Comprehensive Guide to Survey Research Design in Cognitive Psychology

Hello and welcome to the course basics of experimental design for cognitive psychology. I am Ark Whmer from the

department of cognitive science at ID Kpur. This is week two where we are discussing different types of research

designs and uh so far in this week I've been talking about descriptive research designs and some of the nuances related

to them. And this lecture also I'll continue our discussion on different kinds of research design and what are

the different factors that are associated with them. Now one of the most used types of

descriptive research uh has been the survey research. I'm sure you must have come across several surveys in your uh

you know uh lifetime around you and we are constantly filling several surveys anyway. So for example if you go to a

mall there is somebody with a list of questions that they want to ask. Every now and then there are customer surveys

that we fill. uh other times there are say for example uh student reaction surveys, teacher surveys and so on. So

we basically uh something it's it's something that we are very familiar with but what is it and how can we use it as

a research method is what I'm going to talk about now. A survey typically is a series of self-report measure

administered either through an interview or a written questionnaire. So sometimes you know you will you'll get a call from

somebody saying I'm calling from so and so bank or I'm calling from so and so company. We wanted to know your views

about question A to A, B, C, D and so on. So either you were sort of giving this on a phone call or you were filling

a questionnaire. Maybe sometimes on a scale of 1 to5 how much do you like our service on a scale of 1 to5 how

satisfied uh are you with the you know executives behavior this and that. So we we are sort of aware with very aware

with how surveys work and that is basically uh you know testimony to the fact that surveys have actually been the

most widely used methods of collecting descriptive information about a group of people a sample or a general population.

For instance, telephonic surveys or small interview at a shopping mall or you know they sometimes hand out these

small pamphlets with 10 or five questions and you fill them you get a scratch card and you hope that you'll

win something at the end of it. What is what are these used for? What are these survey people trying to do?

Typically uh or the idea behind this method is to produce a snapshot of the opinions, attitudes or behaviors of a

group of people at a given time. All right, it could be consumers. For example, I'm launching a new product.

Let's say a washing uh detergent and I'm standing at the aisle of a particular departmental store which has different

kinds of detergents. I can just ask people questions about okay what is the most important quality of a detergent.

Is it its smell, color, uh amount of bleach therein? Should have chemicals, should not have chemicals, should be

economic, should be machine friendly. So I can come up with five six different questions and also things like how which

detergent are you using at this point. So I get some data about what is the current market share of different

competitors. What are people looking for and see you know how I can interview and create a brand and create a unique sort

of strength for my brand. So this is one. The other is say for example a lot of times we are filing we are filling

surveys for customer service executives. Okay. So uh you go and you take a service let's say you send your car for

servicing and they'll send you a question saying uh how much did you like the behavior of our executives? Was the

car serviced on time? Are you happy with it? Are you happy with how our executives talk to you? Any number of

things. What these questions basically do is they create a snapshot. They create a distribution of a certain

variable uh in the population. How many people use a blue color detergent? How many people would prefer a cheaper

versus a more expensive detergent? How many people pay more importance to executive behavior as compared to how

well the car was serviced? Things like this. Now another type of survey instrument is

basically interviews. So a lot of times uh you know surveys are administered in form of interviews wherein an individual

that is the respondent is presented with some questions either in person or over telephone. Now an advantage of in-person

interviews is that they allow the researchers to develop a close rapper and sense of trust with the respondent

and then basically a lot of information can come out lot of information can be elicited and the person is more relaxed

and and a let's say seamless sharing of information is happening. You might be aware that there are so many podcasts uh

you know that are happening these days. People are uh very popular by making podcasts. What they do is the format is

rather open-ended isn't it? There's a person sitting on the left, sitting on the right. They start with a brief

introduction but then they sort of go ahead with sharing asking questions sharing answers and in in that sense it

gives you a very uh interesting insight into the kind of questions you wanted to ask with that person be it a politician

Bollywood personality a cricketer or sometimes just common people. All right. So uh this rapo that is built and

something that you'll see in most of these interviews it's they typically start with a very informal introduction.

