THE BOOK--Playing The Percentages In Baseball

First, thanks Tom for doing this and here are the ones I have thought of:

Off season/non game physical work ethic - Just seeing if the player attempts to stay in shape.

Off season/non game mental work ethic - Looking to see if Brian Bannister’s stay in the league longer

Game physical work ethic - Players that are always diving or running into walls

Game mental work ethic - knows the game situation, takes extra base, or throw balls into the crowd with only 2 outs and people on base.

Team Player/Clubhouse Cancer - Looking to see if the Milton Bradley’s last longer then the Willie Bloomquists

Overall Heart/Drive/Grit/Desire - Can see which of the previous traits lead to a gritty player

Race

maybe add

*Clubhouse presence
*Game calling (for catchers)

Coachability

Big one I’d try to get is clubhouse influence: positive, negative.  (May just feed back what local media says, but even quantifying that could be useful.)

Also:  tough, plays through injuries (vs. J.D. Drew)

Not really inangibles, but might be good data:
Speed
Weight (assuming we don’t trust official stat)

Fearlessness (Abreu pulling up at the wall, or Eric Byrnes running right through it)

Your overall feelings (love/like/neutral/don’tlike/hate)

LookoutLanding.com ran that poll for the Mariner players, and I thought it was very interesting.  If you are a great player (Ichiro, Felix), you automatically earned alot of love.  For the good-guys / good-players (Figgins, Gutierrez), they got alot of love too.  Then you got some good love for Milton Bradley.  Overall, it was very interesting to see.

***

Of the list above:

“clubhouse presence” is like my #4, in that it’s the fan reporting what the media chooses to report.  I don’t want to have too many of those on the list.  I might knock out mine for this one.  Maybe I’ll have both, but I don’t know about that.

“fearless”: do you think there’s going to be a distinction from hustle?  I like the idea, as long as it’s something separate.

“tough”: hmmm… I like that too.  Ok, we should probably try to get two categories out of hustle / fearless / tough / heart.

weight: ahahaha… GREAT idea.  Yes, I love it.  weight.  It’s ridiculous that we have to guess this number.  We should probably put height too.

“coachability”: probably another media-reported category.  I definitely want to keep that number low.  It might be the one that makes it.

“speed”: I’ll already have fielding-speed from the fielding survey, so I don’t know that we need it here too.

***

Keep ‘em coming.

Does simple ‘intelligence’ fit in any of the already listed categories? I’m thinking of Ken Dryden/Steve Nash types, who appear to be very intelligent in a general sense, not simply in a sport-specific sense.

How about Tattoos?

Depending on the purpose, I would combine speed, weight, etc. into something more ambiguous like “looks good in a uniform” or maybe “looks like a ballplayer.”

A media fawning ratio is a similar kind of catch-all. Presumably players, coaches and front office see something that the proverbial box score misses.

I don’t know exactly what “mound presence” means, but a lot of people report on it.

On-field combativeness or red-ass-ness will be blurred into media stuff, but Carlos Zambrano will never be confused with Ted Lilly. Maybe on-field “emotion level” would work.

"we’re selling jeans?”

I’d be curious to see two versions of the survey:  one that asks for race/skin-color and one that doesn’t.  I think there is a tendency for some people to try to fill a mental quota if they are specifically thinking about race, sort of like affirmative action going on in their minds.  For example, if you ask someone for a list of smart players or players with heart or whatever, you might get a list of white guys, but if you ask the same person to give you the list, and also report the race of the person, they might notice they are only listing white guys and then try to add minority players to the list.  I wonder if people will change their ratings (consciously or subconsciously) when they are forced to consider the person’s race.  Basically, I’d like to see if minority players tend to get higher ratings or not in the survey that asks for race than in the same survey without race.

It might be tough to put two surveys out there and make sure that it is not the same people doing both but still a random sample going to each, or that people aren’t looking at both before deciding which to take.

Kincaid/11 wrote: “I wonder if people will change their ratings (consciously or subconsciously) when they are forced to consider the person’s race.”

