FW: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

[John F Hall]

Juanjo, Tom

 

For information: the R vs SPSS debate continues on SPSS and QM teaching lists under

 

[Quantitative Methods Teaching] {Disarmed} Stats and SPSS at OU]

 

John

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Karen Grace-Martin
Sent: 23 January 2014 21:58
To: John F Hall
Subject: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

 

Do I really need to learn R?

Someone asked me this recently.

Many R advocates would absolutely say yes to everyone who asks.

I don't.

(I actually gave her a pretty long answer, summarized here).

It depends on what kind of work you do and the context in which you're working.

Read the rest of my reasoning and learn more about the upcoming workshop here.

The Analysis Factor

678 Valley Rd
Brooktondale, NY
14817
US


If you no longer wish to receive communication from us:
Cancel

To update your contact information:
Update

[Juanjo Medina]

Thanks John.

I'm not in that discussion list. As I said I moved away from SPSS in my teaching and I'd probably already bored to death with my own reasons!

On Fri, Jan 24, 2014 at 7:42 AM, John F Hall <[hidden email]> wrote:

Juanjo, Tom

 

For information: the R vs SPSS debate continues on SPSS and QM teaching lists under

 

[Quantitative Methods Teaching] {Disarmed} Stats and SPSS at OU]

 

John

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Karen Grace-Martin
Sent: 23 January 2014 21:58
To: John F Hall
Subject: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

 

Do I really need to learn R?

Someone asked me this recently.

Many R advocates would absolutely say yes to everyone who asks.

I don't.

(I actually gave her a pretty long answer, summarized here).

It depends on what kind of work you do and the context in which you're working.

Read the rest of my reasoning and learn more about the upcoming workshop here.

The Analysis Factor

678 Valley Rd
Brooktondale, NY
14817
US


If you no longer wish to receive communication from us:
Cancel

To update your contact information:
Update

--

Juanjo Medina, PhD
Senior Lecturer in Criminology
University of Manchester
http://www.manchester.ac.uk/research/juanjo.medina/

Blog: De delitos y penas

Twiter: @Juan_JoseMedina

[Juanjo Medina]

But some may find this useful: http://www.r-bloggers.com/teaching-with-r-the-switch/

On Fri, Jan 24, 2014 at 7:51 AM, Juanjo Medina <[hidden email]> wrote:

Thanks John.

I'm not in that discussion list. As I said I moved away from SPSS in my teaching and I'd probably already bored to death with my own reasons!

On Fri, Jan 24, 2014 at 7:42 AM, John F Hall <[hidden email]> wrote:

Juanjo, Tom

 

For information: the R vs SPSS debate continues on SPSS and QM teaching lists under

 

[Quantitative Methods Teaching] {Disarmed} Stats and SPSS at OU]

 

John

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Karen Grace-Martin
Sent: 23 January 2014 21:58
To: John F Hall
Subject: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

 

Do I really need to learn R?

Someone asked me this recently.

Many R advocates would absolutely say yes to everyone who asks.

I don't.

(I actually gave her a pretty long answer, summarized here).

It depends on what kind of work you do and the context in which you're working.

Read the rest of my reasoning and learn more about the upcoming workshop here.

The Analysis Factor

678 Valley Rd
Brooktondale, NY
14817
US


If you no longer wish to receive communication from us:
Cancel

To update your contact information:
Update

--

Juanjo Medina, PhD
Senior Lecturer in Criminology
University of Manchester
http://www.manchester.ac.uk/research/juanjo.medina/

Blog: De delitos y penas

Twiter: @Juan_JoseMedina

--

Juanjo Medina, PhD
Senior Lecturer in Criminology
University of Manchester
http://www.manchester.ac.uk/research/juanjo.medina/

Blog: De delitos y penas

Twiter: @Juan_JoseMedina

[Juanjo Medina]

And the linked slids here make a good pedagogical case for some of the advantages of the R environment: https://stat.ethz.ch/pipermail/r-sig-teaching/2013q1/000531.html

On Fri, Jan 24, 2014 at 7:51 AM, Juanjo Medina <[hidden email]> wrote:

Thanks John.

I'm not in that discussion list. As I said I moved away from SPSS in my teaching and I'd probably already bored to death with my own reasons!

On Fri, Jan 24, 2014 at 7:42 AM, John F Hall <[hidden email]> wrote:

Juanjo, Tom

 

For information: the R vs SPSS debate continues on SPSS and QM teaching lists under

 

[Quantitative Methods Teaching] {Disarmed} Stats and SPSS at OU]

 

John

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Karen Grace-Martin
Sent: 23 January 2014 21:58
To: John F Hall
Subject: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

 

Do I really need to learn R?

Someone asked me this recently.

Many R advocates would absolutely say yes to everyone who asks.

I don't.

(I actually gave her a pretty long answer, summarized here).

