0:03 all right so we are continuing our
0:04 discussion in this section on
0:07 experimental design and what goes into
0:10 making a really good experiments on our
0:12 slides in this section you guys are
0:14 gonna see a lot more typing and I
0:16 usually like to do on these slides right
0:18 here this is more just for later for you
0:20 guys to have something to look back at
0:23 and read in your notes but I'm gonna try
0:24 to summarize each of these things as
0:26 we're going through them so we are still
0:28 in the context of our caffeine
0:31 experiments from last time so if you
0:33 guys can think again the idea was I was
0:35 gonna see if having caffeine makes a
0:37 difference in your pulse rates um and
0:39 then we were talking about different
0:41 things I was doing or needed to do to
0:44 make sure my results were worthy of
0:46 actually being analyzed to make sure my
0:49 design was good so last time we only
0:54 talked about one aspect only one aspect
0:57 of experimental design and that was the
0:59 idea of having a comparison you have to
1:02 have at least two groups if you don't
1:05 have two separate groups to compare you
1:08 can't say whether the results that
1:10 you're seeing are because of the
1:12 variable of interest or because of some
1:14 other factor my example last time was if
1:17 I give every caffeine I weight and I
1:19 measure on maybe my awesome stats
1:21 lecture in the meantime is what
1:25 increases that pulse rates so I need to
1:27 make sure that I have at least two
1:30 groups that both experienced the same
1:31 thing two groups that were experiencing
1:33 all the same things except for that one
1:36 thing that matters which is Kathy so I'd
1:38 have a group that got caffeine and I'd
1:40 have a group that did not that all heard
1:42 the same lecture so if my lecture was
1:44 really what was going on I would see
1:47 both groups having a high rate and if
1:48 there isn't still a difference it would
1:50 be because of the caffeine not because
1:52 of that factor that was the same between
1:56 them but even if I have two groups if
1:58 I'm not careful about how those two
2:00 groups are made up I could still run
2:03 into issues so I need to make sure that
2:06 not only do I have two groups but that
2:09 those two groups are evenly balanced we
2:12 have to do what is called R and
2:18 assignments so we randomly assign
2:21 subjects to treatments to create groups
2:24 that are roughly equivalent at the
2:26 beginning of the experiments underline
2:28 and put a star by that line roughly
2:32 equivalent groups so let me talk about
2:35 what this means let's say I was gonna do
2:37 the caffeine study and I realized okay
2:39 yeah god I have two groups got I have a
2:41 comparison but then I just let people
2:44 self assigned to their own groups it
2:47 seems plausible to me that if I just let
2:49 you guys assign to your own groups
2:50 you probably wander over to the group
2:52 whichever group you end up being and
2:53 maybe you wouldn't even know what group
2:55 it is a or b but like you would probably
2:57 migrate with your friend group -
2:59 whichever station you're gonna go to so
3:01 people who are friends in the same group
3:03 would probably go over to the same
3:06 treatments well let's say you and your
3:07 friend group are just going to get
3:09 coffee every single day before school
3:12 you're getting a triple espresso mocha
3:14 whatever and you're just really really
3:16 tolerant to caffeine and all of your
3:17 friends go there together so you guys
3:20 are all pretty tolerant to caffeine um
3:22 so you're all in the same group right
3:23 now and you probably would experience
3:28 almost nothing due to the fact that the
3:30 you're used to the caffeine so what
3:32 could that do to my results if you guys
3:34 work in the caffeine group I would see
3:36 that me like wow caffeine actually
3:38 didn't make a difference at all cuz
3:40 you're so used to it or if you were in
3:44 the non caffeine group um then I would
3:45 be like really confused with what's
3:47 going on or on the reverse let's say you
3:49 never F caffeine your friends are all
3:51 anti caffeine and you guys get the
3:53 caffeine I'm gonna be like what is going
3:54 on here this caffeine makes such a big
3:56 difference we could talk about
3:59 athleticism on people who are physically
4:02 fit going into the same group etc etc if
4:04 I let you choose your own groups that
4:07 concept of confounding comes back I
4:09 won't know if I put you let you choose
4:11 your own groups if it's your caffeine
4:13 status or what blah blah or athleticism
4:16 or blah blah blah or whatever so you
4:17 need to make sure that when you make
4:19 your groups you're splitting it up so
4:21 you have people who have caffeine
4:23 tolerance in both groups so you have
4:25 people who are athletes in both groups
4:28 and the only way to do this just like we
4:30 talked about last section you can't
