[Music] [Music] welcome back having developed a construct of motivation to buy now for your product you need to find out how good it is how well does your construct actually represents the actual motivation to buy for your target customers to do this you use your construct as a foundation to build a questionnaire this module will explain what a questionnaire is and how you will develop them so what are questionnaires questionnaires i'm sure things you have seen around many times and many of you will probably even fill them out questionnaires are fundamentally designed to generate data for you to process and analyze but a questionnaire needs to be designed in line with your research objectives goals and questions because your questionnaire has to collect all the necessary data to answer all of your research questions that satisfy all of your research goals and ultimately fulfill your research objective but they are more than that they are also measurement instruments that allow the researcher to assess differences in responses and answers of your respondents based on the level of measurements used now what exactly do we mean when we talk about measurements measurements are numbers or labels that you assign to objects people or events but in assigning measurements you have to follow specific rules to properly represent the quantities or qualities of the concepts you are trying to measure for example measuring students performance involves not just a letter grade like abc it also involves percentages numbers and even letters and texts in order to assess the quality of contributions and performance to generate the various indicators of a student's actual individual performance you can't just roll a dice so in research you need to know what you want to measure first and with that then you can easier determine how you will actually measure it often the concepts that must be measured are discovered as part of the problem definition process or as in our case are defined during the construct development so since your questionnaire is also your measurement instrument you need to incorporate appropriate measurements into your questionnaire in research or in statistics there are four general levels of measurements that determine how you will measure different questions in your questionnaire many of you might have seen those before no so we are looking at a nominal ordinal interval and ratio level of measurements so the most simplest level of measurement is a nominal level and as the name suggests you can name group or classify characteristics like gender or what is your favorite color but you cannot see any differences or how you can only see how many respondents like blue but you do not know by how much they prefer blue over the second choice of color so for the ordinal measurement here you can order just like the name says you can provide order for the characteristics you are trying to measure from the least to the most from the smallest to the tallest for the favorite from the favorite to the least favorite you are ranking you're ordering your data now you can see the differences in the rank for example who comes first and second and third in a race you will know okay the first one is the winner and then the second runner up and the third one up but you still don't know the real differences between them by how much does the first person came in and how long did it take for the second person came in or for the third person came in it can be minutes seconds or hours you don't know that is the part of ordinal data so for interval scale now it is a little bit different interval scale is very popular in marketing because here you have equal intervals on the scale you are not you're not only able to classify and rank the data but now you can also see the absolute differences for example if i ask you to tell me your preferences for a certain product feature on a scale of one to four one of you might say four one of you might say two then i can say the one who chooses four is twice as more likely to buy the product as the one who chooses a two because the preference is twice as much so here you can see absolute differences but still you cannot see the real differences in order to see real differences you need a ratio level of measurement but ratio is more often used in science or engineering because here we deal with real numbers an example would be the heights if i would ask everyone in the classroom how tall you are i could calculate the real differences down to the millimeter between each one of you so that would be the real difference now as i mentioned earlier the interval scale or the interval level of measurement is the most common one used in market research and most questionnaires apply a liquid scale in one form or another the ligard scale is often used to measure differences in preferences and attitudes but how do you know the amount of points you need to have on your liquid scale when you design your questionnaire should you have three four five six ten points you need to ask yourself a two simple question even or are many or few so when you look at an even or odd scale the differences are simple and even scale has no middle point that means you are forcing a decision if you are after preferences then the target customer or the respondent has to choose they either prefer or they don't prefer which one is it they cannot be neutral to the issue and there is a place for even scale or uneven scale if you have very sensitive information for example now where you need people to be able to choose a neutral point like for example political information you want to collect or talking about biases or other things criminal things where you need people to have an exit where they can stuck in the middle the problem in market research is if you have an uneven scale you have an odd scale and you have this middle point especially in the asian region people like to tick the middle point in asia people are less motivated by opinion than in western countries so my experience doing research here in asia it's always better to have an even scale but now the question is how many should you have a lot of points or should you have only few points on your liquid scale if you decide to have a lot of points then ask yourself does your respondent know the differences between a 9 and an 8 or a 2 and a 3 if not then maybe you should use a smaller scale and it's very simple it's a matter of sensitivity how sensitive do you want the measurement to be for your questionnaires if you know you are addressing experts in the field for example experts in marketing or expert engineers in the product design that will help you to to get certain tools ready then you can be sure they will know the differences between an eight and nine but if you intersect people at the shopping mall and asking hey do you have a few minutes can i ask you a few questions how likely are there experts in the subject matter that you are trying to investigate so in that case it would be better to have a smaller scale so it's also a matter of experience and knowledge the more experience the more knowledge a person has about the subject matter you're trying to investigate the more points you can have when we look at questionnaires we also need to consider reliability and validity reliability is all about consistency how consistent is your measurement how consistent is your question or questionnaire you know and every single question in it do you really measure the same thing over and over if you are trying to measure customer