You don’t have to wonder - there have been numerous sociological studies that have shown that asking about race *always* primes the respondent. Most famously, a study of black University students found that they performed significantly worse when asked to identify their race before starting a test; white students, on the other hand, showed no change in their performance.

But the tricky part, with a survey like Tango’s, would be to control for race. Because you would also have to ask for the respondent’s race, since those same studies have argued that white Americans will generally be less affected by the mention of race than non-white Americans. So that’s four surveys, now.

This is not as simple as it sounds, either. Race is a socio-cultural category, after all, and so definitions of race will vary among respondents, as will a) the relationship they have to their local definitions, and b) the relationship they understand others to have to their local definitions, relationships that might disagree with how others see themselves. Am I white, white-Canadian, English-Canadian, Western European and Slav, or 15/16 white and 1/16 Aboriginal? Is Carlos Delgado black, Hispanic, or black-Hispanic? Would Delgado agree with my assessment of myself? Would Delgado agree with my assessment of him? (I know for a fact that CC Sabathia and I disagree on Delgado...)

And it would also make sense to attempt to control for people who admit to holding racist opinions and those who consider themselves anti-racist and/or colorblind - all three, preferably. Because, otherwise, you run the risk of the biases appearing to balance each other out. (Granted, the existence of various biases would probably be borne out in a rather large standard deviation, but still.)

Which is to say that in order to do it well, you might have to introduce so many controls that it’s not worth doing it at all. We know that race and place of birth present biases, we know that they’ll have an effect, though we don’t know exactly what the effect will be - we just have to remind ourselves that the bias exists when we see the results.

I wrote: “You don’t have to wonder - there have been numerous sociological studies that have shown that asking about race *always* primes the respondent. Most famously, a study of black University students found that they performed significantly worse when asked to identify their race before starting a test; white students, on the other hand, showed no change in their performance.”

That was probably stated confusingly. I said it always primes the respondent and then I said white students showed no effect.

It wasn’t strictly a contradiction - the argument/explanation is that white North Americans and Europeans don’t think of themselves has ‘raced’. When they’re asked to identify their own race, they tend not to reflect on institutional and personal racism and how it affects their lives - it’s just another box to tick off.

Ask a white student about racism or other races, though, and they show a response. It has to be turned into a ‘race issue’ - which is done much more easily, clearly, when one is made to feel and experience ‘race issues’ their entire life.

The way the survey would work is that I’d have a ballot of players for a given team, and you just select “1 to 5” type for each category, exactly as I have done it with all my other surveys.

So, there won’t be a selection bias as Kincaid is suggesting.

I would just omit the skin color question.  The data you get will be of very questionable value, it might bias results, and it will likely cause some fans not to participate at all (it’s not justified, IMO, but many people will find the question offensive).

*

Might be worth including 1 or 2 questions about the fan, such as age.  That might produce some interesting correlations.

The bias would be that when people get to the skin-colour/race part, that will force some people to notice that, for example, they are tending to score white players higher, and they might go back and change things.  Or, they might see that criteria before filling out the responses and then make an effort to balance the scores of various racial groups beyond what they would do if the did not see that race was a consideration in the survey.

Forcing people to consider race will sometimes affect the answers they give, and if you want to consider the impact of race/skin-color, having the respondents fill that in themselves will probably dampen the observed effects of race on perception compared to what answers you would get if you did not ask the respondent to fill that out.  Scoring everyone 1-5 won’t do anything to get rid of that even if you aren’t asking respondents to select their own players.

The only reason I think having the respondents fill in race themselves would be helpful is if you want to look at it for one of the things Neil mentioned, where there are intersections between racial definitions.  Do respondents who consider Delgado black tend to score him higher or lower than those who consider him Hispanic?  Or something along those lines.  If you just want to see how race affects perceptions in general, I think that listing that criteria will end up masking whatever results you might otherwise get, at least to some extent.

i’d say keep it simple

IQ
clutch
professionalism (kind of a catch all for hustle, preparedness, clubhouse affect, etc. i like it as one category since individually those seem somewhat repetitive and their value minor compared to IQ and “clutch")

I also agree with those above about the race question, leave it out but bring it in after the results to see if race colors the opinions of the player.  Or confirm the media’s general opinion that anyone lazy, gritless and/or dumb are all non-white.

