# Sampling

## Key Vocabulary

Sample- this is a proportion of total population.

Population- this is the total group that is being investigated.

## Choosing a sample

If we ever choose to do a survey or to investigate something, more often than not we need to choose a sample from the total population.

When we choose a sample we want to make sure that our sample is representative of different groups within our population. This will make the sample represent the whole population, if it doesn’t the investigation may be biased.

Within these groups we want to select our sample at random to make sure there is no bias in our selection.

We also want to make sure we have a sufficiently sized sample, the bigger the sample the more representative and reliable the data will be.

**How to sample your data:**

**Stratified sampling.**

This method is where you take a proportion from each group within your population to be your sample. For example, if Nigel decides to take a sample of students from his school and he wants to take a sample of 10% of the student body he will take 10% of the students in each year. So if the number of students in each year varies so will the number of students he asks in each year:

Eg.

Year | Number of students in the year: | Sample size 10% |

7 | 100 | 10 |

8 | 120 | 12 |

9 | 110 | 11 |

10 | 115 | 12 ( he has rounded up here) |

11 | 80 | 8 |

Now that Nigel knows how many students to ask he can use a simple random sampling method to choose them.

**Simple random sampling**

This is where individuals from the population will be selected completely at random. If we have decided to use a stratified sampling method as well, this the way that we are going to choose the specific individuals (eg. for Nigel it will be the number of students in each year).

There are several ways that we can choose people eg:

Number each member of the population randomly and using a random number generator select the sample

Put names into a hat and pick at random (blindfolded- having made sure each piece of paper is exactly the same- just with a different name or number)

## Limitations of samples

Even if we are able to be very careful with our sampling methods there will always be limitations to our data, as without using the entire population the methods will never be fully representative.

Additionally, we need to be careful with how we select people for our sample. Key areas where there might be issues are:

- The sampling method itself
- The location
- Time
- Size of the sample

For example:

- Taffy asks a survey about owning pets at a dog walking park, here the location is going to bias the sample!
- Jack asks a survey about people’s favourite types of music. He asks some people at 9pm on a Saturday night and then asks people at 8 am on Monday morning.Each time he gets very different results.
- A telephone survey asks people at 5 pm what their favourite TV programmes are. This is bias as it will limit the population to people who are home at 5pm.

**Phrasing Questionnaires:**

The way that we ask questions can also bias our sample. We need to make sure our questions:

- Give a specific time frame (if appropriate)
- Do not give options that overlap
- Do not bias the respondent to a particular answer.

Eg.

*How often do you watch T.V - does this mean per week?? Per month? Per day?? You need to be specific.*

A better question would be:

How many hours a week do you watch T.V.

*How many times a week to do you to the gym 0-3 3-5 5-8*

Firstly your respondent will not know where to circle if they go to the gym 3 or 5 times per week.

Also what if they go to the gym more than 8 times?? Always make sure there is room for a higher amount if theoretically your question could be infinite.

*My favourite character in friends is Phoebe, the rest are boring. Who is your favourite Friends character?*

This question judges potential answers, making the respondent more likely to say Phoebe is their favourite character.

- Sahil asks a questionnaire about people’s favourite form of exercise outside a swimming pool. What’s wrong with this?
- Your answer should include: Location / Bias / Sample
- What is wrong with this questionnaire question? "How many books do you read? 0, 1-2, 2-3, 3+"
- Your answer should include: Time Frame / Options Overlap