satisfaction is your questionnaire really measuring customer satisfaction every time you conduct or you send out the questionnaire is that sure that two people get the same questionnaire and they're not misunderstanding it and one thinks it's about customer satisfaction okay and the other one thinks it's a kind of a product introduction so in that case uh it's not very reliable maybe because of the language you're using in your questionnaire people misunderstand people don't really understand what you're trying to ask the same uh when you remember back the example of performance evaluation for students how how clear is it that every time i grade an assignment i will grade it in the same way towards a certain standard that i have set at the beginning i can't just give someone an a and another one has less quality and i give them an a as well so they have to be reliable measurements included in that so also how good can i reproduce the same results so whenever i give an a that should uh reward or should deserve an a and not that in the next class then a similar kind of work i give only a b so they have to be a consistency they have to be a reproducibility in there when we look at validity here we look at are you actually measuring what you wanted to measure if you wanted to measure customer satisfaction but you're measuring motivation then you're not really measuring what your intention is and your research goes off track a little bit so here is all about accuracy think about you try to investigate a behavior of coffee drinker but you decide to collect data from a large variety of people without asking whether they're actually drinking coffee it's very likely that you get people into your respondent pool into your sample that actually don't drink coffee that are tea drinker or don't even like caffeinated beverages so if they respond to the questionnaire how valid is their responses they don't know anything about coffee how can they give you valid responses to coffee so it's all about intention what you want to measure and how accurate will this measurement be here are four different situations that you might encounter with your research in terms of reliability and validity the first situation it's not reliable and not valid when you look at the dots representing a measurement or response from a questionnaire and the circle represents exactly what you want to measure the scope in which you need to measure then you can see there are several different dots that are outside of the margin meaning you don't really measure what you wanted to measure with that response and it's also completely scattered every measurement is all over the map giving you an indication that you not consistently measure the same thing whereas in situation two you can see the dots they are grouping around a certain center which shows you that yeah you are measuring what you want to measure you are you you reliably measuring the same thing now so you're measuring customer satisfaction but for example if i would want to measure customer satisfaction for coffee drinkers but i collect data from tea drinkers then i would probably get something like a situation too i have a very good questionnaire that can measure satisfaction but i don't ask the right people in situation three you ask the right people but your questionnaire your measurement instrument is not very good and there are lots of misunderstanding and people don't think about satisfaction or preferences and situation four is what you are trying to aim for you want to have a higher reliable and high valid kind of research instrument so your research instrument in our case the questionnaire needs to consistently produce the same results and ask the right people and this is why questionnaire design is so important so since we have established what a questionnaire is and how important the right measurement is now we can focus our attention on how do we design a good questionnaire so the most important point here is that a questionnaire needs to be relevant to your research objective accurate in its execution but it also needs to be short easy to understand and quick to do how do you do that if you have so many questions right now uh the questionnaire is relevant to the extents that all the information you collect addresses your research questions and then fulfills your ultimate research objective it will also help you the decision makers to address the current problems so the relevance also comes with currentness if you designed a questionnaire two years ago and you're trying to reproduce the results you might need to freshen it up a little bit because something might have changed and you need to address that change so you need to address specific needs data needs with a research that is where relevancy comes in and accuracy now how accurate is your measurement or your questionnaire now you want to be very accurate by by showing the same measurement over time for example a bathroom scale when you step on in the morning and it shows your weight you want to have it show the right way in the right measurement unit now for example kilogram or pound you don't want it to show a different weight every time you step on it or lie to you okay if it's less than maybe good so questions need to be short easy and quick in your questionnaire this will help respondents to understand it easier and better and if respondents understand your questionnaire then they can reply to it much better and very often if you see a questionnaire that has like 20 or 30 pages attached to it how likely are you to actually execute that questionnaire if someone would come to me and ask me can you please take two hours of your time and answer 500 questions we have uh no thank you so your question need to be very focused very brief and very clear in order to keep the questionnaire as short as possible a good questionnaire can be done in maybe 15 or 20 minutes and it's easy to understand use easy and simple language don't try to use big words always think about who is your target customer who are your target respondents and try to cater the language towards them if you know you are in an international setting and english is not the prime language or the first language then you have to make sure that you don't use complicated words this is what often happens in thailand when thais try to interview english-speaking population or you try to interview thais with english you always get some misunderstandings no matter how good the english is so what type of questions should we choose for our questionnaire here the question is what kind of questionnaire are you having are you are you intending to collect qualitative data like an essay or a short answer type of questionnaire then you need to use open-ended question but if you want to use more quantitative way where you can assign proper measurements to the responses then you should use closed-ended questions where you fix the alternatives or the choices and for those close-ended questions you have three different types you have the dichotomous you have the multiple choice and the attitude scale so for dichotomous it's rarely used it finds its users for example agenda now what is your gender male or female it's a two-choice kind of answer in that case it makes sense but if you ask okay do you prefer this feature and you give only yes or no choices then okay you will know people like that feature or they don't like the feature but you don't know the intensity and with a likert scale for example when you ask them to choose between one two three