 Hey, hi everybody. It's Monica again. This is part two of the original live stream I was trying to do where I was trying to show you how to go from having a survey spec to actually having a programmed in survey. I'm sorry, forgot to turn my phone off. A completely programmed in survey into a survey software like SurveyMonkey. That's what I was demonstrating was SurveyMonkey. And why did I stop my live stream? Because SurveyMonkey crashed, stopped responding. So SurveyMonkey is an online app. And because it crashed, I was like, what could I do? So I did the stream and then I went and did something and now it seems to be back up, so let's continue. Now just to remind you what I was doing, I'm gonna share my screen and I'm gonna share the spec with you. So in our last live stream, I had already made the spec for a simple survey. It's mainly because I started doing these live streams and I don't really know what topics people want. So I wanted to do a survey that way anyway. And but I thought I'd take the opportunity to create a few different types of questions and program them into SurveyMonkey in front of you because I wanted to show you not only what you have to think about with these different question formats, but also what the data looks like when it comes out. It looks different depending on the question format. And something that I emphasize in the last live stream and I'll remind everybody about now is that it doesn't really matter what survey software you use, like you could use Qualtrics or RedCap. Those are the ones I was talking about. So long as it's survey software, you don't wanna have forms or like Google Forms or anything, you want survey software with those features in it. And as long as it doesn't matter which one you use, but you should get good at it. And you should, in getting good at it, you should also like choose one that you invest in with your business, like if you have a business, like mine. If you're working at like a university, usually the university will endorse like either SurveyMonkey or Qualtrics or you know what I mean, they'll pick something. Sometimes they don't. But in any case, if you work at a university or another organization, you wanna try to use their account, you don't have to pay. But I have a small business so I set it up with SurveyMonkey and so that's what I've been using and that's what I try to use with my customers unless they wanna use something else. And so I'm better with the vocabulary in SurveyMonkey but regardless of whether you use that or Qualtrics or whatever, you have to make a spec or you don't really know what you're doing. Okay, so I'm gonna go back to SurveyMonkey and continue demonstrating here. Let's see, where did that go? Okay, so last part of livestream, part one I guess I'll call it. I was showing you how all of these things work, right? So this is how you go back and forth. Like this is where we designed the survey and this is where we previewed the survey. And this preview, it didn't collect any data but then I showed you how, now remember this is an online anonymous survey and so I have this link here called test for one of the collectors, you remember it's link collectors and live. And so what I did, I filled out one of the surveys here. So now I'm gonna copy this link again and I'm gonna fill out another one here. Let's see here. Now I'm gonna switch to that. Oh, it's a survey not found. Let's see here, copy URL. I must have copied it around here. Why is this happening? For some reason it's not letting me go in here. You know, I think it might be, let's see here, edit collector. Oh, let me, are you not seeing what I'm doing here? Oh, great, now we have those. Let me see here. Let me show you what is going wrong here. I already took the survey once. So this multiple responses is off and so only allow the server to be taken once from the same device. So allow the survey to be taken more than once from the same device. I'm doing this on test. I'm not gonna do that on live. I just wanna do that so I can test it. Okay, because that's why the survey wasn't up. Let me see here. Okay, this time it came up. So let me go and share the screen. Let's see here. Okay, so now we're gonna fill in our test survey. So last time I filled it in, I pretended I was a public health person from one of my personas. That's what from my audience. Now I'm gonna pretend I'm a new data science learner. Like those are people who haven't had a job before. This is their first time. So in which following software applications often these new people are trying to learn like everything. Usually not SAS, Python, SQL, probably not SPSS. They usually aren't interested in these. And then what files do you use most often? Often they're just totally using Python. Okay, so this persona, and again I'm just filling in fake data just so I can have something to demonstrate and also show you this is a good way to test your survey. When you're filling in this fake data, think of the real people, think of real scenarios because that's when you're gonna hit real issues that your respondents might have. So this is totally what you should be doing. Okay, is pretending, making fake people and filling it out for them. So you're testing it. So okay, so my newbie data science learner, I'll tell you what they always want is career planning advice and software demonstrations. And software demonstrations. And they tend to really like, they tend to like educational things, let's see. Yeah, so I'm just guessing this is what they'd