You can also use R, Minitab, Matlab or even MS EXCEL to draw graphs and conclusion in statistics. These softwares have the capacity to handle all types of data for analysis.
Don't forget about InStat (by GraphPad) as well. It automatically guides you to the correct statistical analyses that you can/should apply to the particular kinds of data set(s) you provide. Although, with just GraphPad alone (with practice), you can accomplish the desired statistical ends as well. But InStat rounds it all out nicely - replete with additional options.
I would strongly recommend against proprietary (and relatively static) programs like GraphPad or Excel. First and foremost, the cost of these programs is usually prohibitive. For instance, a Graphpad Instat academic license is ~US$600. For non open-source programs to develop fast enough to take into account improvements in statistical approaches or visualization methods, we must pay this price. (Practitioners who choose to shirk the system and obtain pirated copies only hurt the development of features.)
More detrimental is the fact that these programs force you into certain analytical pathways. With respect to other contributors here, I disagree that these programs lead one to the correct statistical analyses. More often than not, they allow users to think that the analysis is perfect, simply because a model converges (falsely) or a parameter estimate is returned. This hurts the path towards strong inference.
A case in point: You can look at almost every journal in publication today (and almost every paper) and see standard error bars that dip below zero for positive, continuous data. This is obviously an incorrect scenario. These are often in the cases of data with low absolute values (low counts, low percentages, etc.). Most of the time, authors simply don't visualize the lower standard error, perhaps because it goes below zero. Considering all the attention that is placed on understanding basic distributions of data (and modeling and visualizing them appropriately), the prevalence of this phenomenon is troubling.How is reader supposed to have any belief in the work itself if the authors are suggesting that the body mass measurements they made on black bears or the hormone concentrations in the blood were negative?
My point is this: There is unparalleled utility in having a somewhat linear workflow when it comes to analysis. With the large and wondrous menus of programs like GraphPad, the user is bombarded with a huge array of possibility. The user can skip necessary steps (e.g. examining the original distribution of data for zero-inflation, etc.) and proceed directly to applying the best wrong model for their data. Certainly, it's the responsibility of the user to linearize their own workflow and not rush to the testing phase. But, judging from basic bar charts in the scientific literature alone, this responsibility has been shrugged off a bit.
In conclusion, I think we should be pushing open-source and, for lack of a better phrase, "long-hand" programs R for two reasons: (1) They are often free or very inexpensive. This goes a long way towards encouraging development of high-quality science in every institution around the world (especially those with new science programs). (2) These programs not only guide students towards good data workflows but allow them to think critically about often messy data. These programs are superb for visualization as well. Though graphics often take more time to produce in R than in GraphPad or SigmaPlot, spending time with graphics forces the creator to maximize information content using minimal ink.
Yup agree with Michael, we are using GraphPad Prism also. Alternative you may use SPSS. As both are licensed softwares, most researchers will pick either one depends on their institutes' budget.
Dr. Gallup's suggestion for "S" (distinct from its common implemention as the proprietary "S-Plus" program) is also not sufficient. It is now effectively implemented as S-Plus and R. S-Plus is not free. TIBCO is offering a free personal subscription for free for only a year (and US$99 for a personal license per year). Thus, over one's research lifetime of 30 years, the S-Plus user will spend $3000 that otherwise could have been avoided by using R. R is free and evolves faster than S-Plus.
(I suspect Dr. Gallup was setting some bait for this response. But I still needed to clarify this lest others take it seriously.)
Spoken like a true statistician Dr. Kelley (I am not a Dr.). Unfortunately most humans (including myself) are not bonafide statisticians, so things like GraphPad have become very popular. To expert statisticians such as yourself - this entire thread must feel frustrating. No bait here.
My narrow-mindedness in this matter comes from my limited formal training in stats. So, for graphing, I have chosen GraphPad due to my unfamiliarity with other implementations.
Hopefully others here will find your knowledge of better implementations and the associated pricing as helpful as I have. R is the way to go it sounds, and is free.
@Gallup. I wouldn't call myself a statistician either (after >60 months in the rainforests of Central America)...not to worry. This is thread is far from an insult. I hope I didn't give that impression, and I certainly hope I wasn't rude to anyone. I simply worry about the cost, especially when one of our goals should be to help young scientists in new science programs (or young scientists who have little money) become more involved in science. The cost is amazingly prohibitive. I suspect most of the recommendations for GraphPad come from individuals who joined lab that already had the program in place upon their arrival.
Anyway, good back-and-forth. These discussions are good for ResearchGate. And definitely do keep me in check if I use the letter "R" too much! I realize there are other solutions out there!
if you have never heard of matlab and R before, then i would try PRISM or ORIGIN, for ploting and estimating significance. I am sure that both are suitable and easy to use (if you know what you want, of course). ORIGIN works nicely with ms excel. PRISM is straight-forward like no other and has an excelent dokumentation. The later offers also a full powered one-month free trial.
Graphpad Prism! Get a free trial 30 day download at the website. SPSS is good for clinical data but for straight mouse/cell line data Graphpad is the best for publications and stats. And Remember, "Friends don't let friends use EXCEL" ;)
I won't belabor my previous point about the disadvantages of using programs like Graphpad Prism or SPSS that cost hundreds or thousands of dollars for student, individual, or academic licenses. Though I don't disagree that these programs will suffice for simple visualizations, I disagree that they are advantageous for widely accessible, distributed, and reproducible research. My questions to everyone:
1. Did you purchase your own copy of the program out-of-pocket, or did your university provide a copy?
2. If your university provided you a copy, how do you plan on maintaining your workflow after you switch universities?
3. How do you create reproducible and distributable analyses and distributions with collaborators who don't have access to the same program you are using?
For home I used the trial version and at work I have always been provided the program. The cost of the program for personal use is about 150, however most labs who have institutional copies will allow you to use their copy. I also have not met a collaborator yet who does not have it. It is impossible to publish anymore with excel graphs I have found. Also, that fact that Graphpad final graphs, when pasted, are editable makes it quite easy for quick changes between co-authors or co-workers. I just really like the program and would continue to recommend it.
To modify Haley's great suggestion about use of Excel, I would say "Friends don't let friends hastily use programs like SPSS, Graphpad, and JMP!" (See my lengthy comment earlier in this thread.)
Furthermore, simply having an editable graphic is not ideal practice for distributed and reproducible research. The point is well taken that this is better than nothing, but I wonder if this practice is the kind of ideal that we should be shooting for making research as open as possible.
You should try BioVinci, a free and beginner-friendly statistics web application can produces publication-ready figures in seconds. Just drag and drop your data.
You could try the Two-way ANOVA example here: https://vinci.bioturing.com/panel/workset/build/Two-way-ANOVA