It depends on what kind of work you do and the context in which you're working.

Read the rest of my reasoning and learn more about the upcoming workshop here.

The Analysis Factor

678 Valley Rd
Brooktondale, NY
14817
US


If you no longer wish to receive communication from us:
Cancel

To update your contact information:
Update

--

Juanjo Medina, PhD
Senior Lecturer in Criminology
University of Manchester
http://www.manchester.ac.uk/research/juanjo.medina/

Blog: De delitos y penas

Twiter: @Juan_JoseMedina

--

Juanjo Medina, PhD
Senior Lecturer in Criminology
University of Manchester
http://www.manchester.ac.uk/research/juanjo.medina/

Blog: De delitos y penas

Twiter: @Juan_JoseMedina

[Juanjo Medina]

Oops... apologies to the groups, I thought I was just replying to John.

On Fri, Jan 24, 2014 at 9:13 AM, Juanjo Medina <[hidden email]> wrote:

And the linked slids here make a good pedagogical case for some of the advantages of the R environment: https://stat.ethz.ch/pipermail/r-sig-teaching/2013q1/000531.html

On Fri, Jan 24, 2014 at 7:51 AM, Juanjo Medina <[hidden email]> wrote:

Thanks John.

I'm not in that discussion list. As I said I moved away from SPSS in my teaching and I'd probably already bored to death with my own reasons!

On Fri, Jan 24, 2014 at 7:42 AM, John F Hall <[hidden email]> wrote:

Juanjo, Tom

 

For information: the R vs SPSS debate continues on SPSS and QM teaching lists under

 

[Quantitative Methods Teaching] {Disarmed} Stats and SPSS at OU]

 

John

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Karen Grace-Martin
Sent: 23 January 2014 21:58
To: John F Hall
Subject: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

 

Do I really need to learn R?

Someone asked me this recently.

Many R advocates would absolutely say yes to everyone who asks.

I don't.

(I actually gave her a pretty long answer, summarized here).

It depends on what kind of work you do and the context in which you're working.

Read the rest of my reasoning and learn more about the upcoming workshop here.

The Analysis Factor

678 Valley Rd
Brooktondale, NY
14817
US


If you no longer wish to receive communication from us:
Cancel

To update your contact information:
Update

--

Juanjo Medina, PhD
Senior Lecturer in Criminology
University of Manchester
http://www.manchester.ac.uk/research/juanjo.medina/

Blog: De delitos y penas

Twiter: @Juan_JoseMedina

--

Juanjo Medina, PhD
Senior Lecturer in Criminology
University of Manchester
http://www.manchester.ac.uk/research/juanjo.medina/

Blog: De delitos y penas

Twiter: @Juan_JoseMedina

--

Juanjo Medina, PhD
Senior Lecturer in Criminology
University of Manchester
http://www.manchester.ac.uk/research/juanjo.medina/

Blog: De delitos y penas

Twiter: @Juan_JoseMedina

[John F Hall]

I think SPSS-ers should see this as well

 

John F Hall (Mr)

[Retired academic survey researcher]

 

Email:   [hidden email]  

Website: www.surveyresearch.weebly.com

SPSS start page:  www.surveyresearch.weebly.com/1-survey-analysis-workshop

 

From: [hidden email]  

On Behalf Of Art Kendall
Sent: 26 January 2014 16:07
To: Philippe Blanchard; [hidden email]
Subject: Re: [Quantitative Methods Teaching] {Disarmed} FW: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

 

If people are going to be doing exclusively a single type of research then procedures that are tailored to that single kind of research.
When I taught stat and methods I considered it important that students have some idea of the array of things that could be used.  That does not mean that they had to be able to use them all, but that they had to have some idea of what those tools would be.  This was not an extensive effort, but I had them look at all the sections of the American Statistical Association and all of the titles in the green Sage series of monographs.  I had teams of students copy or write 3 to 5 sentences about each of the techniques in the Sage series. Each team had 6 to 10 pamphlets to deal with.  They then worked together to write a paragraph for each about how it might be used in their discipline. Teams then exchanged what they wrote.

In accord with the idea of using data to aid decisions, it might be a useful exercise for some instructor to put the list of Sage pamphlets in a spreadsheet. Add a column about whether each technique would be useful in her/his discipline. Then see if each package under consideration can be used with that technique. 

One caveat about using R in forensic applications is that some of the output says that the software is developmental etc.  Attorneys on the other side get to see all output used.   They would not hesitate to use such wording to make a result seem less credible to the trier of fact, i.e., a judge or jury.