say. They'd say Python, but I don't teach Python, but I know they just love Python, okay? So now we're done. And okay, so now I filled out two surveys and we're gonna go back to the Survey Monkey window. And we're on the collector details. Let's go back to the collectors here. Now you can see I have two responses. That's enough. So now what I'm gonna do, so what we just did is, I'm pretending you're on my research team with me, is we just designed the survey. We made the spec, we programmed it in and we put in some fake data to test it. And what this means is we really like the way it's running. It's running like it's gonna run in live and we like the way it's running. So we're pretty much done with the survey front end. Now what I'm gonna show you is what I usually do, you don't have to do it right away, but I like to do it before we even deploy the survey because that's a good time to do this before you start, like when you launch a survey, it's very hard to change it. I've launched surveys before where we screwed up the Likert scale. Like it said, strongly agree, somewhat agree, neutral and then strongly agree, somewhat agree. Like agree and disagree were not like right. And you can edit the labels, it lets you, but you don't know what happened with your respondents. And so it's really important to just, I can tell you a million other things that could go wrong. It's good to just test it and make sure that it works. And make sure it's getting you the data you want because the whole point of SurveyMonkey and the surveys is to measure data, get you data that you're gonna put in R or whatever. So if you don't make sure before you administer the survey that you can download the data and put it in R and make sense of it, then you don't wanna deploy the survey. So this next step I do, technically you don't have to do it before you deploy the survey, but it's like if you wanna be sure, sure, sure that you're getting the data you need, you wanna do it, okay? And what it is is where you go and you analyze, kinda analyze fake data. So we're gonna go to analyze results and download this fake data. Let me see if I got any questions here. Okay, I've got this person here. I'm sorry, I think there was a real chat in here somewhere. Oh, hello Siddharth, nice to meet you. I'm sorry that I don't have any moderators and you had to see that bomber there. But thanks for joining us. All right, so now I'm gonna go back and show you, we're gonna go back to, okay, we're on analyze results. So I'm gonna show you how I download data from SurveyMonkey, but this download process is hairy and it's hairy in every software, but it's a little different, okay? So remember in SurveyMonkey, we have this sort of weird menu over here. I don't really, I honestly don't understand all these things in SurveyMonkey, but I know what I want and that is this download, this arrow means download. So I'm gonna wanna, I'm downloading these two data points, okay? Now what's gonna happen, see how it says export data. What's gonna happen is we're gonna do this export and then we're gonna fill in a form that tells us how we want the data exported. And this is tough. So first of all, we have the choices. All summary data, all responsive data, all individual responses. And how are you gonna know what that means even? Like I just know from experience, I want all responses data, but I can't even tell you why. I don't even know what the other ones are. I know when I worked at a mental health program, sometimes they would do surveys where they wanted to produce a report that just had each respondents answers on it. And so I think that's what all individual responses is, but we're gonna choose all responses data, okay? Then we get this. And so we have summary data and all responses data. We always wanna be over here as, because we're collecting, we don't want anything summed up. We wanna sum it up ourselves. Now the choices we have here are XLS, CSV, SVSS and PDF. And I'm telling you, I just always go with CSV. It's always just easier. As much as I like XLSX and Excel files, you could always convert a CSV to an Excel. A CSV is just easier to use. And then it says data view, current view, original view. I don't even know what that is, I believe it. Then you've got these choices. And again, this is different for every software. You just have to look at the software. So it says columns, it says condensed and expanded. And I'm telling you, when these come out, with the condensed, they're still wide. I don't know what's up with them. So leave it there. Then cells, here's where things get hairy. Do you want the actual answer text? Like, if people say SQL or Argui, do you want it to say SQL or Argui? Probably not. If you're a data scientist, you're gonna want the numerical value. So what's the numerical value? Right, we didn't program that in. That's why we're doing this step, is we've got to figure out how does SurveyMonkey take our front-end data because that survey is basically a front-end and store it as back-end data. We've got to figure that out. Because it's making numerical values, what are they, right? And then it's gonna say, automatically it says file name, and it's a zip file. So I'm gonna show you, so zip file, of course, there's gonna be multiple files in it. So I'm gonna show you what comes out. So we're gonna do export, and then here's what happens. It says your export is started, and it starts this. Now, you know how much trouble I had earlier in the stream because it's crash, right? Sometimes this, I