YT YouTube2Text

YouTube2Text → Transcripts → RM lecture correlational and quasi experimental research with narration

RM lecture correlational and quasi experimental research with narration — Transcript

by DrSamFiala · 5,050 words · 763 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:01welcome to the lecture for research
0:03methods uh be talking about
0:05correlational and quasi experimental uh
0:08research uh first of all it's important
0:10to kind of distinguish how these these
0:12types of research these
0:13research designs are different from
0:16experimental research which we've
0:18already um talked about a good bit in
0:20terms of correlation research uh it's
0:22different from experimental in terms of
0:24you don't really do anything to the
0:25participants there's no uh manipulation
0:27of an independent variable You're simply
0:29um measuring
0:32um variables traits or characteristics
0:35of your participants you're not uh doing
0:38anything to
0:40them uh in quasi exper research um
0:45things may uh be done to participants
0:48but you can't assign individuals to
0:51levels of the independent variable if
0:53this happens with quad experiment
0:55research it can also be sort of like
0:56correlational where you're not doing
0:58things like if you're looking at um
1:00difference between are there gender
1:02differences on some
1:04trait some would argue that that's quasi
1:07experimental um I would argue it's close
1:09to really correlational correlational is
1:10really when you're trying to do an
1:13experimental type thing and you're
1:15trying to causing effect but you just
1:16can't assign individuals to level to
1:18levels of the variable um for some
1:21particular
1:22reason and for both of these you cannot
1:25directly infer causation from your
1:27results all you can say is you know
1:29either there is a relationship or there
1:31is a difference between groups dep if
1:34you're doing correlational or quasi
1:38experimental a little bit uh more
1:39closely correlational research uh the
1:42goal is to identify the degree to which
1:44uh variables are related you know so you
1:47know does a person's amount of u x tell
1:51you anything about the amount of Y that
1:53they will have or show and is there a
1:55relationship between these two variables
1:57um for example is there a relationship
1:59between level of educ
2:00and
2:02income um so when you're talking about
2:05these relationships generally thing you
2:07know is it a strong relationship and if
2:10uh if you have a strong relationship
2:12then knowing a person's level on some
2:14variable and say variable X if you know
2:17that then you're confident you could
2:19predict their level on variable y right
2:22if there's a strong relationship between
2:23X and Y also if a person's level on
2:26variable X changes a particular amount
2:28then you're confident you could predict
2:30how their level on variable y would
2:32change um or would have changed as well
2:36an example would be uh if you're
2:38measuring uh intelligence get an IQ
2:40score from the the Whisk right a child
2:44measure of intelligence uh and then also
2:47looking at that same person or same
2:48group of people's IQ scores on the
2:50Woodcock Johnson just a different IQ
2:52test they're pretty similar they have
2:54people doing similar things people who
2:56score high on the Whisk will score high
2:58on the Woodcock Johnson people that
2:59score low on the Whisk will score low on
3:01the Woodcock Johnson um and these are
3:04really two different measures of the
3:05same thing so not surprising that you
3:07should expect a strong relationship
3:09between scores on those two
3:10tests uh things may be related but may
3:13not be that strong of a relationship
3:15right could have a weak relationship um
3:17this is whenever if you know a person's
3:19level on variable X and you have some
3:22idea uh what their level on variable y
3:25would be but the confidence and accuracy
3:27of that prediction is limited example be
3:30uh IQ score on the Whisk and um level of
3:33creative achievement they're somewhat
3:36related but there are plenty of really
3:37creative people that have pretty high IQ
3:40scores and some that have fairly low IQ
3:42scores there's enough of relationship
3:44there that we can say yes there is a
3:46relationship but it's not a real strong
3:48one so knowing that knowing um about one
3:51variable doesn't tell you a whole lot
3:53about um the other
3:56variable excuse me uh and then also
3:59there's possibility there could be no
4:00relationship between um two variables um
4:04so knowing a person's level in variable
4:06X doesn't tell you anything about what
4:08their level in variable y might be
4:10example would be IQ score on the Whisk
4:13and weight there's no correlation
4:16between those two variables um no
4:19differences in intelligence or
4:21systematic differences in intelligence
4:22related to differences in
4:25weight um okay when we're looking at
4:27correlation resarch
4:30uh we can look at the relationship
4:31between two variables or among multiple
4:34variables depending on the exact type of
4:36study you're looking at uh either way
4:38you're doing the variables that you're
4:40looking at can fall into one of two