4:32 trust yourself as an experimenter to
4:35 break that up you have to trust a random
4:38 process the only way to be sure that it
4:41 wasn't because of this or this or this
4:43 or this that are confounded with your
4:45 variable is to break your people up
4:48 randomly why does that matter if they
4:50 were randomly assigned you should get
4:52 about the same number of athletes the
4:54 same number of caffeine people in each
4:57 group okay is it possible that you'd get
4:59 a few more in one than the other well
5:01 yeah it is it could work out that way
5:04 but if you have enough data you're gonna
5:06 be okay and basically what random
5:08 assignment does is it stopped people
5:10 from being like oh did you consider this
5:13 or oh what about this oh did you think
5:15 about what people are eating for
5:16 breakfasts and you can be like hey I
5:19 randomly assigned treatments so random
5:21 assignment is like a kind of just quick
5:24 boom Hey I did I did random assignments
5:28 I'm good so let's move on to our next
5:30 the principle of experiments in design
5:32 Oh actually wait I forgot I did this
5:36 first so let's talk about designing
5:39 experiments here and the concept of
5:43 completely randomized design when you
5:45 have a completely randomized design in
5:47 your experiment it's basically what you
5:49 probably think of or what I just
5:51 described on my last slide where I break
5:53 you guys up into groups randomly if I
5:55 asked you to describe a good experiment
5:57 there's a pretty strong chance it would
6:01 be this kind um kind of a difference
6:03 type of experiment that we will talk
6:05 about later in this chapter is blocked
6:07 experimental design a randomized block
6:10 design and we'll get to that but this
6:13 generalized randomized design is when
6:14 your experimental units are assigned
6:18 treatment entirely by chance and you
6:19 will be asked sometimes on the AP test
6:22 to describe a randomized design either
6:25 by like a lists or by making a diagram
6:27 and I prefer writing down like a
6:28 bulleted list like boom-boom-boom this
6:30 is what I would do but I'm actually
6:33 gonna do it via diagram right here so at
6:35 the time I'm making this video I don't
6:37 know who all is in I don't know the
6:39 number of kids in my class yet so I'm
6:39 gonna say there
6:42 are 18 of you guys who are in my
6:45 hypothetical caffeine study I start out
6:49 with my 18 students and the first thing
6:51 I need to do with my 18 kids is I need
6:53 to break you guys up into the treatment
6:55 groups so I need to figure out somehow
7:00 how to do random assignments that could
7:01 happen a lot of ways I could give you
7:04 guys each a number from a 1 to 18 and I
7:06 can randomly picking 9 numbers no
7:09 repeats those 9 numbers those people go
7:10 in the first treatment group then the
7:13 remaining 9 go in the net I could
7:15 instead take index cards or papers put
7:17 your names on the papers put them in a
7:18 hat shake and let pull out in the name
7:20 and there's nothing that says you have
7:22 to do the same a lot of people in each
7:24 group but unless there's a compelling
7:26 reason to do otherwise that's what I'm
7:28 gonna do right here so I'm gonna use
7:29 some sort of random assignment and I'm
7:31 gonna break you guys up into two groups
7:37 of 9 so we're gonna Group one um and the
7:39 variable n is one we use a lot and his
7:40 sample size so I'd have nine people in
7:42 this group right here so it's the size
7:45 of my group and idea of group two which
7:49 would also have nine people in it each
7:50 of those groups is then gonna get their
7:53 treatments so the treatment in this case
7:54 it doesn't really matter which group is
7:58 which say treatment one is gonna be the
8:03 caffeine group so I give you guys
8:06 caffeine and then treatment to you guys
8:08 are not gonna get caffeine we're gonna
8:10 talk later in the slides about how
8:12 giving like a caffeine free soda is
8:14 better than giving no soda at all and
8:16 we'll talk about that later but for
8:18 right now I'm just gonna write down no
8:19 caffeine and leave it at that
8:23 and after we actually get our data so we
8:26 do the caffeine we do the no caffeine
8:28 the part people always forget when
8:30 describing experiments is that you need
8:31 to make sure you actually compare the
8:33 results between the two groups so
8:40 measure pulses measure your pulse change
8:47 and then compare your two groups the
8:48 whole point of doing the experiment is
8:51 to pit with this group against this
8:53 group to see if there's actually a difference
8:53 difference
8:55 in pulse rate so once you collect your
8:57 data you need to make sure you actually
8:59 compare the two groups to see if there
9:03 is a difference alright so that is an
9:06 example of an experimental diagram right
9:07 there um you could have done that same
9:09 thing by making like a little list of