I’m not sure how the race question will bias the results.  The ballot would be something like:

JETER
Clutch:
Yes! Probably.  Dunno.  Prob not.  No way.

Hustle:
Yes! Probably.  Dunno.  Prob not.  No way.

Playing Smarts:
Super smart.  Smart.  Average.  Not smart.  Stoopid.

Skin color:
White.  Black.  Brown.  Yellow. Red. 

I *will* have skin color in there.  I’ve been needing this data since forever. 

So, how is answering the first three questions going to be pre-biased by answering the 4th one?

Kincaid/16: ok, I get you now.

I don’t think it will happen, but I get where you are coming from.  I don’t think someone is going to say “Torii Hunter”, mark him as dunno-clutch, smart-player, probably-hustle, get to the skin color, and think “hmmmm… I never thought about it, but Torii is black, and look, I’ve got my black player as probably-hustle, when he’s a Yes!-hustle.”

Maybe I’m too Canadian in how I see skin color.

In a similar spirit, if I did hockey players and asked if the player spoke french or swedish or russian, etc, I don’t think the responder is going to say “huh… Sedin is Swedish, and those Swedes are not that tough, but maybe Swedes are tough and I’m blind to it”.

I need that skin color question in there.  If that somehow invalidates everything else, well, let’s see.

"I need that skin color question in there.  If that somehow invalidates everything else, well, let’s see.”

How will you ever know?  You certainly won’t know it’s impact on cooperation rate. 

And putting skin color 7th on the ballot only solves your problem for the first player rated.  Assuming you want fans to rate many/all players on a team, after they do the first one they know the question is coming.  So, it’s not really a solution. 

I think you’re seriously underestimating the risk you run of hurting the whole project.  Look at your example:  “Skin color: White.  Black.  Brown.  Yellow. Red.” “Yellow?” Are you trying to detect jaundice?  This has been politically incorrect for at least 20 years—Asians, and young people generally, will be offended by this.  “Red?” Do you mean sunburned, or Native American?  Again, people will be offended, and refuse to return your ballots.  And that introduces a new bias—response skewed by respondents’ ideology, age, race—you have no way of ever measuring or accounting for. 

You’re just asking for trouble here....

Would it help to have the non-race questions come first and then, once those are submitted and cannot be changed, then ask about skin color?  I doubt anybody would change their perception of skin color based on their other responses but I understand how someone could vice-versa.

Guy, well, I need a solution that requires the race/skin category in there.

I based the selection list on google hits for:
Color terminology for race

I’m open for suggestions for how to handle it.... as long as race/skin is one of the categories.

ElBonte has a good enough suggestion, except I’m not really setup to do a two-part survey.  I may have to.

Kincaid/16 wrote: “having the respondents fill that in themselves will probably dampen the observed effects of race on perception compared to what answers you would get if you did not ask the respondent to fill that out.”

But it’s equally likely to create a new effect - to compel people to prove (if only to themselves) that they’re not racist by either judging white players more harshly or non-white players more generously than they otherwise would have thought was fair.

I should add, too, that the least important detail would be to ask respondents about their own skin color. Given that asking about a player’s skin color should get everyone thinking about race, (if they weren’t already) there are a number of fairly accurate assumptions that we can make of the respondents that makes accounting for the race of the respondents less important. (And I can explain that in more detail if anyone cares to know, but it’s complicated - suffice to say, I have reason to believe that the question isn’t necessary and controlling for the race of the respondent to a poll on this blog is unlikely to be additionally useful.)

