Speaker 1: Hello Great 10s. In a previous video we joined MacGyver as he chose his sample method for his questionnaire on whether the use of uniforms should be stopped. We will join him again now as he draws up his questions. He is having problems with phrasing the questions so that the person isn't led into an answer. Let's see how he deals with this problem.

Speaker 2: MacGyver will be joining us again for this lesson. MacGyver, why don't you give us a

Speaker 3: quick reminder of what we did last time? Sure. Last time we started drawing up a list of rules for planning a survey and to make sure that the process is fair and unbiased. We saw that it is important to have a big enough sample size and to select a sample randomly. Also, that if there are clearly identifiable groups with different opinions in the population, you may need to stratify the sample to make sure that each group is

Speaker 2: fairly represented. Great. Now we can have a look at how we can avoid bias in a survey questionnaire. To start us off, let's look at a question that you

Speaker 3: used in your small survey. Okay, here it is. School uniforms are expensive and not everyone can afford them. So do you think that school uniforms should be abolished?

Speaker 2: This is what we call a leading question. You have put some information in the question that persuades the respondent to answer in a certain way. Well, it's true. Uniforms are expensive. It's not a lie. No, but one of the important principles about writing a question for an interview is that your choice of words need to be or should be completely neutral. You want to find out what they really think, so your question should not encourage them to answer in one way or the other. Do you think you can rephrase this question

Speaker 3: so it is more neutral? What if I just ask, should school uniforms

Speaker 2: be abolished? That is much better. Now just to show you how much difference a question can make, try this. Rephrase your question so that whoever you ask is likely to say that uniforms should

Speaker 3: not be abolished. Okay, let me give it a shot. How about this? Do you think we should give up discipline by abolishing school uniforms at our schools? Oh, you learned fast. People might not agree

Speaker 2: on giving up on discipline, even though they might think that abolishing school uniforms is a good idea. So they might answer no, even if what they really want to say is yes.

Speaker 3: Oh okay, so should I write that down as another rule in the list for planning surveys, which

Speaker 2: I started in the first lesson? Yes, that can be your second rule. The survey question should avoid bias by not suggesting a preferred answer. Right, now there are a few other things to consider about how you ask people questions. We want them to answer freely, so you need to make sure that everyone in your sample knows that they should give their true opinions. Perhaps by keeping their answers anonymous? Like when people vote in the elections? That's right. That's one of the ways that they ensure that the elections are free and fair. It's also important that every person is approached in exactly the same way. What do you mean? Well, how a person is approached can influence the results. It can help to ask them to fill a questionnaire which you collect in a closed box, rather than answering you face-to-face. There are some situations in which this is very important. For example, in a government election or a referendum, people might not want others to know who they are voting for or what they think about an issue. They may not answer as honestly if the vote or the

Speaker 3: survey is not anonymous. That makes sense. So we can write, approach people in such a way so that they're able to respond freely. The questionnaire may be

Speaker 2: anonymous if necessary. You also need to make sure that the responses you get can all be used. What would you do if some people responded to a question by saying something like, I don't know, or it depends, or maybe?

Speaker 3: I'm not sure. I suppose I'd throw out some of the answers. Because, I mean, surely you're

Speaker 2: only interested in yes or no answers. But that would cause a problem. If you throw out answers, your sample size might decrease. It might decrease in some groups more than it does in others. And this would lead to? To bias. Well done. Ignoring some results could lead to bias, and it reduces sample size, which makes the sample less representative. So your planning must always include a way of collecting a range of responses so that everyone's answer can be included. We should allow people to give a not-sure response. This is better than a non-response, which reduces our sample size. One way is to create a set of check boxes like these. It's also easier to organize data that has been collected in this way. Another advantage of using check boxes is that it allows people to answer quickly and easily, which makes them more likely to respond. That's true. I've seen other examples of check

Speaker 3: boxes in other questionnaires. Here's an example I found in a magazine for teenagers. Rather than asking somebody what their age is, you could give a range of age groups and ask them

Speaker 2: to check the correct box. It's quick and easy. It's important that everyone responds so that their sample remains representative. When you design your survey, you should think about ways to make it easier for people to respond, so that you increase the response rate.

Speaker 3: So our next rule is, design the response form so as many people as possible from the sample will respond. To do that, we need to make sure that all the answers can be recorded, and that it's easy to fill in.

Speaker 2: Mama Kava, how do you feel about your original statistics? Do you think that they represent the general opinion of all the interested parties?

Speaker 3: I guess not. I haven't asked teachers and parents, so I can't draw any conclusions about those two groups. And the sample group of learners I used was not only biased, but too small, so you can trust that all learners would agree either. I guess I'll have to start again. And don't forget that you've also changed your questions to allow for free responses. I think I'm also just going to concentrate on learners only. When I'm finished with my research, the statistics will show me what learners think. And when I tell people which results the group relates to, that should be honest, shouldn't it?

Speaker 2: Yes it would. Now, let's have a look at what it means when we have designed our survey properly. It means that we have data that is likely to represent the whole population. So it allows us to generalise the results.

Speaker 3: So I guess we can use it as a general statement. If most people in the sample respond yes, then most people in the population will respond yes.

Speaker 2: What we've learnt in this lesson is the importance of designing a survey carefully to eliminate the effects of bias. Let's sum up what we've learnt in these two lessons by looking at the rules for designing a survey. First, use a representative sample. To do this, choose an appropriate sample size, use a stratified sample if there are identifiable groups in the population, and select the sample randomly. Secondly, the survey question should avoid bias by not suggesting a preferred answer. Thirdly, approach people in such a way that they are able to respond freely. The questionnaire may be anonymous if necessary. Fourthly, design the response form so that a range of results can be collected, and to ensure a high response rate.

Speaker 1: This shows us how important it is to phrase our questions so that people feel free to answer honestly. That's all for this lesson, Grade 10s. Why not try some of the tasks in the Data Handling Task Video? You will also be able to learn more about data handling on our website www.mindset.co.za forward slash learn. Goodbye.