They typically start with something that will break the ice, something that will endear the respondent and the

interviewer with each other. And the idea is that once that ice is broken and that is why you'll see in a lot of cases

before the interview goes online, before the interview is getting recorded, the interviewer and the respondent have met

and interacted, they have been briefed uh about each other's likes and dislikes. The interviewer tells okay

broadly these are the questions I will tell you. uh the respondent says okay you can ask me anything but please don't

ask me x y and z so that there is a level of comfort within those degrees of freedom uh the interview sort of

operates and in that sense uh to that level it seems like a very interesting uh experiment

okay uh they say it leads to more honest and open responding so that is also something that is a advantage of taking

in-person interviews and building rappers between the respondent and the interviewer

also So face to face interviews are actually expenses expensive to conduct isn't it? Uh you will have to go travel

to a certain place uh find a you know place to record find a nice place to sit maybe in a restaurant which rather noisy

so you probably rent a studio and this and that. So in-person interviews if you want to do it at scale are rather

expensive to uh set up. You can go to different people's houses but typically people are not very comfortable getting

a stranger in their house for whatever interview that you want to take. So for that reason you will see that for most

purposes so for example credit card uh uh bank loan home loans or you know other kinds of things uh telephonic

surveys are much more common than face-to-face interviews at least in the market in the scale.

Telephonic surveys are therefore and as as you can see they are more economical uh they are efficient and several

interview interviewers can be located in the same place and the interview can be conducted in a coordinated manner. So

for example if you are a sales if you are a you know sales or a survey company you can employ 20 people to make uh

let's say 500 calls each day. So these people will uh you know make use of the telephone data. They'll call 5 minute

people and they'll basically be able to collect information at a much larger scale. All right. So in that sense while

telephonic interviews you know while uh in-person interviews have their own advantage telephonic interviews are

typically chosen for their you know ultra efficiency and and uh for being less expensive.

Also the format of the interview is something that we can talk about. We've talked something similar about this I

think in one of in in the last week one of the lectures but let's do this in a little bit more detail. So we know there

are two kinds of interviews structured and unstructured. So in an unstructured interview what happens is that the

interviewer typically can talk freely with the respondent about a range of topics. So I'm sure uh you know if

you're watching podcasts and if you're watching interviews these days on YouTube and some of these other channels

you'll see that a lot of interviews these days I'm sort of uh you know I remember uh watching uh you know Nillesh

Mishra's uh slow interviews with a lot of uh you know famous personalities and those interviews are very open-ended

they are they are sort of self-paced and about a range of topics there's no structure there's no uh clear path path.

There might be a broad path point A, point B, point C, but the exact steps you take uh you know in going from point

A to point B is not extremely restricted. Okay. So uh in unstructured interviews what you will see is that the

interview interviewer is talking freely with the respondent and he can talk about a range of topics. Yes. uh while

the interviewer uh you know uh may have a broad range of so this is what I was saying that while the interviewer may

have a broad range of preddecided topics during the interview the researcher uh you know that researcher may focus on

only those topics that the respondent is uh most interested in talking about say for example you can have a list of 20

items in your head but when you start talking to people let's say when you're uh you know interviewing people who have

survived a calamity maybe an earthquake maybe a flood All right. Uh you have any number of things in mind but when you go

the person is so agrieved they are sort of uh you know uh telling you about their struggles about their struggles to

get food. Uh you know their houses might have been destroyed. There might have been deaths in the family. So what

typically would happen is you'll not push your list. You'll basically ask a question and let the person respond and

take the let the person take whatever trajectory he or she wants to take. when the time is right then you pose the next

question and that question can be typically you know following up from what the person has just said. Okay. So

it's broadly it's a good idea to have a list of things that you want to cover in an interview. But in unstructured

interview unstructured interviews it is very important that people let the respondents talk and the interview and

the direction of the interview is typically uh you know shaped by what the respondents want to talk about. So there

there can be you know there's no sequence here there's no uh requirement of linearity here and there is nothing

like okay I have to cover this entire syllabus also as the questions addressed in the

interview may differ across different respondents the interviewers are typically trained to ask questions in a

way that would elicit most of the you know information that is typically you know uh you're seeking for you requiring