Also:

Tango, the best option is to let people fill in the blank, as frustrating as that may be. I noted above that race is too sociological, too relative, too particular - my Pakistani-Canadian friend is considered black in England, brown in Canada but is often mistaken for Italian, considered simply Canadian by own his family, and jokes about how he’s actually white. What would I select for him, if I were given only a list to choose from? Where’s the room for ambiguity or multi-raced people? Your results will be tainted by a massive assumption - that your respondents have the same definitions of race that you do and see race the as it’s embodied in - and as it’s been culturally inscribed on - each player the same way you do.

Now, you could also try to define each of your categories but the definitions will risk looking ridiculous and there will always be people that fall outside of those parameters. And then respondents will choose one of your options, but for reasons you didn’t anticipate (ie. Delgado is called ‘brown’ because the respondent only considers Africans or Afro-Americans ‘black’ - but that wasn’t made explicit by the respondent because there was no space to do so.)

I’ve studied rhetorics of race and racism for years, now, and if there’s anything I can say with certainty, it’s that the data gathered from quantitative polls of race and ethnicity is always useless. You have to go with a qualitative approach on this one and hope that it works out and that you’re able to make sense of it and connect it with the other results in a meaningful way. The interpretation will be complicated and necessarily a bit arbitrary, but it’s the only good option.

Neil:

I will never again do a “fill-in-the-blank” survey.  I did it once, and the kind of data put in there was so unprocessable that it took me months to do what should have taken me minutes.

There are many people from the Mediterranean that would consider themselves “white”, but that others would call them “brown” or even “black”.

And, I can’t make it complicated with instructions that no one is going to read.

I want to be able to run a study that looks at hit batter records, and ask “are batters perceived as black more or less likely to be hit by pitchers perceived as black?”

There are alot of good studies that can be done with skin-perception data.

You guys tell me how best to get it from the readers.

Best approach would probably be to follow the US Census, but with less detail.  I’m sure you can find 2010 survey on line.  In general, the approach is usually to ask separate questions on:
1) ethnicity (Hispanic or not Hispanic)
2) race (white, black, Asian/Pacific, Native American, other). 
(That’s rough, not suggesting this exact language)

I don’t believe any researcher anywhere asks about “skin color,” and I can’t imagine a way to do that.

I would prefer skin color, because I am interested in perceptions.

But, if I have to go with race/ethnicity, and infer skin color that way, then that’d be fine.

2010 Census:

http://2010.census.gov/2010census/how/interactive-form.php

Question 8 and 9.

2001 Canadian Census, long form, page 10, Question 19:

http://www.statcan.gc.ca/imdb-bmdi/instrument/3901_Q2_V2-eng.pdf

Choices on Census that appear in both USA / Canada:

White BOTH
Black BOTH

Chinese BOTH
Japanese BOTH
Korean BOTH
Filipino BOTH

SouthAsian Canada (Asian Indian USA, Vietnamese USA)

American Indian USA (Appears in separate section in Canada)

LatinAmerican Canada (Appears in separate section in USA, for ethnicity)

Canada Only:
SoutheastAsian Canada
WestAsian Canada
Arab Canada

USA only:
Pacific Islander USA (Hawaiian, etc) - not in Canada

***

From this then, I would make it:
WHITE
BLACK
LATIN/HISPANIC
ASIAN

But I really need Vladimir and Endy Chavez selected as black, not latin/hispanic.  Jeter, ARod should be latin/hispanic.

Am I going to offend people if I say
“dark-skinned Latin/Hispanic”
“light-skinned Latin/Hispanic”
?

The other thing I could do is:
“Would he have been able to play in MLB before Jackie Robinson broke the color barrier?”

That last question should be safe enough, no?

The problem there is forcing people to choose between “black” and “Hispanic.” Many players are both.  And some of your most interesting data may be looking at players considered Hispanic and black vs. Hispanic and white/other.  So you need 2 questions, OR you have to specify “White/Not Hispanic” and “Black/not Hispanic”. 

And I’d include an “other” category (for Damon and Jeter, at least!).

