’. This avoids that, for example, in a specific age bracket we might have an excess of men or women in relation to their presence in the population. Crossing gender with age in the quotas allows us to have a more balanced sample.

Crossing quotas has advantages, but also disadvantages. Using quotas makes it more difficult to find people who meet each and every one of the conditions of the quota, which makes fieldwork more complex and time-consuming. If quotas are too restrictive, it is possible that the quota will not be met in a reasonable amount of time. When this happens, the sampling design may need to be modified to make it more flexible, or a posteriori weighting may need to be done to over-represent respondents who meet the quota criteria in the total.

How are quotas set?

To determine the quotas you want your survey to include, follow the steps below:

• Define the relevant variables. The total number of quotas you set in quota sampling directly influences the quality of the sample. As you increase the number of quotas, you can achieve higher quality data, but you also make the fieldwork more complex and costly. This is why it is important to select the key quotas that really matter for your analysis. Avoid adding unnecessary quotas that will not add value to the survey. Gender and age are the most common, followed by geographic areas. Socio-economic class can be interesting depending on the object of analysis. Here you will find a complete list of socio-demographic data that can be used.
• Consult the baseline data. Look for statistics that reflect the composition of the population or surveys that reliably describe the characteristics of the universe to be studied.
• Calculate the proportions. For each segment, calculate the percentage it represents in the total survey population.
• Adjust for your total sample size. Once you have decided how many people you want to interview, distribute them in your quota sampling using the same percentages as the survey population. This will give you the number of people you should interview in each quota.
• Cross-reference the variables if necessary. Use simple quotas when your survey does not need too much detailed segmentation and the objectives are relatively general. If you want more rigour in the sample selection, allowing you to do a more detailed analysis of the different segments, choose cross-quotas and calculate the weights for each of the crosstabs.

How are data weighted when using non-proportional quotas?

When non-proportional quota sampling is used, it is necessary to weight the data to correct for inequalities arising from over- or under-representation of certain groups in the sample.

Weighting adjusts the data to better reflect the true proportions of the general population, allowing for more representative results and avoiding bias.

To do this, follow these steps in your quota sampling:

• Calculate the theoretical weight of each group in the general population. The first step is to collect from reliable sources the proportions of each group in the total population (e.g. the percentage of men, women, youth, adults, etc.).
• Determine the weights for each group in the sample. Since you have a non-proportional sample, you will need to assign a weight to each share, depending on how much it is over- or under-represented relative to the distribution of the universe. For example, if in the population 60% are women and in your sample only 40% are women, this means that to balance gender representativeness, each woman in the sample should count as 1.5 times more than a man.
• Apply the weights to the results. Once the weights have been calculated, they should be applied to the responses of each group. This is done by multiplying each person’s responses by the corresponding weight. In the example above, applying the weight of 1.5 to the women’s responses would stop at 60% weight in the total. To weight your data, you can use spreadsheets or advanced statistical calculation packages, such as SPSS, Stata or R. All of these have advanced functions for weighting. They all have advanced functions to make weighting easy. To do this, you will need to download your raw survey data and load it into the software of your choice.
• Check the weights. Once you have calculated the weighting, it is important to check that there are no excessive weights and that at the end of the weighting, the sum of the weights of all cases is approximately equal to the sample size. This ensures that the adjustment has not altered the total observations significantly. If the weights you obtain in the weighting are excessive, you should consider whether the definition of quotas is optimal. Learn lessons to define quotas for future surveys so that they are more proportional to the universe and thus obtain better quality data to begin with.

Common mistakes when using quota sampling

Although quota sampling is a very useful technique, some researchers may make mistakes that affect the validity of the results. Here are some of the most common mistakes made in quota sampling:

• Define redundant quotas. For example, crossing age with generations does not make much sense, since generations are a classification of specific age brackets. If we do so, we are adding complexity to the survey without improving the reliability of the results.
• Failure to adjust quotas over time. In ongoing surveys, quotas should be reviewed and adjusted periodically to ensure that the weights do not become outdated and continue to correctly reflect the population structure.
• Do not take population distribution into account when calculating quotas. There may be good reasons to over-represent a quota, but if there are not, it is very important to stick to the weight of the survey population. This reduces the need for weighting to make the total representative.
• Do not weight when there are non-proportional quotas. If non-proportional quotas are used, weighting is essential to use the sample total as representative of the population.

Quota sampling with We are testers

In We are testers, we can do quota sampling easily. When we do the survey for you, you don’t need to worry about anything. Just let us know in the survey briefing and we’ll take care of everything.

If you want to carry out the survey yourself, the research platform allows you to set quotas by gender, age and geographic area, where you can choose quotas by province, by autonomous community or even by Nielsen areas, and you can also choose whether you want to use simple quotas or cross-quotas, and which is the main variable you want to use to define them.

When you select quotas, the platform will distribute the quotas by default by dividing the total sample between all of them. Depending on the objectives of the survey, remember that you will have to adjust the sample you want in each quota so that they are representative of the population. To do this you should consult the official population data published in your country (INE in Spain). And remember that if you encounter any difficulty when designing your quota sampling, we are at your disposal to solve it.

Launch your survey using quota sampling or contact our experts for help today.