9:11 things to do and I like that better but
9:13 now you've at least seen one of those
9:15 diagrams let's move on to our next
9:18 principle of experimental design which
9:22 is a concept called replication so
9:26 replicate up replication is what I want
9:29 you guys to write on this blank up here
9:32 on the top and replication in a nutshell
9:34 means making sure you have enough data
9:36 you got to make sure you have enough
9:39 data when you do a study let's say I
9:41 have 4 people so I only have 4 people
9:44 that I'm gonna do my experiment on if I
9:46 only have 4 people and I randomly break
9:48 them up into two and two there are so
9:50 few people that it's totally plausible
9:52 that these two are athletes and these
9:54 two are not these two like coffee have
9:55 it all the time these two never have
9:58 caffeine when your groups are too small
10:00 random assignments doesn't get to kick
10:03 in and do its job you gotta have lots of
10:05 people and if you have lots then when
10:07 you break it up the chances of one group
10:09 being drastically different than the
10:11 other very very minimal but when your
10:13 groups are too small you could have a
10:15 flukey thing where this group is
10:17 actually different in some way than this
10:18 group over here
10:21 so having too little data is a huge
10:23 issue in experiments and honestly 18
10:25 people and class really probably isn't
10:27 enough for good study no scientist is
10:30 going to publish my study because we
10:32 just don't have the data at onions to
10:35 make it work in addition to not having
10:37 enough data points the other thing going
10:40 on with replication is it needs to be
10:43 possible to repeat the process not only
10:46 in like your one trial but to take the
10:48 whole thing and to do it again those of
10:50 you who are involved in science research
10:52 you've probably heard the concept of
10:55 replicating results in an experiments
10:57 basically if I'm a scientist and I do
10:59 some crazy study I find just the most
11:02 amazing results if other researchers
11:05 can't repeat my process and produce the same
11:06 same
11:09 it's my results are worthless okay you
11:11 hear about like like I feel like in the
11:13 news all the time there are studies
11:14 about the next new one your drug and
11:16 this awesome thing that's gonna solve
11:17 all these problems and you hear about it
11:18 in the news but then it kind of goes
11:22 away um it's something where the results
11:24 worked at the one time but they couldn't
11:26 be replicated reliably so that's a
11:28 problem that you need to you need to be
11:30 able to do it again and again for the
11:33 results to actually be significant so
11:35 when we do have good replication it can
11:37 help us feel more confident more
11:39 confident applying the results of our
11:44 experiment to a population all right so
11:46 let's keep things moving yeah and talk
11:48 about our next principle of experimental
11:50 design and this is a huge wall of text
11:53 right here so you can pause me if you
11:54 want to to read through this or read
11:56 through it afterwards but you read it at
11:58 some point right here I'm gonna give you
12:01 guys kind of a summary though of this
12:05 principle which is called control so if
12:07 you can go ahead and write control on
12:10 that top link I will now kind of break
12:13 down what that means control is a really
12:15 big concept and it's not the same thing
12:17 as having a control group control group
12:19 is like a buzz word oh you got to have a
12:20 group that doesn't get anything at all
12:23 and then people just think that's what
12:24 this is about
12:26 control is a much bigger issue than just
12:29 that what it basically means is making
12:32 sure everything is the same between your
12:34 treatment groups except for that one
12:37 thing that you're actually testing okay
12:40 so let me give you several examples of
12:41 that and you guys can see on the slides
12:43 right here there are two primary
12:47 benefits to control one is that it
12:49 prevents compoundings and one is that it
12:51 reduces variability so I'll talk about
12:54 those two things separately let's start
12:56 out with confounding this first one
12:59 right here so um still haven't gotten to
13:02 why but it's not a good idea for me to
13:04 give one group caffeine and one group
13:05 absolutely nothing
13:07 it's the placebo effect which we'll talk
13:09 about later on in these slides right
13:11 here so let's say I have these two
13:13 groups and I have the caffeine soda and
13:15 I've done this before actually I've gone
13:17 to Chinooks or whatever and bought some
13:20 caffeine Pepsi and then all I could find was
13:20 was
13:24 Diet Pepsi decaffeinated so that it's
13:25 harder to find and all they had in stock
13:28 to his diets well then I've got caffeine
13:31 regular Pepsi and I've got diet
13:33 caffeine-free caffeine Pepsi has sugar