Art Kendall

Social Research Consultants

On 1/26/2014 4:03 AM, Philippe Blanchard wrote:

This long SPSS vs R discussion is interesting. From the continent, I feel delighted by such rich exchanges. Let me take a turn:

1. Juanjo's arguments about R are true. I may add that the strongest asset of R is the quantity of well trained people (mainly academics) who develop packages directly related to their research needs. No company, not IBM nor Google, would be able to do so. This is why R is definitely superior for research purpose--until the day a new language will take a turn. Unless a user is happy with using just a few functions, at the level reached by a "closed" (or at least less powerfully open) software at a given time.

2. As for teaching, all depends on what we want and what level we consider.
2.1. As for PhD students with a strong emphasis on statistics, I agree that R is better. But they do not need us to tell them that.
2.2. As for social sciences undergraduates, whose goal will be to manipulate numbers in a large range of organisations (cf. the Q-Step programme), as well as for most Master and PhD students (except statistical majors), neither SPSS, nor R, nor Stata or any other statistical package should have the lead. The most used tool, by far, is and should be the spreadsheet--namely, Excel and its equivalents (OpenOffice, etc.). 95% of all graduates not following up on the research track will be requested to use Excel and al., against 5-10% for other tools. Graduates will not have the time to prove their boss that other packages can compete with Excel, even less to convince the company to buy SPSS or to train colleagues to use R.
There are plenty of reasons for this, including Excel's commercial advance, the omnipresence of Windows, organisational inertia and the comparative assets of Excel (intuitive, versatile, not too bugged, easy for visualisation, durably maintained). These are also the reasons why SPSS, Stata, SAS and others will not disappear before a long time from institutions specialised in statitstics. They are rather adapting to the new deal by developing R-plug-ins and focusing on niche audiences.
Naturally spreadsheet software have drawbacks: Excel cannot make a scatterplot; regression results are a bit hard to read; and its programming language, Visual Basic, is not widely spread. But it is the best compromise I know betweeen stability, versatility and intuitiveness, at introductory level. Then, if the time allowed to QMs is enough, the instructor may upgrade to another software. Students are accustomed enough to computers transfer statistical concepts from one interface to the other.
2.3 As a consequence: in undergrad. schools, the choice can be made according to criteria external to research, such as the coherence between simultaneous and successive courses and the adaptation of the software to the probability of students to take up a research curriculum.
2.4 One example: in my Faculty of social and political science, for a compulsory undergraduate one-semester "Introduction to quantitative methods", we keep Excel as an introduction, then we move up to SPSS or R, according to the teacher's choice. Concretely, most teachers chose R, under the pressure of teaching assistants who more and more do not know SPSS and feel happy with that.

3. Now: Why do we debate at length about SPSS and R? Because we are researchers and we like it. Do we need to make a choice for the whole world? No, several tools can coexist. Up to the ones in minority to adapt to the demands of the ones in majority, for example like free spreadsheet adapt to Excel's norms. Should we feel sad about our favorite software's collapse? No, at least because new software always get inspiration from previous ones. This is the healthy cycle of intellectual creation.

4. Do students need to hear about this debate? No, what they need is (following up with Adian Kelly):
- relevant and doable research questions
- good quality datasets
- statistical concepts and theories
- knowledge about how computers translate and operationalise statistical concepts.
I do not think we should focus on the last ingredients. I personally try and promote software versatility among my students, simply because this will be a crucial factor of their employability and adaptability of the (many) future technological changes.

Best,
Philippe

Philippe Blanchard

IEPI-Crapul
Université de Lausanne
Geopolis 4340, CH-1015 Lausanne
[phone]

http://pblanchardweb.free.fr
http://www.unil.ch/iepi
http://www.unil.ch/iepi/crapul
http://www.unil.ch/sequences2012/
http://new.ecpr.eu/MethodSchools/MethodSchools.aspx
http://www.concepts-methods.org

 

[John F Hall]

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Philippe Blanchard
Sent: 28 January 2014 03:00
To: [hidden email]
Subject: Re: [Quantitative Methods Teaching] {Disarmed} FW: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

 

Thanks Mike. I realised my data were jammed by missing values, coded as categories ("no answer"). In this case Excel insidiously changes the data selection and truncates the data. This is the inconvenient of a software intended to a wide audience: it tries and anticipates more than students need.
I might use your manual to teach this semester, if you do not mind.
Philippe

Le 27.01.2014 17:37, Mike Griffiths a écrit :

Thanks for that analysis, Philippe.  For the foreseeable future I'm sticking with SPSS, but I do think it is a good idea for students to be familiar with Excel, at least for its flexibility with graphs.

 

A small point, but I think a typo crept in when you said that "Excel cannot make a scatterplot".  It can, easily.  The things I find most fiddly with graphs in Excel are error bars and histograms (if you call up a default error bar in Excel it is unlikely to be the one you were expecting; and histograms are quite frustrating if you are used to doing them in a few clicks in SPSS).  My attached document is a manual on graphs in Excel including how to do scatterplots (page 4), error bars (page 11) and histograms (page 12)  amongst other things.