from the respondents. So this is also something we don't talk about a lot. Interviewers need to be trained to

conduct the interview. They need to be trained how to conduct a unstructured interview versus how to conduct a

structured interview. All right? So, uh interviewers may need to be trained to be sensitive to what the respondent is

saying. A lot of times you'll see in TV interviews the respondent is saying something else and you know your

interviewer is sort of stuck on his own or her own uh trajectory and it sometimes causes a discord between the

two and it it's very apparent as a classical data collection method. It is also it does not work that way. So

sometimes these face-to-face unstructured interviews can be uh held in a in such a manner that uh you know

uh this is basically being done with a group of people at once. Okay. So the interviewer is there the interviewer is

asking question but there's not a single respondent but there are a group of respondents. So what happens in this

case uh is that the respondents are obviously responding to the interviewer but at the same time the uh respondents

are talking to each other as well and a particular uh bit of knowledge about that topic is emerging. A very common

example that uh you know emerges in my head is that of roundts. So you might have seen that a lot of roundts and

panel discussions etc are conducted uh very common these days on on the news and on social media and so on and what

happens is there that the question is posed the interviewer typically poses the question and moderates its direction

but the group of people that are there let's say if all Bollywood artists are there or if all cricketers are there or

if all politicians are there they will basically give their own answers sometimes their answers will encroach

into the answers of others sometimes they're responding into each other but a certain degree of knowledge is emerging

all right and that is basically what is called a focus group if if you do it you know more formally sometimes it is done

informally and again when I'm giving examples from social media news channels and so on please don't consider them as

uh typical academic examples I'm basically trying to give you uh more handy examples so that you could

understand the overall paradigm now so uh unstructured interviews can actually provide in-depth information about the

specific concerns of an individual or a group of people and therefore can also produce ideas for future research

projects or policy decision. For example, let us say a bunch of academicians are you know having a

discussion about uh how to uh you know uh how to sort of invigorate the interests of students in a given subject

or how to ensure that all students are attending the classes. What kind of policies be formed that all students are

uh you know uh interested attending classes working very hard in this and that. Now what would typically happen is

that you know there will be academicians there will be administrators there will also be maybe some student

representatives and if uh you know if a good interviewer is basically bringing out the most relevant questions and then

moderating the questions so that everybody gets a representation in in the overall answer the emergent answer

then obviously these kinds of uh panel discussions these kinds of group uh you know discussions can actually lead to

very interesting insights about future projects what what is the next thing we can do or say for example even making

policy decisions because uh different stakeholders are all at the same place at the same time and they're answering

engaging with the same important questions. So that is in some sense uh you know the power of these unstructured

interviews both group wise and individual wise. Now the interesting part or slightly problematic part about

unstructured interviews is that it is difficult task to train uh you know interviewers to ask questions in an

unbiased manner. So a lot of times you will see that when uh you know people are taking interviews uh their personal

inclinations and their personal biases start factoring in. Suppose I'm talking to you about a particular issue and I

have person A in the opposition and person B uh in favor of that particular topic and I as a person am in opposition

of that topic. Now what happens is you'll see it it happening a lot of time that the interviewer because he has or

he or she has a personal bias will ask certain kinds of questions to the opposing party and certain kinds of

question to the favoring party and in that sense what happens is that the interview does not remain neutral

anymore. the interview does not remain unbiased anymore and whatever the result of that kind of interview would be it'll

always be or more likely be lopsided than a neutral uh result coming out of a you know honest discussion among the

stakeholders. So that is again something I'm sure you have seen uh several times across social media, YouTube this and

that. But it is extremely important when we are talking about this in terms of aademic interview. If you want to study

let us say the effect of uh you know uh people getting displaced you know migrations you know European migrations

or migrations within our country because of let's say uh you know lack of labor lack of this that and so on and if you

want to evolve some uh you know important actionable information out of it. Uh it is very important that the

issue be dealt with academically uh without biases and uh whatever stakeholders are there everybody gets

their fair share of representation uh which will illuminate the topic which will illuminate and create you know

possibilities for making policy decisions. Now let's talk about structured interviews. We say we

basically were discussing that okay unstructured interviews are good and this and that but there is this