13:36 in its diet does not have like
13:40 artificial sweeteners so everybody who
13:42 got caffeine also got sugar everybody
13:44 who didn't get caffeine also got like
13:47 fake sugar so if I see a difference in
13:50 my groups its confounded with sugars
13:51 confounded with caffeine I can't break
13:53 those two apart and if I see a
13:54 difference I don't know if it was the
13:57 sugar or if it was the caffeine that
14:00 made that happen so it's dangerous when
14:02 you have studies to let more than one
14:04 variable run wild because if you do that
14:07 you can't say for sure which of the
14:09 variables is actually responsible for
14:12 that difference okay so that is one
14:15 factor you definitely need to consider
14:18 the other kind of thing going on with
14:21 control in addition to like compounding
14:23 and having that going on if you don't
14:26 control four variables you will have
14:30 higher variability in your responses so
14:33 let's say I decide that I'm gonna do my
14:35 caffeine study but I'm not gonna specify
14:37 how much caffeine you have you guys just
14:39 go wild take however much so so do you
14:40 want you want to drink like this much
14:42 that's cool you want to drink a whole
14:44 two-liter have at it it's just up to you
14:48 you decide well then logically the
14:51 people who had more caffeine if you
14:53 drank a whole two-liter during that time
14:56 period you're gonna experience more of a
14:59 change presumably if you drink your soda
15:01 really fast so instead of just on the
15:03 health of soda being the same if you
15:04 just chug your soda I don't know maybe
15:06 that makes a difference and makes you
15:09 feel the caffeine more than if you sip
15:13 it gradually other things at play if
15:14 like certain cans of soda are at
15:16 different temperatures maybe the colder
15:20 soda lowers your pulse or whatever I'm
15:22 recording this video summer of 2020 so I
15:24 don't even know yet if I'm gonna be
15:26 seeing you guys in person for this
15:27 lesson or if it's gonna be like a
15:30 virtual deal if like some of you are at
15:32 home doing this study and you've got
15:34 music playing in the background
15:35 that's pumping you up and some of you
15:37 don't or somebody's yelling in your
15:39 house there are lots of different
15:41 variables that I wouldn't be able to
15:44 control for and those variables changing
15:46 will affect responses and just make my
15:48 results more spread out than I want them
15:52 to be okay so this slide talks about
15:55 that in a bunch of detail so let me
15:58 expand like um kind of explain a little
16:01 more that second thing about reducing
16:03 variability and why we want lower
16:06 variability in our responses so this dot
16:09 plot on the next page has two
16:12 hypothetical scenarios this is focusing
16:14 specifically on the amount of caffeine
16:16 so making sure that I keep the caffeine
16:19 consistent for everybody so in my first
16:22 little dot plot right here these people
16:24 were given the same amount of caffeine
16:26 and these people were given the same
16:28 amount of like let's go with caffeine
16:32 soda and then diet or not diet sorry
16:34 caffeine free same soda want to make
16:36 sure they drink the same amount of soda
16:38 not more or less depending if you have
16:40 caffeine or not same temperature same
16:43 everything same rate of drinking only
16:45 difference is these guys got X amount of
16:48 caffeine you can see in the results of
16:51 this study that these people over here
16:55 have positive changes in their pulse now
16:56 when you talk about a change in the
16:59 pulse I'm assuming they took the after
17:03 minus the before although didn't super
17:04 specify right here but that wouldn't
17:07 seem logical to me so if my pulse beats
17:09 for you times and however long that time
17:12 forward afterwards and it was 35 before
17:15 that means it changed by five so this is
17:17 the difference of their pulse rates
17:19 right here these numbers for the
17:21 caffeine europe are mostly higher than
17:23 these numbers right here for the new
17:25 caffeine group so I would look at this
17:28 and be like oh man caffeine made a
17:29 difference that's what this would lead
17:32 me to believe so it seems like caffeine
17:39 seems to make a difference okay that's
17:41 what I would think if I saw this these
17:42 results right here
17:45 meanwhile in the second scenario this
17:47 one kids were allowed to choose their
17:50 my caffeine so if you look at this right
17:51 here what you're gonna see is more
17:54 variability or spread in your results
17:55 results are a lot more all over the
17:58 place this kid over here probably drank
18:00 a two liter of soda where this kid was
18:02 probably sipping on just a little bit
18:03 right there so if caffeine actually
18:05 makes a difference you see a lot more
18:09 spread in your results okay spreading