can't get this down. I'll just be honest with you. This has been a problem with SurveyMonkey ever since I started with it, all right? So, but now it says your export is complete, and we can download the zip file. So I'm gonna download the zip file, all right? Let me go back here and see if anybody else's got any questions or anything. Okay, so I've downloaded the zip file, and now I'm gonna open it, and I'm gonna show you what's inside it once I get it open here. Okay, now I'm gonna change to the zip file window here. I don't know if you can see anything in there. Well, I won't show you the window, but I'll tell you what's in it. So it's zipped, all it's zipped has in the zip file is a folder that says CSV, okay? So when I open the folder, I get two CSVs. One is called CollectorList, and I'm actually gonna open that one and show it to you. Let's see here. So this is CollectorList that I've opened here. And actually, let me just make this really big. Okay, remember we have two collectors and they're named Test and Live. Let me make this big. See, it's assigned a collector ID to each of them. And what you'll see is that if I keep the test data collector filled with data and the live data collector filled with data, you know, like people actually fill out the survey, and then I go to download the data, I'm gonna have the test data in here and it's gonna have this collector ID. And of course that's gonna be in the data. So it's important to clear the test collector before you administer it. You know, I mean, you can always remove the test records later, but it's better to just clear it. I just clear the test collector a lot. Right now I'm not clearing it because I'm trying to use this test data to help me understand, curate the data. But I first wanted to show you, this is what happens. So if you make, let's say you make an anonymous survey and you wanna send it on social media. You want it from Facebook or from Twitter, from maybe you have multiple Twitter accounts, whatever. You can make a collector for each and they have a link for each and you just use that link. The respondents don't know. And easily you can download the whole data as one data set and just look at that column, which I'll show you next. I'll show you the other spreadsheet that comes in with that. Let me open it up here. And that's your real data, right? So the other CSV that comes in here, I'll go back here, share my screen. So this is, this is the other CSV and this is the real data, okay? And I'm just gonna scroll, like you can't really see the whole thing. This is a short survey, so there's not that many columns. But one of the things I want you to notice is that there appears to be two rows of header column, okay? And here are our two rows of data, okay? So then, let me close some of these windows. So, and actually, let me make this a little bigger so we can look at the format of this data, because this is the next problem, right? So you can see that this is a CSV and I feel like I wanna format this. I can hardly read it. So I am, I'm gonna save it, we'll just save as, find a place to save it here. And I'll save it as fake data, maybe test data. Well, I saved it in a CSV, but I actually meant to save it as an axle loss because I'm gonna start editing it, right? Or I'm not editing it to keep it edited, but I'm editing it because I want to, why did I just do this? Is I wanna just, see I'm changing it now to an SLX. I just wanna format it and it's to hold the formatting because if you format a CSV, like you make things bold or whatever, it won't hold the formatting. And I just wanna look at it. So I'm gonna go up here. So those of you know me, I use Excel for everything. So I click up here and I'm gonna make all the borders. I'm just formatting so I can read it. And then I've got, let's see, wrap text here. Ooh, that's kind of bad. Okay, let's make this a little smaller here. So you see that these are kind of wide here. Let me, that's making a wrap text like kind of not work very well on here. Oops, let's, let me see if I can do, make this a little shorter here. Okay, so now it's a little bit easier to read. This is still a problem here. Let's make this one wide here. Okay, so see what I'm doing. I'm just formatting this so I can see it. And again, this is just fake data that I'm using to try to figure out the structure that SurveyMonkey is doing with these, okay? Okay, so sometimes what I'll do is I'll, let's make these a little wider, is I'll make this bold cause it's like the top. It just helps me be able to tell, actually remember the top two are headers, so we'll make them bold. Just helps me remind me when the data actually starts. So let me show you what data fields came in here. First, we had a bunch of canned fields, like I didn't make these. You didn't see me make these in the spec or whatever. We have this respondent ID, and this is automatically assigned, here you can see it up here in the, this is automatically assigned by SurveyMonkey which is kind of good. And then here's a collector ID. Remember that big number I showed you from the collector? This will say that this is the test collector. And this is a start date and end date of the, and this looks like it's local time. Remember when I set those options on the collectors to not collect the IP address? That's why this is empty. These are all empty. You can add custom data fields, you usually don't really get it. Okay, so as we go on here, here's question A1. Now remember question A1 was multiple choices. Like