4:41categories predictor variables and
4:43Criterion variables so predictor
4:46variables um this is the variable the
4:49variables you're predicting with right
4:51so let's say if you know X and you want
4:55to predict y then X is your predictor
4:58it's the thing you know predicting with
5:02uh example if you're trying to predict
5:03um which NBA team um will will win uh
5:09the championship based on the average
5:11height of players on each team so I'm
5:14trying to predict who's going to win
5:15based on height of the players so
5:18average height of the players is my
5:20predictor
5:21variable um and the Criterion variable
5:25um is the outcome you're predicting
5:28right so in the pr example which team
5:30will win that's what I'm trying to
5:32predict that's the Criterion the
5:35outcome um with correlation resarch we
5:39we're looking at um you know
5:41relationships and there are different
5:42types of relationships kind of
5:43mathematically speaking uh we can break
5:45it down into um two uh two basic types
5:49uh the most common we look at in terms
5:51of the stats that you probably are
5:53familiar with would be looking at linear
5:55relationships between uh
5:58variables um
6:00this is one thing way you think of is if
6:02you chared the relationship between two
6:03variables the relationship would look
6:05like a line right so changes in variable
6:07X will will lead to a set amount of
6:10changes in variable y right and it's
6:12always the amount of change is a
6:15constant um so you know IQ score in the
6:18Whisk and IQ score in the look like
6:19Johnson it's a linear
6:21relationship um but the might also be
6:24interested in curval linear
6:25relationships and Le at least two types
6:28of Curves we think about uh kind of an
6:30asymmetric uh curve think about one that
6:34um uh starts uh kind of flatten and gets
6:37more steep or starts Steep and then gets
6:40flatter uh when it's that either way
6:43basically um you have the changes uh
6:46well one example would be where
6:48initially small changes in X correspond
6:51to small changes in Y at low levels of X
6:54but as you get to larger levels of X
6:57then small changes in X correspond to
6:59large changes in y an example of an
7:02asymmetric curve linear relationship
7:03would be um relation between amount of
7:05stimulus and The Sensation uh it
7:08produces so this is a example from the
7:10web looking at as the stimulus increases
7:14how do you um what do you how how much
7:17do you notice the the difference right
7:21and it's not a straight linear
7:22relationship it's a curval linear
7:25relationship uh the other type of curval
7:27linear relationship
7:29would be U any type of kind of symmetric
7:32curve uh this happens when this happens
7:35the direction of the relationship
7:36changes depending on the value of x
7:39example show here is um classic one of
7:41anxiety and performance so at low levels
7:44of anxiety there's this positive
7:46relationship right where at low levels
7:47of anxiety you start to increase anxiety
7:49performance gets better but to a point
7:52then you know this kind of dimin returns
7:53where now at higher levels of anxiety as
7:56anxiety continues to increase
7:58performance then begins to
8:04decrease and depending on what type of
8:07relationship um you're looking at will
8:09impact what statistical test you use um
8:12here I kind of focus more on just linear
8:14relationships even among linear
8:15relationships there's different types of
8:17test the most common we think about
8:18would be B variate correlations right
8:20looking at the relationship between just
8:22two variables um and the type of byar
8:25correlation you use depends on what form
8:28the data in most common both variables
8:30are um continuous right when that
8:33happens use a Pierce and r u there are
8:35different correlations if your data um
8:37are ordinal um or in any other kind of
8:40format you might have different um
8:42different correlational statistics but
8:44the most common one again is that the
8:45pier and R and it generates they all
8:48generate a correlation coefficient right
8:50that R which ranges in value from -1 to
8:53one and both -1 and one represent
8:57perfect relationships right
9:00where um we have perfect predictability
9:02if we know uh there's a constant an
9:06expectable level of change uh change in
9:09X corresponds to an expectable level of
9:11change uh in y um so the two things are
9:15perfectly perfectly related as that
9:17correlation gets closer to zero away
9:20from the ends of negative 1 and one it
9:21draw us to the middle the correlation
9:24gets closer to r equal Z that indicates
9:26there's no relationship right at least
9:29no
9:30linear relationship or system there
9:32changes in one variable aren't
9:33systematically related to changes in
9:35another
9:36variable uh so looking at the
9:39correlation coefficient the sign tells
9:41you the direction of the relationship a
9:43negative correlation uh is a negative
9:46relationship uh where as values you
9:48increase on one you it decreases on the
9:51other or if you decrease on one it
9:54increases in in the other you you can
9:56look at it either way just think of as
9:58any kind of inverse
10:01relationship like um if the the amount