18:12 your results is not desirable you don't
18:14 want that if you can avoid it and the
18:16 reason for that is the first one it was
18:18 pretty clear that this group is bigger
18:21 than this one but in the second one look
18:23 at how many of these people experience
18:26 very similar results these people on
18:28 these people got about the same results
18:31 right here there are a few people on the
18:33 high end and a few people on the low end
18:35 right here but if you have like
18:37 basically that makes this data less
18:39 relevant look at how many fewer data
18:41 points we're looking at now is opposed
18:44 to everything as a whole so if we do
18:46 that picture it's like one of these two
18:47 these people who are not in this study
18:49 anymore all of a sudden you really can't
18:51 tell that this group is better than this
18:54 one so when we look at that plots like
18:56 these we don't want to see a lot of
18:58 overlap not a lot of overlap means
19:00 there's probably a difference lots of
19:01 overlap between that means it's harder
19:04 to see what's going on so if I saw these
19:07 results I would be unsure of what was
19:09 actually happening okay
19:12 so you have to have control to lower
19:14 control allows you to lower variability
19:16 in responses if you don't control you
19:18 get increased variability which makes
19:21 things unclear I have one more question
19:24 on this scenario right here would
19:25 students weight be a confounding
19:28 variable in this experiments so what do
19:31 I mean by that weight possibly seems
19:33 like it would make a difference in no
19:35 change from your pulse rate if you weigh
19:37 more the caffeine that you have if
19:38 everybody's giving the same amount of
19:40 caffeine that caffeine probably isn't
19:41 gonna have as big of an impact on you
19:43 because you've got more mass for the
19:46 like caffeine to spread out across when
19:48 you weigh more in general things have
19:50 less of an effects on the other side of
19:52 things if you don't weigh very much that
19:54 same amount of caffeine may hit harder
19:57 for you and have more of an effect so
19:59 the question was is caffeine a
20:01 confounding Miriam
20:03 now remember what it means to be a
20:05 confounding variable it means that you
20:07 have these two factors that are tied
20:10 together that you can't separate to say
20:12 which variable is actually causing the
20:14 change that you're seeing so in this
20:17 context it would mean that weights and
20:20 caffeine are tied together in a way that
20:22 I can't tell which one made the
20:24 difference that's not what's going on in
20:28 this study I randomly assigned people to
20:30 treatment at the beginning so if I
20:32 randomly assign people to treatments
20:35 people who more should be balanced
20:36 between this group in this room who
20:38 don't weigh as much balance between this
20:41 group and this group random assignment
20:44 is basically our way of ensuring that
20:47 variables don't get links like that and
20:49 we can actually tell what's wrong not so
20:53 would weights be a confounding factor no
21:06 treatment was randomly assigned so sorry
21:09 treatment was randomly assigned so
21:18 weight should be balanced the balance
21:26 I just can't write between groups so if
21:29 you randomly assign treatments you avoid
21:31 that possibly you can sidestep the worry
21:33 of oh gosh which of those variables
21:36 wasn't the group should be even except
21:37 for the caffeine the thing you actually
21:41 care about so let's keep things moving
21:43 on our slide right here and the next
21:45 thing we have to talk about is more text
21:45 right here
21:47 I've already referenced one of these
21:51 things so I'll remind you guys again um
21:54 when you do a survey it's important that
21:57 subjects in both groups are blind to
22:00 what treatment they receive now I'm
22:02 gonna put blind in light quotes right
22:03 here this doesn't mean like physically
22:07 lines um the idea is that if I'm doing
22:09 this caffeine study I shouldn't know if
22:11 I'm getting caffeine you're nuts if I
22:14 know I'm getting caffeine my pulse may
22:14 actually start
22:17 to raise just because I expect it should
22:18 your mind can actually play tricks on
22:21 you would cause real responses by in
22:23 your body just because you think
22:25 something is supposed to happen that's a
22:29 very famous concepts called the placebo
22:33 effects so if you are not blind to
22:36 treatments you run the risk of being
22:38 fooled by the placebo effect if somebody
22:40 expects something to happen it may
22:42 actually happen even if there wasn't
22:44 caused by the caffeine you may have a
22:46 higher pulse just because you think
22:48 you're supposed to that can happen with
22:49 medications too even with like crazy
22:52 things like it can be very concrete
22:54 things like anxiety but also like I
22:56 don't know some sort of illness that you