four then you can see the intensity of their preference and this would be an attitude scale you can measure their attitude with a liquid scale so now you know uh by how much do they prefer the choice of other they really really like it or they're yeah okay i can i like it so you can see there are differences in intensities that you can measure with an attitude scale and the multiple choices just like an uncommon multiple choice exam you have certain choices you can ask your respondents to choose from choose more than one or choose only one now when we look at how do you word your questions there are a lot of things you need to consider and we will discuss more at a later point so thinking about ambiguity ambiguity is a different word for unclear so try to avoid unclear phrases unclear wording try to keep a sentence as short as possible without fluff words you don't need to i usually buy my coffee at starbucks you don't need that you can just say i buy my coffee at starbucks you can cut out usually or some kind of those fluff words over demanding recall over demanding recall means you are asking your respondents too much to remember and very often you will get wrong or biased answers because people will still answer even though they don't remember now for example how many bottles of water have you had this week i can't even remember how much i had this day so how much can you ask them to remember if you would ask okay how many bottles of water had you today all right that might be easier to to think about but if you ask them how much did you bought this year or this month no one will be able to remember that so you have to be careful when you ask for something of your respondents to recall are they capable of doing so make it as simple as possible the human brain is lazy by nature don't ask your respondents to put too much work into it then they will suddenly stop in the middle of your questionnaire or they start lying or you know just ticking whatever they feel comfortable with overemphasis sometimes the way we say things can bias the way we we would answer a question for example words like uh let's discuss the current financial crisis when you talk about the crisis this already provokes a certain sense of urgency in your respondents but if you talk about situation instead of crisis situation is very neutral and as a researcher you need to be neutral you cannot sway your respondents into one way or another so be careful how you phrase some of the questions in your questionnaire also reading now sometimes the way you phrase a question can lead to answer don't you think you should buy this no i mean that is kind of pushy kind of even borderline not very polite so leading questions biases are very common occurrence in questionnaires double barrel double barrel is something when you ask two concepts in one question do you like a car to be fast and safe now those are two concepts fast and safe once if it's in one question and you choose one answer for that question how do i know which concept are you addressing of course i want it fast but i also want it safe or some people they don't care about safety they only care about speed so which one is it once you have them in one question and the respondents have answered you cannot separate them so it is a better way to ask two separate questions do you prefer a car that is fast well and then you give a liquid scale one two three four or do you prefer on the next question do you prefer a car that is safe one two three four this way you can assess each different preferences to the two different concepts and the last one sensitivity sensitivity is often overlooked and i have seen lots of research where it could get hot in here no sensitivity when you ask people about things they are not comfortable to talk about for example something illegal uh hey what do you think about your gambling problem or you know how often have you stolen something from a grocery store people will not tell you that whether they did or not so when it comes to sensitivity you have to try to find ways around asking around it don't really ask directly in their faces and often if you do you will not get accurate information so there are many different ways how you can improve your questionnaire just by going through all those different notions and just keep in mind what kind of situations can happen if you choose a wrong wording and here's the problem because most of the questionnaires are designed by people in a language that is not their native language if you are here in asia so you might need to have different language versions for your questionnaire thai chinese english german whatever you might need but you can't just translate the questionnaire you have to be very careful when you do translations we can discuss that at the later point so one more thing for questionnaire design is the layout there are many different opinions about how a questionnaire should look like should the demographic questions be right in the beginning or not how much should i have how many questions there is no right or wrong answer i would like you to look at this kind of a layout it starts with an introduction of course you need to tell your respondents who you are and what you are after don't tell them everything i'm after your motivation to buy might be a little bit off-putting a bit too much but you can talk okay we want to investigate preferences for product features try to be general explaining what you are doing and who you are also introduce your product if you have a new product people might not know it introduce a little bit show some pictures or anything that can help people to envision what you are trying to ask them later on in the questionnaire and tell them how long it will take okay do you have 20 minutes time but don't lie if you ask them hey can you have 20 minutes to do our questionnaire and then it takes an hour they will hate you for it i promise you or some people might not even finish the questionnaire and then it is useless for your analyzers and the next up is a screening question one or two screening questions this will help you to raise validity because here you are asking who has actually experienced with the product or similar experience about a similar product that you are trying to ask questions in your questionnaire so screening questions are very important because they can weed out people who have no idea what you are talking about and can raise your validity and then you start warming up your respondents asking some general questions here you can put in a lot of things like where do you normally shop for your coffee uh what type of coffee you buy general things easy to answer easy to remember and then you start with specific questions here is it where you put your construct about motivation to buy here are all those complicated questions measured on a liquid scale so when you look at all the other aspects of your questionnaire the screening questions the general questions they are mostly multiple choice or even dichotomous for example demographic questions all your private information what is your agenda your personal income your hobbies um what is your marriage your status what is your occupation etc those are all multiple choice and dichotomous questions but when you look at the specific questions you will want to use a liquid scale because only when you use a liquid scale you can use advanced statistics to do the analyzers with nominal and ordinal data you cannot do that much in the next session we will talk about [Music] sampling