10:04the amount of
10:05time uh you spend uh watching TV and uh
10:09your GPA should be inversely correlated
10:13where people who watch more TV score
10:16have lower gpas and people who have
10:17lower gpas watch more TV right um and if
10:23it's a positive correlation you know
10:25above zero positive correlation that
10:27means that the co vary in the same
10:30direction so as one increases the other
10:32increases or as one decreases the other
10:34decreases so study time and GPA should
10:37be positively correlated the more you
10:39study the higher the GPA the less you
10:41study the lower the
10:43GPA so the sign tells you the direction
10:46the further away it gets from zero
10:49um the the stronger the relationship in
10:52terms of kind of really quantifying how
10:55strong the relationship is we look at
10:57something called the coefficient of
10:58determin ation um which for this is just
11:02the square of the R so if you have a
11:03correlation of .5 the coefficient
11:06ofation is .5 * .5 which is equal to
11:090.25 and that's helpful because the two
11:12things have a correlation of
11:130.5 the coefficient termination is 0.25
11:17which is 25% right if we convert it to a
11:20percentile and that's significant
11:22because with a correlation of 0.5 that
11:24means that the variability in X accounts
11:28for 25 5% of the variability in W so if
11:32two things are perfectly correlated
11:33correlation of one or 1 the coefficient
11:36determination is one and you can account
11:39for all the variability in X with the
11:41variability in y and really that only
11:43happens if you're measuring the exact
11:45same thing two different ways like
11:46you're measuring uh height using
11:48centimeters and height using inches as
11:51long as you're measuring accurately
11:53those things those two sets of
11:54measurements will have uh r equal um 1.0
11:59correlation the variability uh in one
12:02group in height in inches is the same as
12:04the variability in that group in um
12:09ctim um okay so pretty straightforward
12:12for for B correlations uh and there's
12:14some other types of correlations we
12:15won't go into right now but the other
12:17kind of big group of tests uh would be
12:20um different types of regression
12:22multiple regression uh this is whenever
12:24you want to look at um multiple
12:27predictors how they relate to
12:29uh as they relate to each other how they
12:32predict some
12:34criteria uh when you're looking at
12:36multiple aggression uh it'll generate
12:38these things called beta weights which
12:40are conceptually like little
12:41correlations right for each predictor in
12:43the Criterion um separately and doing
12:46multiple regression looking at the data
12:48weights you can tell which predictor has
12:49the strongest relationship with the
12:51Criterion uh and you can see how well uh
12:55prediction of the Criterion can be
12:57improved by adding or deleting
12:59predictors right so if you're saying
13:01trying to figure out okay what predicts
13:03suicide uh suicide um suicidal actions
13:08and you have all these variables you can
13:09put them all on the multiple regession
13:10and figure out okay all these 10
13:13variables it's significant ression
13:14that's great but really I predict just
13:16as well with five of those variables as
13:19with all 10 so in real world sense you
13:21say okay I'm I'm only going to ask
13:23questions about those five variables and
13:24not waste the time with the others
13:26because they don't add anything to the
13:27predictive validity and that's something
13:29that multiple pression uh can do for you
13:31is tell you what things um are needed to
13:33predict some Criterion
13:36right excuse me okay um so when looking
13:41at the results of your statistical test
13:42you're always as we've talked about
13:44before looking for statistical
13:46significance uh and sometimes you won't
13:48find statistical significance you won't
13:50find kind of mathematical evidence for
13:53relationship you won't find a
13:54significant correlation or a significant
13:57um regression but in reality there is a
14:00relationship present and that can happen
14:02uh for a variety of
14:04reasons uh one of the big ones is a
14:07truncated range this has to do uh
14:09usually with how you selected your
14:11sample um so like if we say um does uh
14:16GRE predict um how well you do in grad
14:20school and the answer for most is no and
14:25for uh for you all and for most people
14:27in graduate school no it doesn't predict
14:30well at
14:31all but and here's where it get the
14:34beginning if we gave the GRE to
14:37everybody children older adults all
14:41kinds of people it would predict pretty
14:43well who would do well in uh in grad
14:46school but the thing is people who want
14:49to go to grad school who have taken a
14:50lot of Under courses who know a lot of
14:52stuff already are the only ones that
14:54take the gr so you have this trunk head
14:56you're looking at only one portion of
14:58the people in terms of academic skills
15:00and knowledge and all these things
15:02you're restricting the range this trun
15:03headed range of individuals and you're
15:06trying to predict GPA in grad school