nonlinearity about unstructured interviews that we were talking about. there is this uh problem that not

everybody has method to be completely unbiased and neutral and conduct the interview. So some of these things are

handled better when the interview is very structured when you know exactly questions 1 to 10 have to be asked and

they have to be asked in exact the same in exactly the same order. Okay. So in order to elicit and organize the data in

a more systematic manner, researchers often utilize what is called the structured interview format. Here the

questions are prepared ahead of time and the interviewer reads the question to the respondent one by one in exactly the

same sequence. This obviously as I said has the advantage that it allows for better comparison of responses across

participants. It allows you to uh you know get all the knowledge that you wanted because you've basically asked

all of the questions that you wanted to ask. Also the response formats are controlled isn't it? It also say for

example uh you know the time frame the sequence the uh kind of response that is coming

the time that you are giving to each question all of this gets standardized across participants. So if you want to

collect the data for from 20 interviews or from 50 interviews, you have a better chance of being able to do it in the

structured interview format because everything is done in a comparable and systematic manner. So this is one of the

major advantages of the structured interview format. Now uh but interviews are not that common. They are at least

less common as compared to questionnaires. The most common thing that we come across is questionaires.

you know fixed format, self-report items but questions printed questions in and nowadays printed basically means uh

digitally written and so on also so you get a PDF you get a Google form that you have to uh you know sign uh you know

choose one choose two choose three and so on so printed does not really mean but printed basically means uh digital

or paperbased uh set of questions that one has to ask okay so questionaire include a set of fixed format

self-report items that is completed by the respondents at their own pace. So you send the questionnaires. A lot of

times what you'll see is uh people receive questionnaires in their emails. You know uh our students psychology

students uh do send to a lot of people uh questionnaires which are relevant to their research in in a Google link and

they expect that the people will respond to it. Uh they're given a broad timeline that please send the responses to me by

two weeks or one week or it or if it is urgent by the end of the day something like that. And it is interestingly

because this is done without supervision. You send the link of the questionnaire to somebody and you then

wait that in a day in 2 days in 5 days the responses will come back. So no uh supervision here. This is uh probably

the most economical way of conducting uh interviews or asking question to people because now what you can do is you can

add in your mailing list 5,000 people and send the same uh email with the link of the let's say Google questionaire to

everybody. Okay, there are also some interesting uh you know characteristics of these

questionnaires. For example, uh it is possible or it could be possible that questionnaires will produce more honest

responses than interviews because uh particularly say for example if the questions are questions are about

sensitive issues let's say uh intimate intimate details uh financial details uh personal uh details and so on uh then

the presence of an interviewer sometimes uh you know acts as an inhibiting factor. But if it is a questionnaire and

you have uh you know for the most part uh you know taken your attempt to anonymize it then it is likely that you

will get the best and most honest answers. There's no guarantee but at least the hope is that okay also

questionaires are less likely to be influenced by the characteristics of the interviewer because they are standard

which voice modulation I'm using to ask this question does not get manifested in the questionnaire. Okay. Uh which party

or position I support does not get uh mixed up with my interview but it is in in the same standard uh digital you know

sterile format reaching the people anyways and they're ask answering the question like they would want to. So uh

if and there's an example for example if the topic of your questionnaire concerns race related attitudes it it has been

shown that the respondents answer do vary are affected by the race of the interviewer. Okay so it could be

anything race so uh social uh categories it is race uh in rest of the world it could be religion it could be cast it

could be soio economic status it could be a simple thing like teacher and student. So if a uh teacher is asking a

student about what they feel about their teachers, obviously that is going to impact. When a student is asking another

student, oh what do you feel about your teachers? The answer might be completely different. All right. So this is again

something that uh you know is handled or sort of uh you know uh skipped around by questionnaires.