22:58 have getting better faster your mind
23:00 actually is very powerful and just
23:01 because you think you're getting
23:04 treatments it may actually cause like
23:06 you to get better just because your mind
23:08 convinces it to happen so you want to
23:09 make sure you don't know what treatment
23:12 you're getting as a person involved in
23:15 the experiments sometimes it also makes
23:18 sense for the researcher to not know who
23:21 gets the treatments and in that case it
23:26 is called a double-blind study so when
23:28 the researcher doesn't know somebody
23:30 knows if nobody knows what's going on
23:32 and how are you ever gonna find out what
23:33 treatment you got it's like the results
23:35 are lost to you so there's somebody
23:36 behind the scenes or it's recorded
23:38 somewhere but let's say I'm the
23:40 researcher sometimes it may be
23:42 beneficial for me not to know either in
23:44 the caffeine problem it doesn't super
23:46 matter if I know where if I don't what's
23:48 going on because I can't have any sort
23:50 of influence on the results we're
23:53 counting your pulse rates um but let's
23:56 say instead I was measuring how relaxed
23:57 or help calm you looked or how anxious
23:59 you look there's something like that if
24:02 I knew what treatments you got I may
24:04 subconsciously be like oh they got the
24:05 good treatment so yeah they do look more
24:08 relaxed or they look more anxious and my
24:11 own biases can kind of cloud my
24:13 judgement right there so you want your
24:15 study to be double-blind not all the
24:17 time because it's not always helpful but
24:24 it's useful when the researcher must be
24:28 subjective so when the researcher must
24:30 the subjective you want to make sure
24:32 they're blind so they can't have their
24:35 own bias he's coming to play all right
24:38 couple more disclaimers here not every
24:40 single experiment needs to have a
24:42 control group that gets no treatment
24:44 needs to have a placebo as long as you
24:47 have a comparison so people mistakenly
24:49 say oh not a good experiment there was
24:50 no control group
24:54 there was no placebo um basically let me
24:56 give you a few examples of why that may
24:58 be the most important thing is that you
24:59 have a comparison you have to have
25:02 multiple groups but you don't always
25:03 have to have a group that gets no
25:07 treatment at all okay so let's talk
25:10 about medication if I'm testing a new
25:12 cholesterol medication and I want to
25:13 know if this new medication is better
25:16 than what's already out there it's kind
25:18 of unethical for me to make people who
25:21 need medication having no medication at
25:22 all hey guys you're not taking any
25:23 cholesterol medicine I'm giving you fake
25:25 pills so you won't know what's going on
25:28 that's not right so instead of testing
25:31 new pill versus no treatment at all
25:34 you can test new pill against treatment
25:36 that's already used and is already
25:39 commonplace so you can test the status
25:41 quo against new treatments it doesn't
25:43 have to mean new treatment against
25:46 nothing at all and even easier example
25:47 let's say I'm testing two kinds of
25:49 paints and I want to know which one does
25:52 better and feigns less in the rain I
25:54 don't have to have a section of pavement
25:56 with no paint at all before I do my
25:57 first paints in my second paint I can
26:00 just do first verse ii have to have a
26:03 comparison don't always have to have a
26:06 group with no treatment at all and this
26:08 last part right here as long as you have
26:10 these things that I'm talking about the
26:11 big ones that we talked about our lesson
26:14 here in your study um your group should
26:16 be about identical and then you can get
26:19 actually to testing what you want to
26:22 test so if you see changes in your
26:24 variable in your response variable then
26:27 you can begin to talk about what whether
26:29 the explanatory variable actually cause
26:31 those to happen now it talks at the end
26:34 here about there being two possibilities
26:36 and stuff we'll get to those later on so
26:38 there's a little disclaimer for that but
26:39 all this bottom part right here is
26:41 saying is if you do a good job
26:44 designing your experiments then you can
26:45 actually begin to talk about cause and
26:47 effect talking about whether the
26:49 explanatory variable cause the
26:52 difference you are seen so to close
26:56 things out in this section No close
26:58 things out in this section I have just
27:00 really quick here my little thing moves
27:03 but it's just a summary of the four big
27:05 principles of experimental design so
27:07 these four things that are written right
27:08 here make sure you read through these
27:09 make sure you understand them and
27:11 anytime you're designing your own
27:13 experiments these are all things you