15:08with this test score and it doesn't work
15:10for that group if we had put everybody
15:12in there and everybody had gone to grad
15:14school and had gotten the GPA then yeah
15:16it it would have a significant uh
15:17correlation but for graduate students it
15:20doesn't predict well because of the
15:22truncated
15:24range
15:25um an even kind of maybe more obvious
15:27example um
15:30uh I looked at a measure of aggression
15:33in the number of times uh you've been
15:37arrested uh most people say yeah that
15:40probably that makes sense that more
15:41arrested people probably get in trouble
15:43with law more often and more likely to
15:45get arrested probably not a big
15:46correlation but big enough sample I
15:48could probably find a small small
15:51correlation if I looked at just the
15:53people in this class probably wouldn't
15:55find that because again Trump had range
15:57and even more than that i' probably have
15:59no
16:00variability probably in this class um
16:04nobody or very few people have been
16:06arrested at least for violent crimes uh
16:09let's hope um and I can't find a
16:12correlation if there's no variability if
16:13everybody is the same on either one of
16:15the variables you're not going to find a
16:17significant correlation you can't
16:18predict variability with no variability
16:21so you have to have uh some diversity in
16:23your sample on the variable you're
16:24looking at to find significance if
16:26everybody's at the same IQ level and
16:28you're trying to predict something with
16:29IQ even if there is a relationship
16:31between those two variables you're not
16:33going to find it if there's no
16:34variability you've got to have
16:35variability in both of your variables to
16:37find any kind of relationship okay so
16:40you've got to have um a wide range of
16:42people and you have to have variability
16:44in both of the variables that you're
16:46you're looking
16:47at uh and then thirdly you might be
16:50looking at the wrong type of
16:51relationship
16:55so uh
16:59let's say you're looking at uh the
17:00relationship between um uh climate
17:04between heat uh you how hot it's outside
17:07and aggression and you get uh a wide
17:09range um of temperatures um from very
17:13hot to very cold and you get a wide
17:16large population and you find zero
17:20correlation could happen if you're
17:22looking with just kind of a regular
17:24Pierce and R linear relationship type
17:26thing because
17:29the relationship between heat and
17:30aggression seems to be curol linear
17:32right so aggression goes up as it gets
17:34hotter but then when it gets to a
17:36certain amount of heat aggression
17:38aggressive acts start to come down and
17:39it's too hot to go out and get into
17:41fights and because it goes up and down
17:45when you look at it with a in a linear
17:47way it's going to cut right across the
17:49middle of that curve and look like a
17:50correlation of zero same thing the other
17:53way if you're looking for a curve linear
17:54relationship but the actual relationship
17:55is linear you won't find simp results
17:57either so you want to make sure you're
17:59looking at the right type of
18:02relationship um okay so with uh
18:06correlational
18:09research
18:11um we we've got some limitations in
18:13terms of what we can uh conclude right
18:16we can't infer causality because it's
18:18not an experiment and because we we're
18:20kind of missing that piece and we're not
18:22assigning people to groups we're not
18:23doing stuff to people the importance of
18:25measurement validity
18:27really um increases right because this
18:29is where you can exert some uh some
18:32control and some uh some skill in
18:34designing study to make sure that you're
18:36measuring what you say you're measuring
18:38because what you're able to conclude
18:39about these about a relationship between
18:41variables hinges almost entirely on how
18:44you measure those variables right so
18:47figuring out how to measure something
18:48really well and really accurately with
18:50high with a high level of construct
18:51validity becomes very very important
18:54when doing correlational research
18:58so we know we can infer a causation well
19:01why not as you I'm sure have heard
19:04before but I'll just iterate one more
19:06time uh two main reasons directionality
19:09right if there's a correlation between
19:11variables A and B are are related we
19:13don't know if a caused b or B caused a
19:16right so if we find a correlation
19:18between um sun exposure and mood where
19:22more sun exposure is associated with
19:24more positive
19:25mood okay well is it that being out in
19:28the sun makes people feel better or is
19:30that when people feel better they're
19:31more likely to go out and be in the
19:33sun people who are depressed stay home
19:36in the dark we don't know which variable
19:38which way the the the flow of causality
19:42goes if we didn't assign people to go
19:44outside or not go outside right we're do
19:46a correlation we're asking people hey
19:48how much have you gone outside what's
19:50your mood like if we're just asking
19:51people stuff we're not manipulating
19:52anything then we don't know about the
19:55direction of the causal