OK, so:

White / Non-Hispanic
White and Hispanic

Black / Non-Hispanic
Black and Hispanic

Asian

***

I still like my Jackie Robinson question.

I like it too.  But too many fans, especially younger ones, would have no idea what you meant or how to answer it.

I think light-skinned/dark-skinned Hispanic (however you want to word that) would be fine, and probably a necessary distinction for the kind of data you seem to want.  Otherwise, I think there will be too much intersection between black and Hispanic.  With the distinction, you can make sure you get data that will let you classify Vlad, et al as black if you want.  I’ve seen it presented that way in surveys I’ve taken before, so I don’t think you’re really going out there if you do that.

I agree with Guy that you run the risk of a lot of people not knowing how to answer the Robinson question, particularly with Asian or light-skinned Hispanic players.  Possible even Jewish players for a small percentage of respondents.  If all you want is a simple breakdown of how frequently a player is perceived as minority/non-white, that might still work for you, and you can take someone with a 20% yes or someone with an 80% yes on that and treat them however you think is appropriate.  I’m sure you can find a way to deal with it if that’s how you really want to do it.

On a side note, I don’t think Jeter is even remotely Hispanic.

Maybe a subdivision of non-game work ethic, but I’d ask about public relations… does the player sign autographs, appear in local TV commercials, donate to charity, visit hospitals, etc. In other words, do fans see that player as part of the community that cares about the fans or not…

I’m confused. Tango/18 makes it seem like the voter chooses Jeter’s skin color, not his or her own. Is that correct?

Never mind my last post...needed to read the rest of the thread.

Guy/25 wrote: ‘I don’t believe any researcher anywhere asks about “skin color,” and I can’t imagine a way to do that.’

A lot (a LOT) of researchers ask about skin color because skin color tends to be the first thing we process in a person we don’t know. And for those people that we see but will never really know except in a fleeting or casual sense - the vast majority of the people in our communities, much less our countries - it’s also often the ONLY thing we know about them.

The danger in asking about skin color only, though, is that people will end up conflating ‘skin color’ with all aspects of race and answer accordingly. They’ll effectively out-smart the test by reinterpreting the question ("it must ACTUALLY be asking me about race and ethnicity...") and answering the question that wasn’t asked ("even though he LOOKS white, i know that he’s REALLY...").

Unless race is presented in a more nuanced way, that is.

Tango/24 wrote: ‘I want to be able to run a study that looks at hit batter records, and ask “are batters perceived as black more or less likely to be hit by pitchers perceived as black?”’

I highly doubt that the question can be answered easily, and it certainly can’t be answered through quantitative polling like this.

Only a qualitative analysis would yield, for instance, the subtle but undeniable distinction that Sabathia makes between himself (as black and Afro-American, which he believes to be the same thing) and Jose Reyes (who is Dominican and Hispanic, which he takes to be the same thing, and not black, since he doesn’t think that non-Afro-Americans can be black). Further complicating Sabathia’s racial logic is that he considers Grady Sizemore, who had one black parent, to be black without qualification, making him more black than the darker-skinned Reyes.

So if Sabathia threw at Sizemore, Jeter, etc. a disproportionately high number of times, and at Reyes, Delgado, etc. a dis proportionately low number of times, your poll would lead us to conclusions that are very well the opposite of what they should be.

...

Tango wrote: ‘There are alot of good studies that can be done with skin-perception data.’

Agreed. But it’ll be difficult to claim that this data tells you anything about how your respondents actually perceive a player’s race, given your restrictions.

A potentially big problem is that you’re polluting those perceptions because your list is too restrictive, and so it compels the respondent to perceive race in the very particular way that your list of options allows, whether it is an accurate reflection of how they would normally perceive race or not.

But there are some things you can try, short of allow a write-in. You can increase the number of options in acknowledgment of the different divisions that your respondents may draw and/or allow people to click all that apply. I’d even go so far as to recommend seemingly redundant categories: if you include all of Asian, Japanese, and Korean, you allow for those who don’t differentiate to simply click Asian and for those who feel the difference is as important as non-Hispanic black and Hispanic black to have their particular perspective recognized. (Because my guess is that, as with people who differentiate between particular black ethnicities and those who don’t, there will be important differences in the results that would be otherwise lost.)