There's an interesting thing by the way a property of questionnaires which is uh you know noteworthy is that people don't

really respond to questionnaires all the time. you know the response rates are typically lower. I remember uh you send

out 200 questionnaires to 200 people or let's say well one questionnaire to 200 people and the number of people that

will actually fill it would probably be less than 100 sometimes 20 30 only. Among those 20 30 40 people who have

filled your questionnaire uh how many people have answered all the questions correctly and correctly there's no

correct incorrect here but how many of them have answered the questions in all their honesty and you know being upfront

with this will probably be an even smaller percentage. So this is is a bit of a problem with questionnaires and

unsupervised question answering because it may lead to incorrect conclusions as the people who return the questionnaire

may have responded differently than those who did not return the questionnaire. Also investigators can

sometimes you know as investigators we try to do this. We try to increase the response rates by

providing gifts or monetary compensations for completing the questionnaires. Sometimes we have class

credits. Sometimes we have monetary compensation. Sometimes we sort of appeal it in an emotional way this and

that. Uh you know or say for example we can have we can uh sort of work with the style of the questionnaire. We don't

want a very long questionnaire. We make a short questionnaire clear statements uh you know not a lot of choices three

four choices and so on. So in that sense how are the people responding to these questionnaires is a is a important

question which we are all talking about. Also uh in question is you are never sure that the participant is taking the

questions one by one or is doing one and doing 21 and doing 15 and doing 27 that the order in which the questions have

been answered is something that you cannot really ensure and in some cases it might have an effect on how you

interpret and analyze the data or how you interpret the answer. So say for example if a if the last question is

answered first versus the first question there might be properties of those questions that will basically have uh

you know a cascading effect to rest of the questionnaire and in that sense the entire questionnaire gets corrupted. So

the data becomes you know not usable and you're left with uh you know nothing but to sort of discard that data because it

is not useful anymore for your analysis. Now, so this is the types of instruments and the type of questionnaires we were

talking about, but there are other interesting things that we can uh do as well. So, how do we make our surveys and

questionnaires and interviews more effective? Okay. So, uh let's let's look at some of these ideas. So, Mitchell and

Jolly in their book research design explained uh offer us some suggestions and I'm taking them up uh because I sort

of also want to highlight how to do this. So, survey research may run into problems because the researchers were

not really clear. So one of the major problems that comes in survey research is if the researchers themselves are not

really clear about what is it they are seeking from the you know results. What do they want to find out? How do they

want to find out? Do they have a hypothesis? Do they have their key research questions already outlined and

clearly stated? If that is not the case, the items that you have made will sort of be all over the place. The responses

that you will get, you will not find a way to systematically organize them. And in that sense what will happen is you

will not really get uh you know your research per se will not be that successful.

Okay. Also uh they say that survey research may be flawed because sometimes the questionnaires or tests have uh what

is called poor construct validity. Okay. What is construct validity? This can basically happen say for example when

the questions uh seek knowledge that the respondents do not possess. You know sometimes uh in all our wisdom we make

very interesting questions but we don't take into account the target population. If you're asking about something to an

individual it should be something that that individual can answer. If you're asking say for example class 10

questions to a class one person uh how how class one student this will not really work. Okay. Also sometimes you

frame the questions wrongly. to frame the questions in a way that they hint at the answers that the researcher anyways

wants. So what people will do is they'll basically say this is what you want to hear tick mark and I'm done. Also

sometimes the questions are not clear to the respondents. So for example sometimes people use very flowery

language very complex language and the respondents may misinterpret the questions and they'll give you the

answers from that vantage point of the misinterpretation and in that sense that item becomes useless for you. Also

sometimes it is possible that the researcher uh you know misinterprets or miscodes the respondent's answer and it

may lead to completely different conclusions than the one you intended with.

Also survey research will have little external validity because the people who were questioned may not represent the

target population. If you've not sampled nicely, if you have not if your sample does not represent represent the target

population, then there's no external validity. Say for example uh you know uh if a telephonic survey in Uttar Pradesh

uh you know is basically conducted on the issue of petrol prices it will not uh you know uh elicit the representative

responses if the responses are only sought from a section of population maybe extremely rich people uh who for

whom it does not matter or extremely poor people who do not own vehicles. Okay. So if you ask questions from uh

you know uh people who do not fall in your target population who do not represent your target population then

whatever the results of the uh survey that you're going to get is going to be faulty and it's not going to work.