19:57relationship uh the other day would be
19:59the third variable where if there's a
20:01correlation between A and B it could be
20:04that it's explained by some C
20:06relationship where a causes C which
20:09causes B um and the example I always
20:12cite is that there is a correlation
20:14between a number of asay in people's
20:17houses and um likelihood of getting
20:20cancer and having asay doesn't cause
20:23cancer but asays are associated with
20:25smoking and smoking smoking is
20:27associated in a causal way with cancer
20:30so there's that third variable that
20:31explains the relationship between
20:33between two other
20:37variables so if it can't you can't infer
20:39a causation well why would you use it
20:41why wouldn't you just do everything uh
20:44experimental uh well lots of reasons one
20:47would be you might be interested in
20:49multiple levels of a variable right so
20:53uh in an
20:54experiment um think about like a a drug
20:57study we we do uh with the drug without
21:00the drug that's two groups well say I
21:02know I want to know about um how much of
21:04the drug affects people well I could do
21:07okay um no drug low dose high dose now
21:10I've got to have a certain number of
21:11people in each of those groups right and
21:13as I well I want to do no drug really
21:16low slightly low medium slightly big
21:19really big the more kind of I divide it
21:21up the more people I have to having
21:22groups to maintain statistical power and
21:25the more tests I'll be doing to compare
21:27group a to Group B to group C Group D so
21:29on and so forth so if I really want to
21:31know about well at all levels of the
21:33drug people at really high slight Less
21:36in that what's what's the relationship
21:39with these symptoms it might make more
21:42sense to do a correlational study where
21:44I just look at how much of the drug
21:45people are taking and what their
21:47symptoms are like and then I get all
21:49kinds of variability uh and see if there
21:51is any kind of linear consistent
21:53relationship and I might find that
21:55there's not um a linear Rel maybe it's
21:57this kind of step function where uh
22:00between 0 and 10 mg there's no uh no
22:05effect then between 10 and 20 everybody
22:07in that group has the same amount of
22:09symptoms maybe maybe not but looking at
22:11the correlation I can find that I can
22:12find if it is is this weird kind of Step
22:14thing like looking graphically at the
22:16data or is the smooth linear function or
22:18it even a curv a linear
22:21relationship um so I might be interested
22:23in multiple levels of the variable um
22:25that I'm I'm interested in um sometimes
22:28times I just can't uh manipulate the the
22:32independent variable uh for ethical or
22:34or practical U
22:36reasons right if I want to know uh the
22:41relationship between um well uh
22:45infidelity repeated infidelities and um
22:48partner uh
22:50violence I can't go out and tell people
22:52okay I need you to cheat on your spouse
22:54uh six times I need you to cheat on your
22:56spouse three times I need you to cheat
22:58your spouse zero times and I want to see
23:00how often um partner violence occurs in
23:02your relationship right can't can't do
23:04that but I can go out and I can give
23:06service people just measure how
23:08frequently they um cheated on their
23:10partner and I can measure I can ask them
23:12how frequently they engaged in or were
23:15victims of partner
23:18violence excuse me and then um sometimes
23:22I don't want to manipulate an invariable
23:25or I don't want high levels of
23:26experimental control right maybe I want
23:28a more naturalistic design I want to see
23:31what happens in a natural setting right
23:36um you know this is what I really I'm
23:38maybe focusing more on external validity
23:40and not that concerned with internal
23:42validity maybe i' they've already
23:44established that you know a can cause B
23:46and I want to find if it happens in the
23:48real world right people found okay well
23:50um this kind of a silly example but um
23:53people run faster uh when being chased
23:57uh by a bigger bear
23:58right and they found that in the lab
24:00setting my show people pictures of bears
24:02and measuring the heart rate and they
24:03kind of extrapolate from that well you
24:05probably run faster but would you really
24:08is that higher heart rate really
24:09correspond to faster running or not well
24:12one one way to do it I could go out and
24:14watch people that come across bears and
24:17again ethically I can't uh um and it
24:19wouldn't be realistic to tell people
24:21okay you're going to go on the forest
24:23and there's going to be a bear I'm going
24:24to watch how fast you run that's not
24:26real world real world would be watching
24:29campers and whenever a bear comes up the
24:31person starts to run measure how fast
24:33they run and measure how big the bear is
24:35and now I've got a very naturalistic
24:37design and I can see um if that kind of
24:40lab finding really does generalize to uh
24:43the real world okay silly example but
24:47anyway U quasi experimental research um
24:52similar in many ways but also maybe a
24:54bit different uh its goal broadly um
