Choosing a Statistical Test for Your IB Biology IA

this is a video about choosing a statistical test for your IB biology internal assessment dunt dunt done really it's not so bad there are two main features of your experiment to consider in the first is what's the purpose of your research question there are two major purposes you can have the first is comparison the other is relationship comparisons try to understand whether there's a difference between the group's relationship tries to find a connection examples of comparison could be perhaps males versus females control groups versus treatment groups or grouping individuals by color preference in every single example we have at least two groups and we're trying to find a difference relationship on the other hand could look like for example looking for an equation between height and flexibility or asking whether age predicts muscle mass we're trying to see if medication dosage is linked to recovery time and every one of these scenarios were seeking out correlation or causation or some sort of prediction from one variable to the other the second thing we have to evaluate after purpose is the type of data we're looking at and the two main categories are categorical and continuous categorical data is qualitative even if there are numbers as text those values don't represent any numerical meaning continuous data on the other hand is quantitative or numerical as you increase the value that represents an increasing amount of that property categorical qualities could look like political party or even a yes-or-no answer or it could be which gene was expressed on the other hand if your data involves things like heart rate age or number of bacterial colonies then you're looking at continuous data even in that last example the number of colonies would only be an integer value but that still qualifies as continuous data in statistics once you figure out your experiments purpose and type of data you can choose between three main families of statistical tests those families include the chi-squared family the t-test or ANOVA family and the correlation family remember purpose has two choices comparison relationship type of data could be either categorical or continuous if your experiment is trying to draw a comparison or find a difference and you only have categorical data then you can almost guarantee that you will be working in the chi-squared family on the other hand if you're drawing a comparison we're finding a difference but your data has categorical and continuous types that would be the t-test family what does it mean to be categorical and continuous well maybe you are trying to compare the mean height height is continuous of different groups of people the groups qualify as categorical height that's continuous on the other hand if you're trying to find a relationship between continuous variables then you are definitely in the correlation family here are our three families and to summarize here are the typical characteristics once you know which family you're dealing with you can start to drill down to find a particular test that's appropriate for your experiment chi-squared is easy because there are the same two tests for any number of groups or levels in t-test family we have to know how many groups does our independent variable consider if you're only looking at one sample group you would use a one sample t-test that would be the case for an experiment where you're trying to compare the mean height at your high school to the US population for teenagers in this case you have a known average you know the population average for all teenagers in the US and you're comparing it to one sample from your high school on the other hand maybe your experiment involves two groups or two levels of the independent variable for example perhaps you're comparing the mean height of men to women and you're trying to figure out whether there's a difference in that case you would use a two sample unpaired t-test there is another kind a two sample paired t-test and that's for scenarios where you test the same group twice for example maybe you have a group of people and you give them a memory task before doing an exercise and then you give them another memory task after doing an exercise and you're trying to find out did the exercise make a difference you're looking for a difference in that case your data points have specific pairing and you're looking for a difference you're looking for an improvement so this is the scenario where you would use a paired t-test but it doesn't have to be the same group for example your paring might involve mothers and daughters or it could be any two groups the experimenter specifically paired together it you have more than two groups so three or more you'll be using a one-way ANOVA test this stands for analysis of variance this test looks to see whether or not they are all similar if you have any one group that is different you'll get a statistically significant result from this test but the thing about one-way ANOVA tests is it doesn't tell you which of the three groups or which of the four groups are five or however many it doesn't tell you which one is statistically different or if there're multiple that are statistically different for you from each other this test only looks for whether or not they're all similar or at least one is different on to the final family we will consider just the scenario where you have one independent variable and one dependent variable in those circumstances there are two primary tests you might look for Pearson's correlation and regression correlation tries to figure out how closely connected the two variables are is height a predictor of flexibility does age correlate with muscle mass so it tries to understand how well those variables go together if you know one can you predict the other one really really well do they correlate strongly or do they correlate poorly that's what correlation tries to understand on the other hand regression tries to figure out a specific mathematical equation that describes the relationship so it doesn't just want to know can height predict flexibility it asks what's the equation is it y equals 4x minus 2 and that's what regression is after it tries to find an equation so that it can make predictions for data points you may not even have measured now a really important thing to understand is that many of these tests make assumptions about your data the most common assumption is that your data is disturbed normally or it follows a Gaussian distribution that's another name for a normal distribution so if you don't satisfy that assumption you would not want to use the test because you'll get bad results your results won't be reliable the good news is that many of these tests have something called a nonparametric alternative or counterpart and here are those alternatives or counterparts for some of the tests I've shown before we go a quick caveat this video is primarily aimed at ieb biology students that's a high school class that's going to be at a college level but it's intended for a group that may never have taken any statistics classes so there are a lot of simplifications and things that we've just not discussed at all in this video for example Spearman's correlation can apply to categorical data we haven't looked at ordinal data which can sometimes classify as categorical and sometimes it's treated as continuous so there are things tests groups there are sorts of classifications that we have not included at all in this video and it's really aimed at organizing the main tests that an IB biology student would encounter so with that caveat thanks for watching I hope this has helped and stay tuned for more videos where we look at how to perform the statistical tests and calculations that we need

Channel: Daniel M

Views: 322,367

Rating: 4.9673381 out of 5

Keywords: ib, biology, sl, hl, statistical, test, statistic, which, one, how, to, decide, know, use, do I, ia, internal assessment, figure out, understand, chi, square, squared, t-test, t test, anova, regression, correlation, vs, vs., or

Id: ulk_JWckJ78

Length: 9min 57sec (597 seconds)

Published: Sun Feb 17 2019