which of the following software applications are you currently trying to improve your skills? Select all that apply. And this was the list here. And look at these field names. You know, they're not good field names for SAS or R or anything or SPSS. But they're good field names for humans because you can see what I put. You know, R, GUI, RStudio, SAS, whatever. And remember the first person who filled this out was kind of a public health person. So they were learning R and SAS. But, and so they chose those. So you can see this is kind of odd. SurveyMonkey does this, but it doesn't really bother me because I like that it makes separate column for each of them. Sometimes they don't, you have to separate. Like they'll just have a comma with all of these in one big field, it's such a pain. But this separates them. It's just that it doesn't put like a one in the field when they check it. It just puts whatever number it is. So when I'm processing these things that I'm making flags and stuff later, I'm like, where this is not null. I just say where it's not null, it's checked, okay? So this indicates what was checked off in these different ones, okay? But see this column here? There's nothing in this. Oh, this is none of the above. None of the above is in here, I guess. So I guess if this is checked. But you see, there's two different headers. This is what confused me about this, okay? Then here's A2, which of the following software do you use the most often? Remember that was the drop-down box and there's only one really answer? You know, it's in here, whatever answer that was. Then it says, please rank the following potential general live stream. Now remember, these were the ones, the five things that I wanted them to rank. So you can see that the public health person put academic subjects, because she's busy studying that at school and the career and the data scientist, she put career planning advice because she's trying a good job, right? She put that one. So as you can see, that's how the data are recorded, but how easy is it gonna be for me to analyze it, right? I could just as easily take whatever people say, one the most about, but that's maybe not the right thing because I have these different segments, right? So I might look at the SAS users and see what they want because SAS users are often like public health people or healthcare somewhere. And then I might take the RGUI users and see what they want. But again, this is just the data and then here's our open-ended question, okay? So you see this fake data comes in and it's a little hairy. But one thing that's good is we know that we can get our data out. I know that I, from doing this walkthrough, I know I can figure out what collector they're from, Winston, I probably don't need to know that. I know that I can figure out which ones were checked off here in this question, which one they chose in this question. And again, four and seven, I assume it's the fourth one. Like the, so this was response number four to A2. Let me look at the spec. So that should be Python, okay? And remember how we just, what am I doing here? We remember how we just filled this out, that this one up here is the most recent one. This was the data scientist lady and she wants Python, right? Cause they all want Python. I don't know Python. I guess I got alerted, right? And so I don't know, maybe somebody's got to teach it to me. I'll make my intern teach it to me cause I don't know. She owes me one. But in any case, I can see that, like it says Python here because I was kind of joking. All they want is Python. But I know I chose Python here. So this helps me know how the data land. So this is a big problem in surveys. And that's why I made that big free online course is helping you design the survey and make sure that whatever data you specify in your spec that it's landing in your data. So now that I know that my data is landing and I can curate it, that's the next thing I do. So let me start my data curation. I'm gonna go to new. I don't know what that looks like to restream that I go and make a new file. No, I guess it doesn't look at it. So I'm gonna, actually let me go back to my, the test data before I stop. I just wanna show you the first thing I'm gonna do with this test data when I start my curation is I'm gonna highlight these top two columns, the two header columns, okay? And I'm gonna put copy. And then now I'm gonna go to, I'm gonna go to the next, here's book one. So see, this is just a blank one. So the first thing I'm gonna do is actually, first thing I'm gonna do is just format this a little like I did the other ones and do this wrap text. And I hate Calibri, how is this arrow? Okay, but now I'm gonna go up, I guess I'm gonna go to the second one. And oh, it lost my pace. Let me go back and I'll get, you saw what I highlighted. Okay, I'm gonna go to the second one and I'm gonna go up to paste special. And here's what we're gonna do here is we're gonna paste transpose. I don't know why it just did that. And we're gonna paste values, okay? So here's what it did. It pasted the values from those first two columns starting up here, down here. See that? And so what I usually do then, I'm gonna insert a column here. I usually start and put order here and I just do one, two, three. That means I can sort it and get it back into order if I need to. And I actually will continue on because we're gonna add probably to this. Oh, and you know, I like to make this bold here. And then also I go up here, I'm clicking here. I'm