24:57stated is to identify group
25:00differences right so um are men and
25:03women different on something
25:06are uh people who
25:08have previous military experience
25:11different from people who don't have
25:12previous military experience right where
25:14again I can't assign people to one group
25:17or the other but I I want to know these
25:18groups uh are
25:20different uh now with some quasi resarch
25:25the goal is to examine some sort of
25:27cause effect relationship where you want
25:29to exam cause and effect but you can't
25:31really do it uh in the same way as you
25:33would with an
25:35experiment um so you have to be cautious
25:38about how you report results of a quas
25:40experimental study but to make uh your
25:44argument for a cause VOR relationship
25:46stronger there's some things uh you can
25:48do right you exert as much experimental
25:50control as possible obviously then the
25:52other big thing is to measure relevant
25:54variables right so
25:56um if you have
25:59um two you want to compare two treatment
26:03groups right uh people that uh go to AA
26:07and people that go to um a a treatment
26:10program without any spiritual
26:12component and you can't if you can't
26:15assign people to go to AA or not go to
26:17AA right you you're just looking at
26:20these existing groups and it's a quasi
26:21experimental study and you and you want
26:24to uh argue maybe that there's some
26:26effect of the spirit component of AA
26:29these people will be different after
26:31doing their group who did some other
26:33group that didn't have a spiritual
26:34component so what you have to do is
26:35anticipate what people might say about
26:38that difference so if you found the
26:40groups different and you said oh it's
26:41because the spiritual component other
26:44people say well yeah but U maybe the
26:47people that went to
26:49AA um were of higher SCS or lower SCS
26:53people went to the other group right so
26:54you try to anticipate any kind of
26:56relevant variables that people might say
26:57could explain your group difference and
26:59you measure those and you try to
27:01establish that your two groups are
27:03equivalent um before you measure the
27:05dependent variable because that's the
27:07whole uh the one of the main goals of
27:11assigning individuals to levels of the
27:12infinite variable is to
27:14establish equivalency of groups before
27:17the intervention that way after the
27:19intervention is done if they're
27:20different it's because of the
27:22intervention not because of some
27:23pre-existing difference so measure any
27:25kind of relevant anything that might be
27:26relevant um pre-existing differences
27:29that people would try to uh criticize
27:31your study with and then you would say
27:32oh yeah well it could be but these
27:34groups were actually exactly the same on
27:36that
27:38variable excuse me
27:44um okay uh one other thing with the
27:48Quasi experimental uh
27:50research uh statistically uh very
27:53similar to um experimental studies right
27:56so uh frequently the exact same stuff so
27:59if you're looking at you know three
28:01groups uh you're going to be using an
28:02anova if it's two groups you probably be
28:04doing a T Test um if it's just one point
28:07in time all
28:10right um which can be confusing because
28:12then you think okay um I'm comparing
28:14these two groups using a T Test uh and
28:17therefore um you know I'm
28:19comparing may this AA group to this
28:21nonaa group do a test and okay then
28:25clearly um the spiritual a caused this
28:28difference you can't say that you may be
28:30tempted to say that because you use a t
28:31test and you think oh T Test is
28:32causation no all it is is a group
28:35difference then you can try to make the
28:37theoretical kind of rational argument
28:39for why that difference uh occurred and
28:42your argum probably be for some cause of
28:44relationship but you have to be very
28:46careful and justifying that and the data
28:49don't justify that you have to kind of
28:51rationally justify it based on other
28:53data and other kind of um theoretical um
28:57arguments
28:58okay uh so correlational quas quasi
29:01experiment
29:02research um probably more commonly used
29:07uh in our field than uh experimental
29:09research just because it's hard to
29:10assign people to groups and doing doing
29:12clinical work um so important to to know
29:14about these these designs uh and the
29:18thing to keep in mind is just because
29:19it's not experimental doesn't mean that
29:22you have you can forget about all the
29:23stuff we talked about with experiments
29:24in terms of um
29:28um uh uh focusing on uh validity and
29:32doing things the right way to uh ensure
29:35that you're making valid conclusions
29:37draw you can draw valid conclusions from
29:39your results all those things still
29:42apply um to both quas exp and
29:44correlational research okay take

About this transcript

This page contains the full transcript of RM lecture correlational and quasi experimental research with narration by DrSamFiala, generated from the public captions YouTube serves with the video. The transcript has 5,050 words across 763 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.