Now to avoid falling prey to these mistakes the you know we can perform simple checks we can maximize the

overall uh you know e efficacy of our survey instruments. Let's see first is the first and the foremost and the most

important thing is what is the hypothesis? What is the research question that you want to go ahead with?

What are the hypothesis in terms of this research question that you have framed? And in that sense are your items in the

questionnaire related to these hypothesis or not. So this basically this loop is the first starting point of

creating any of these surveys. So researchers must and that's basically what the second point is are the

questions included in the survey related to the hypothesis. Okay. So I basically talked about both of them at once that

researchers must take care to ensure that the questions must be focused towards testing the hypothesis else the

generated data will be rendered useless. Okay. So these are the first two and the most important things when you're

designing the survey. Also if the questions are not in sync with the research questions and the hypothesis

and will participants will respond inconsistently uh then you'll get a mountain of data which will be useless.

You will not be able to systematically analyze or interpret them at all. Now uh what kind of cause effect

relation you want to build. So for the successful outcome of the survey research researchers must have a

descriptive hypothesis. For example, you cannot have a cause effect because cause effect is handled best in uh experiments

not here. So the hypothesis that people should have here should be descriptive hypothesis about say for example a

group's characteristics. How many people like wearing a blue shirt? How many people like buying a green detergent?

How many people are comfortable with the petrol prices? Those kinds of things. Okay. And they've basically described

that there are four types of hypothesis that people can have. Say for example snapshot type hypothesis. So sometimes

researchers conducting a survey must be interested in knowing that how many people in a given population possess a

certain uh characteristic or support a certain uh opinion position. For example, a social worker might want to

know uh you know through a survey uh that how many uh you know adolesccents in that community have been exposed to

drugs. So just yes no answer you get a description that okay uh in this particular neighborhood among you know

the age group so to so this percent of people have already been exposed to drugs or alcohol abuse

profiling uh hypothesis for example sometimes researchers typically conduct a survey to uh develop a detailed

profile of certain groups. So you want to know about various characteristics of these groups. Say for example diner and

singleman celigman sorry diner and seligman found that a that very happy college students were more more outgoing

than others. So you want to say profile outgoing students uh college students you can say okay uh you know what are

your activities what are your play types this that and on the basis of that you can get a uh you know detailed knowledge

about different characteristics that uh they have relational hypothesis for example

sometimes you can ask questions about you know how the two relation how two or more variables are related to each other

for example Davis and colleagues use a survey to test the hypothesis that attachment styles you might be aware of

secure or insecure attachment styles are related to sex drive and reasons for having or not having sex. So again you

can basically ask uh have a questionnaire which asks question about one side and the other side and

eventually you can perform correlations and so on which will help you to get this idea. Hasselton and colleagues uh

they use surveys to test their hypothesis that how upset people will be if their partners lie to them. Now,

interestingly, uh they basically uh wanted to check that they said they had this hypothesis that it will be related

to both the type of lie and the gender of a person. They predicted something very interesting. They said research uh

women would be more upset by men lying about their income and men will be more upset by women lying about their past.

Okay, so again something they collected data. I don't know what they found but questions that you can ask. Finally,

Oishin colleagues use surveys to test the hypothesis that beyond a certain level, happiness is not associated with

financial and educational success. So again, you can basically uh you know uh ask questions about financial and

educational success and ask questions about how happy do you feel and then you can basically you know have a broad

description through plots and so on that how are these things related or unrelated to each other.

Finally, you can also have predictive hypothesis. For example, sometimes the surveys might be conducted with an

intention to predict how people will behave in different situations. For example, a lot of times news

organizations perform surveys to predict how people will vote in a certain election. Uh market researchers conduct

surveys to find out which products are more likely to uh you know sell uh quickly and which will you know receive

a lukewarm response. Now remember this that while surveys may be an excellent uh tool to create a descriptive snapshot

cause effect correlational hypothesis uh the surveys are not best suited for them. They can give you the initial data

but the rest of the thing uh has to be done through other methods that we'll talk about in the next lecture. Now uh

there can also be questions about okay uh when we're using this self-report method how useful this self-report

method is uh will it lead us to correct knowledge okay so Nisp and Wilson conducted some research and they