gonna go over view freeze paints. That way if I scroll down, that stays okay. Then what I usually call this is serving monkey header one and SM header two. And then R, I actually, I usually analyze this stuff. I actually am gonna name the R variables here, okay? And I'm gonna name all of them even the ones I'm gonna get rid of because the idea is what I wanna do is I wanna insert a row to replace these headers and just put these in. So onto my CSV so I can read it into R, right? Would that be nice? So I can do like this, REST ID, all ID, SDID, IP. And I'm just naming them so I can read them all in and get rid of them later. Email, even though they have name, surname, custom. Okay, here is where I start creating naming conventions, right? So I have, you know, I'll probably do something like, like this is improved skills. Maybe I'll do SKL none, SKL are gooey. SKL are, I won't go on, but I'll show you why. Let's see here. And then we got this one. Which one do you use right now? Probably maybe I would call it use now. But this is, oh, and I usually call this one main. I should probably save it. So this is a data dictionary, right? So I'm gonna save that here. And I'm gonna say data starter. I'm gonna call it starter, like, you know, yogurt starter or whatever, because I'm not gonna fill in the whole thing. But this is the way I would start it. And here I put main, over here I put, like I usually create a column called values. And like here, this would be like integer. But over here, like this would be, this is the number that, let's see here. If this is not null, they selected. You know, I usually put something like that there. But this one, I'll put the values as being use now or use most or something. And I usually just copy this to get the format in and then erase everything, like just delete, just so I have the format in. And then I'll have the values that I put in, which were our GUI, our studio, SAS. Remember that list? So let's see here, hold one, two, three. And then ask. And remember that was, I'm looking at my spec now. Our GUI, our studio, and SAS. And theoretically, I would fill in all of those. And then that way, when later, when I'm going to analyze the data, here, let's just make this a little wider. When I go to analyze the data, I would know like the values here I called this use now. I know just to go over to this list and see those are the values. So whenever I have a dropdown box, that's what I do. I just under values, I just make a pick list. I call it a pick list. And I just put what the values are over here. And that way, like it's really easy to analyze the data. Okay, so imagine that I had filled in, well, let's actually go through this because like here I would have to do, this one is rank. So I might do R and K software or soft demo. See, I'm using camel case. You can use however, R and K. I'm at a good time then. So you fill these in. Oh, and the other thing I do is I'll put in, I keep adding the variables I keep forgetting. I'll put in source. In this one, I'll just say survey monkey for all these, because they just came in from survey monkey. But that ends here. Let's say that I wanna make a variable in R like usually what I'll do is I'll redo these variables. Like I'll make a new, like I'll call this like flag, like I'll say none flag are gooey flag. And then over here for the values, I'll say like a flag for none, like for are gooey and I'll put R here. And of course I haven't made them yet. I'm just designing them. And I'll say says one if they checked the box, otherwise zero. And this is like little code to myself. I might write this out more in the data picture of sharing it with people. And I need to really communicate, but this tells me that when I get the stuff into R, I'm gonna create a set of flags. I didn't even start naming them here, or I put that in the wrong place. This is where I'm naming them. It doesn't belong under here. Let's see here, flag for none. Oh, I'm putting this in the wrong place. This is the description. Let's see R variable. I don't even have the description. Where's the description? Ha ha ha, that's here. Description. You know, in a survey, you don't even need much of a description, but when you get down here, the variables you're making, you need a description. So what did I just say? Life for choosing none. And so on. And you know, I can always sort this in a different way and then go back to the same order. But this is how I'm gonna document the variables I'm gonna make. And you know, I'm just gonna keep building this out. Now, what I want you to understand or imagine is, let's see here. Imagine that you filled in all of those columns. Then what you could do is something like this. I'm gonna go back to our actual fake data. Let's see here. So remember how we have, like, I'm gonna make this a little smaller here. We have these two headers, right? What we kind of wish we could do is go in here and insert and just copy in another set of headers that had our R variables in it and then get rid of these and then save this as a CSV and read into R. Well, actually you can do that. If you fill in all of these, like imagine all of these were filled in all the way, you know, just down to the native variables. I'm doing copy. Now I'm gonna come over. Oh, you probably didn't see that. Like I copied the native variables. Then I'm gonna come over here, you know, from that column and do paste special. And this time I'm gonna do transpose again and values, but now it's bringing it back the other