I think you might need to follow Guy’s advice about having more than one question, too. Note, for instance, how the Canadian census (in addition to a fill-in-the-blank option) asks about country of birth, citizenship, cultural group, race, and languages spoken. That’s obviously too much for your poll, but I would think that nationality - or maybe just a region, like North American, Central America, the Caribbean, etc. - could be useful, especially since the WBC has raised our consciousness of players’ nationalities.

The additional problem of one question about race/ethnicity/nationality, as well as having one question with so few options? It makes the respondent more self-conscious of what you might be trying to extract from his response and what his response might be saying about him. In short, it *will* screw with your results.

But, if nothing else? You need to add an option for multiracial players or allow for the option to click every one that applies. But I’m going to add that I’m skeptical that this data will be able to convincingly prove what you want it to prove.

As long as the data reflect some truth, and the rest of the data is wrong, but random not systematic, then it’ll give me what I need, much like BABIP gives us what we need.

Latin/Hispanic is not a race, such as white (European)/black (African) or Asian. It’s a language group. Make Hispanic or not a separate question, as they can be Native American, white, black or any mix of those three.

Tango wrote: “As long as the data reflect some truth, and the rest of the data is wrong, but random not systematic, then it’ll give me what I need, much like BABIP gives us what we need.”

The problem with using words like ‘random’ when you’re talking about race and racism is that it’s not at all random - it’s systemic. And because it’s systemic, it can be accounted for if you’re willing to put in the effort.

Again, I understand that you want it to be uncomplicated. But from what I’m seeing and hearing, this isn’t BABIP that you’re proposing. This is batting average.

...

Brian wrote: “Latin/Hispanic is not a race, such as white (European)/black (African) or Asian. It’s a language group”

Race isn’t technically anything at all - it has no actual scientific definition - and so it’s easily adjusted to suit popular usage. (It first popped up in English more than 400 years ago, as a method of differentiating the English ‘race’ from the Irish ‘race’. And until about 200 years ago, race didn’t even strictly refer to biology - it was much more of a cultural designation.) That Hispanic has caught on as a racial designation is wholly unsurprising - think of how little sense it makes that “Jewish” has been considered ‘race’ for hundreds of years, too. Think, too, of how little sense it makes that all people who inhabit Asia, by virtue of inhabiting a space that was arbitrarily demarcated by Europeans, can be so easily collectivized as Asians, as if that suggests they’re all similar in some way.

Of course, using the word ‘race’ at all tends to upset people easily, so it might be better to avoid it altogether.

"On a side note, I don’t think Jeter is even remotely Hispanic.”

Is this a joke?

Brian: you are espousing what I think is a USA view.  Look at the Canada census form.

1.  Gives back to the community (involved in charities, has a restaurant in town, etc.)

2.  Articulate.  Gives a good interview.  Presents himself intelligently.

3.  Humble.  Appreciative of his status as a Major League ballplayer.

4.  Fluid.  *Looks* like a good baseball player.  Makes everything he does look easy.

5. Mystique.  Has a presence about him that makes him seem larger than life - almost mythical.  I’m thinking Joe DiMaggio, Eddie Murray, Reggie Jackson, Jim Palmer.

This might overlap a bit with what has been listed previously, e.g., “playing smarts” and “coachability”, but one thing you always hear said to make the difference between success or not, whether the player has lots of talent or not, is whether they can adjust when the league adjusts to them.  Or whether they can adjust when they lose fastball speed or bat speed.

So, ability to make adjustments.

Tango,
Trust me. You want ‘race’ on a separate survey.

Alex: it’s going on, and the chips will fall where they may.

“On a side note, I don’t think Jeter is even remotely Hispanic.”

Is this a joke?