basically provide us some pointers about uh why in a lot of these surveys uh people's self-report might be completely

inaccurate. Let's see participants never knew and will never know the answer to the questions posed. If you ask a

extremely abstract metaphysical kind of question for example sometimes people are in you know by themselves stranger

to their own uh feelings. If you ask them oh why do you like broccoli or why do you like carrot or why do you not

like broccoli? Some people don't like mangoes for example why do you not like a they'll not be able to come up with a

clear answer or at least address an estimate of that. So the question should be posed in such a manner that they can

elicit exact responses and not uh estimations because a lot of things people don't even know about themselves.

Also sometimes see memory is prone to error. Memory is something that fades over time. So it is possible that

people's responses are inaccurate because they do not any longer remember the information needed to answer the

question. For instance, this is interesting. Obese people tend to over uh you know tend to under reportport

what they have eaten. Students tend to over report how much effort they have put in their studies. So again sometimes

it's a matter of time and people forget or people are sort of trying to hide in in in whatever manner. Also something

that you know sometimes it happens that participants don't even know the correct answer of a question at that point. Say

for example if you ask questions in the format what if what would you do if you were in this situation? No uh the person

has not been in that situation and they're not extremely good at predicting future. So these kind of questions

there's an example here that when people were asked what would they do if they were put in milligrams experiment uh

most of them said they would basically not obey uh but actually 60 to 80% of them ended up uh you know confirming and

obeying the commands. So again very interesting also uh sometimes what happens is that

the participants are not giving the correct answer knowingly because they are affected by certain biases. For

example the social desiraability bias participants basically want to appear good. They want to appear likable and

sociable and so on. So uh they would sometimes understate unpleasant things, exaggerate pleasant things and they

would likely uh you know most likely give an answer that makes them feel good. Okay. For instance, if one were to

believe uh the US adults uh survey responses, 40% of the US adults regularly attended religious services in

the early '90s. So this is when the survey was done. This is what everybody said. When this was matched with actual

attendance data in uh you know religious uh uh gatherings or religious uh services basically going to the church

on Sunday and other uh you know prayer ceremonies it was actually found that it was not as high as the people had

reported. It was actually less than half of the reported data. So social desiraability okay uh how punctual are

you? How disciplined are you? Everybody will give answers that will present them as very punctual and very uh you know uh

hardworking and so and so but actualities might be something else. Abiding by demand characterist people

are extremely smart. They know what you want to hear from that survey and a lot of times what happens is people will

give you the answers that you are looking for. So that's why a very impartial framing of the questions and

you know sort of a non-person uh way of delivering questions might be int might be uh useful for example based on the

perceived political alignment of a journalist uh voters will indicate that oh uh you know this party is good or

that party is good. Okay. Also fixed response tend to sometimes people are just lazy. They they choose

that okay I will give this this this and this answer and whatever question you uh ask them they'll basically present a

very similar answer did uh you know displaying a fixed response tendency. They want to answer 1 2 3 4 or let's say

they want to answer four four. So whatever the question is they're not even reading the question. uh they're

just answering four four four collecting all the answers that are with four. So this is again something which minimizes

the efficacy of uh all of this. Okay. Now uh we've seen the strength of these methods. We've seen some of the uh

loopholes, some of the pitfalls uh of this research. But say for example uh do we do this or not? Certainly we do it

and a lot of people are doing survey research and these kind of descriptive tools are being uh you know adopted and

used extensively. But one last thing that is very important uh you know is that who are

you collecting this data for? What who are these results relevant for? Are they relevant for policy decisions? Are they

relevant for uh basic research? Are they relevant for uh you know uh some kind of uh you know change that you want to

bring up that is important. So a most important consideration here is to keep in mind uh that you know the target

population and the generalizability of the sample. If your question is about policy decisions, then the people giving

the survey should be the ones that the policy is going to impact most. All right? If the uh uh survey is about some

medical things, then the medical stakeholders should be the ones uh who should be answering these questions. So

this is something that I'm sure uh people should keep in mind when they are using and designing these instruments.

I think that's all for today. I'll see you in the next lecture. Thank you.