way, right? See, remember how I filled these first ones in so they showed up here, but these other ones weren't. And so what I'll often do is in the data dictionary, I'll create a tab. I'll create a tab that says headers. At the end, you know, the values tabs with all the picklists and stuff. I'll create a tab that says headers and all it has is that row at the top. That's it. It doesn't have anything else. And that way, when it's time to download the data after everybody spilled it in, then I can go just easily convert whatever it is to CSV, copy that in. And I've already got this data dictionary and I've already designed some of the variables. Like I've written down which variables I wanna make an R. I mean, I'm gonna keep adding to the data dictionary as I go. Another thing I wanna show you, like actually let me go back to my spec, right? So you're building these documents as you go. So here's the spec we had originally. Remember how we're making field names for each of these things? So I just made a field name for this one, right? And I called it school none, right? Or no, this was none of the above with school none. Okay, let's put that in a different highlight. And then I also did skill are gooey, right? And so that's this one over here. And you can make a little like kind of a legend that says, okay, the green is the field names or whatever that I'm picking. And then you remember this one was just to choose one. So what did we do for that one? We did use now. So I could add that to the spec. So these are kind of living documents while you're doing this. It's your keeping track of everything you're doing, especially if you had to change the survey or you had to add something. So the spec is being built now and the data dictionary is being built. And so now you've got, what have you done? You've done, I'm going back from the beginning. We did the spec in Word, then we built it in SurveyMonkey. Then we used the little browse and testing feature in SurveyMonkey, but after we finished that, we actually made a test collector and a live collector, but we haven't used the live collector yet. In the test collector, I filled in two fake people surveys based on like my customers, like what I think my customers would say. I wanted to make it real. Also, that's a good place. If you've got a complicated skip pattern during that testing is a good place to make sure that's working. So now, then we, I finalized the survey, I said everything is fine. So I downloaded the fake data I made and I just documented it. I didn't finish the data dictionary, I didn't finish every the spec, but I just showed you what I would do to finish it. I'll have to finish it offline. Okay, so now we're gonna go back and be like, okay, now is the time we're ready to actually deploy the survey. So I'm gonna go back into SurveyMonkey and say, okay, we're all ready with the survey. The analysts downloaded the fake data. They said that the fake data's good, they can get in an R. So now we're gonna go back. So I'm getting ready to deploy the survey. So I'm gonna go back to the collectors and I'm just gonna clear this collector. How you do it over here is you clear all responses and it's gonna give you a message about it because they don't want you. Now here is live. So I'm gonna copy live and guess what, I just programmed a survey in front of you. I'm gonna put live in the chat here. So this is a real survey. Go ahead and fill it out, okay? And I'm gonna post it, I'll throw it all over the place here and there. You know, it's such a short survey. That way I can get a better idea of what people wanna know from my live streams. And for what I'm doing, maybe it's just good enough for me to use this and just kinda watch, my survey has no responses, but just use this. I might not need to use R to analyze it, but I just wanted to show you this as all the steps from the very beginning to the very end when it comes to programming and deploying a survey, like designing one for online applications, survey applications and then programming it in, testing it and then deploying it. A lot of people don't realize there's like a hundred steps, right? That I just showed you. And this is a little baby survey with no skips or anything. So if the next time you're thinking, oh my gosh, I need to do a research project. Oh, I'll do a simple survey. Just remember this is not simple, okay? If you do research right, it's usually not simple. Like this is the simplest little survey you'll ever see and it wasn't simple. So research is like always a big project. It's like renovating parts of your house or something, oh, it'll take an afternoon and it won't take an afternoon. All right, but the main reason why I did this live stream was, I did this live stream, not a lot of people showed up, but that's okay. You know, if you're watching this later and you got anything out of it and you now know how to do your surveys or whatever, thank you, I'm glad, please like it and please put comments. You know, obviously if I'm making a survey, I want feedback from people. And you might want to follow my channel or find my blog or whatever because I'm always giving tips and tricks about things like this. All right, and lately my customer's been asking that so I thought maybe you guys would benefit from it. All right, well, it's a Saturday afternoon and thanks for hanging in there with me when the survey monkey crashed, but it came back and we could finish our lessons. So have a nice weekend and I'll see you soon. Thanks for showing up.