MAN YOU SAVED MY LIFE WITH THIS! My stats course in undergrad was complete bull crap and now i’m doing my MS and needed to do this. And you saved me. Cheers from Auburn!
There is a far easier way to add an '*' to indicate significance 1. Select the Column 2. Right Click > Select add Label - the label will be the value of the column 3. Click select the box with the value in > delete the number > change font to larger size > type '*' > move to top of error bar
The DPI issue is easily worked around on a Mac: right click on excel graph, save as pdf, open file with 'preview'. In preview, select file -> export. Select tiff/png/whatever, and enter DPI you want. I have done this using 1200 Dpi and it looks great.
This is incredibly helpful. I used to use Prism in my old lab, but my new lab is reluctant to shell out the money for it. Thank you for making my life easier!
@@VivoPhys I past the course!! Comment of the teacher: Format Nice figures, but labels were not always explained well. So can you do a topic about labels as well XD
Great tutorial! Thanks. One thing though: Since your data set does not have a normal (bell-shaped) distribution nor is it a continuous variable, you cannot use a T-Test. It is a discrete variable and statistical testing should be performed with either a Chi-square test or Fishers exact test.
Thanks so much for making this video!! Everything was explained so well and I was able to follow you step by step!! You helped make my graphs for my lab report look ten times better!!
Thanks so much! I’m doing drugged studies with age and cholesterol…some biostatistics and l am not a math person. So it’s drug A and Drug B effects on cholesterol with age
Hi Evan. Great video! I had a quick question about what to do for calculating the SEM in a different circumstance in regard to the sample. So I have my total sample (n = 71); however this sample has been split into 5 different groups (n = 15, 14, 15, 13, 14) respectively. When calculating SEM for the n=15 group for example, do I use the total sample (stdv(of each group)/'root'71) ) or (stdv(ofeachgroup)/'root'15). I'm really confused what to use! Also, if there are outliers/anomolous results in my data, do I discount these from my calculation of my mean, and subsequent SEM? Just trying to get my dissertation finished and it's stressing me out! Hope you can help. Thank you for the video!
You would use the sample size that matches the mean value you are calculating SEM for. If it is a group mean with 15 subjects, then that is the sample size for SEM. If removing outliers from the mean calculation than it makes sense to do the same for SEM.
Hey, thank you a lot for this video. I have a question about the statistic you are using, my project looks a lot like your table. I wanted to ask you about benefits of using SEM instead of SD. Thanks!BTW (my sample is just 11 subjects)
I find that most purist suggest SD. However what you use will likely depend on the norms of your field of study. In my field SEM tends to be more common. To learn more: www.sportsci.org/resource/stats/meansd.html
It depends on how the authors choose to display their statistical results. You need to read the text below the graph to see if a symbol is defined to mean statistical significance or not. If symbols are not defined it may or may not be statically significant and you need to read further in the text.
@@VivoPhys Awesome, thank you so much for your well written response! So if they have star at the bottom to represent values of significance but the data set states ns or does not have a star, then it is not statistically significant? Does this indicate a flaw in the research/conclusion presented?
SEM is the traditional convention in physiology research. What is thought of as "correct" varies by discipline and academic journal. I generally use standard deviation for my more recent work.
Hello again. I am reading an article right now about graphing in physiology journals (link below). It states that 77.6% of bar graphs in top tier physiology journals display standard error as the error bars and 15.3% display standard deviation. I hope this is helpful! journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1002128
I now know that Excel can do error bars. I am not sure whether this info will ever be useful to me. It's better to leave these type of things to specialized tools such as R or Python. Excel is for office guys and these people don't know what standard error is (from my experience).