 

If anyone has any better ways of doing any of these things, or any requests, I am happy to receive comments, on or off list.

 

Mike Griffiths

---

From: [hidden email] [hidden email] on behalf of Philippe Blanchard [hidden email]
Sent: 26 January 2014 09:03
To: [hidden email]
Subject: Re: [Quantitative Methods Teaching] {Disarmed} FW: Do I Really Need to Learn R? (and Linear Regression in R workshop is open)

 

This long SPSS vs R discussion is interesting. From the continent, I feel delighted by such rich exchanges. Let me take a turn:

1. Juanjo's arguments about R are true. I may add that the strongest asset of R is the quantity of well trained people (mainly academics) who develop packages directly related to their research needs. No company, not IBM nor Google, would be able to do so. This is why R is definitely superior for research purpose--until the day a new language will take a turn. Unless a user is happy with using just a few functions, at the level reached by a "closed" (or at least less powerfully open) software at a given time.

2. As for teaching, all depends on what we want and what level we consider.
2.1. As for PhD students with a strong emphasis on statistics, I agree that R is better. But they do not need us to tell them that.
2.2. As for social sciences undergraduates, whose goal will be to manipulate numbers in a large range of organisations (cf. the Q-Step programme), as well as for most Master and PhD students (except statistical majors), neither SPSS, nor R, nor Stata or any other statistical package should have the lead. The most used tool, by far, is and should be the spreadsheet--namely, Excel and its equivalents (OpenOffice, etc.). 95% of all graduates not following up on the research track will be requested to use Excel and al., against 5-10% for other tools. Graduates will not have the time to prove their boss that other packages can compete with Excel, even less to convince the company to buy SPSS or to train colleagues to use R.
There are plenty of reasons for this, including Excel's commercial advance, the omnipresence of Windows, organisational inertia and the comparative assets of Excel (intuitive, versatile, not too bugged, easy for visualisation, durably maintained). These are also the reasons why SPSS, Stata, SAS and others will not disappear before a long time from institutions specialised in statitstics. They are rather adapting to the new deal by developing R-plug-ins and focusing on niche audiences.
Naturally spreadsheet software have drawbacks: Excel cannot make a scatterplot; regression results are a bit hard to read; and its programming language, Visual Basic, is not widely spread. But it is the best compromise I know betweeen stability, versatility and intuitiveness, at introductory level. Then, if the time allowed to QMs is enough, the instructor may upgrade to another software. Students are accustomed enough to computers transfer statistical concepts from one interface to the other.
2.3 As a consequence: in undergrad. schools, the choice can be made according to criteria external to research, such as the coherence between simultaneous and successive courses and the adaptation of the software to the probability of students to take up a research curriculum.
2.4 One example: in my Faculty of social and political science, for a compulsory undergraduate one-semester "Introduction to quantitative methods", we keep Excel as an introduction, then we move up to SPSS or R, according to the teacher's choice. Concretely, most teachers chose R, under the pressure of teaching assistants who more and more do not know SPSS and feel happy with that.

3. Now: Why do we debate at length about SPSS and R? Because we are researchers and we like it. Do we need to make a choice for the whole world? No, several tools can coexist. Up to the ones in minority to adapt to the demands of the ones in majority, for example like free spreadsheet adapt to Excel's norms. Should we feel sad about our favorite software's collapse? No, at least because new software always get inspiration from previous ones. This is the healthy cycle of intellectual creation.

4. Do students need to hear about this debate? No, what they need is (following up with Adian Kelly):
- relevant and doable research questions
- good quality datasets
- statistical concepts and theories
- knowledge about how computers translate and operationalise statistical concepts.
I do not think we should focus on the last ingredients. I personally try and promote software versatility among my students, simply because this will be a crucial factor of their employability and adaptability of the (many) future technological changes.

Best,
Philippe

NCRM Quantitative Methods Teaching mailing list [hidden email] Email: [hidden email] to unsubscribe or [hidden email] to subscribe

 

--

Philippe Blanchard

IEPI-Crapul
Université de Lausanne
Geopolis 4340, CH-1015 Lausanne
[phone]

http://pblanchardweb.free.fr
http://www.unil.ch/iepi
http://www.unil.ch/iepi/crapul
http://www.unil.ch/sequences2012/
http://new.ecpr.eu/MethodSchools/MethodSchools.aspx
http://www.concepts-methods.org

 

--

Philippe Blanchard

IEPI-Crapul
Université de Lausanne
Geopolis 4340, CH-1015 Lausanne
[phone]

http://pblanchardweb.free.fr
http://www.unil.ch/iepi
http://www.unil.ch/iepi/crapul
http://www.unil.ch/sequences2012/
http://new.ecpr.eu/MethodSchools/MethodSchools.aspx
http://www.concepts-methods.org