No, Jeter is not Hispanic (unless by some distant connection).  His dad is black and his mom is white, but neither one is Hispanic.  Here is his family:

http://janeheller.mlblogs.com/Jeter.parents.jpeg

I am not sure why so many people assume he is Hispanic.

Tango, I looked over the linked census form. A whole lot more detailed than the current US form that still so many people were upset about. Through genealogy, I have seen all filled out US census up to 1930, and there have been many more questions asked in the past.

Back to the topic at hand, the Canadian Census never mentions ‘Hispanic’, instead listing ‘Latin American’. I believe the ‘Hispanic’ term was coined by the Nixon administration to lump the various Spanish speaking peoples into one voting block.

In my own records, I list race (white, black, Asian, Native American, mulatto, mestizo, etc), then nationality (Cuban, Dominican, Korean, etc), then lastly Hispanic (yes or no).

I have observed that very few Latin American players, especially when excluding those who grew up in the US, are white. The CIA World Factbook lists Cuba as 64% white, 10% black, 25% mixed. Almost all of their national team is black. A majority of the National Series players who have pictures posted online are black, with only a few white. Dominican Republic is 16% white, 11% black, 73% mixed, but only a couple Dominican players are light skinned (A-Rod, Wandy), most are very dark. Same with Puerto Rico, 76% white, 10% black, 4% mixed, but most players range from light brown to black.

With the race/hustle bit coming up with the Royals recently (Bloomquist and Getz made some completely boneheaded stupid plays), I think you run the race one before the hustle one, maybe even before the defensive one.

Another thought I have here is to maybe bring in some outside help.  We are always bitching here that others could have saved time and effort if they would have asked for a little help first.  This may be a time we should do the same thing and reach out for help.

The fielding one is going to run first, around Aug 23, as usual.

I don’t like to run surveys like this too quickly one after the other.  The intangibles one would probably run close to the end of the World Series.

***

What outside help would we need to ask, and who would we be asking?

Is there any reason you wouldn’t use a quantitative measure for skin color, and use the survey to ask for perception of nationality or race? 

There isn’t a perfect way to get the quantitative data, but almost all major leaguers have head shots available online.  I don’t personally have the expertise to program anything, but I know for certain that it could be done (maybe this is some of the outside help you could ask for).

You could even go so far as to have some codified (not exactly quantitative) measure of “race”.  I’ve worked with installations that determine not just skin color, but age, sex, and “race” from video footage, the same could be applied to these head shots.

"Is there any reason you wouldn’t use a quantitative measure for skin color”

I can get the qualitative data with virtually no time effort on my part.

Forgive me if this was already mentioned, but what about asking people to rate skin color on a scale of 1-10? You could include pictures of what a 1, a 5, and a 10 look like as benchmarks. So, using the Yankees as an example, maybe Mark Teixeira is a 2, A-Rod is a 5, Robinson Cano would be a 6 or 7, and CC Sabathia would be a 9. If you show people what certain numbers correspond to, it makes it easier to judge.

But it’s more three dimensional (or four dimensional).  For example, where do you put Ichiro?

I suggest using a few categories, probably the census ones, and make them checkboxes where the user can select more than one.  So for Jeter you’d check white and black, for Jose Canseco white and hispanic, for David Ortiz black and hispanic, Johnny Damon gets white and asian, and if we extend to golf check all the boxes for Tiger Woods.

Ahhhh.... checkboxes.... not bad.

having a few check boxes will also help you spot some unwanted ballots (for example, if they check off all of: white, black, hispanic, japanese, etc. that ballot might be a bad apple).

Another item I just thought of after seeing Matt’s article:
http://www.baseballprospectus.com/article.php?articleid=11861

Does player get shifted?

***

All the ballot questions will have a “1 to 5” kind of drop-down list, except for the race/skin one, which will be a combination (you can select multiple answers).

This is a last call for anything anyone wants to add.  I’ll probably run it in about 3 weeks.

Here’s what I’ve got:

================
Rate 1-5 scale
================

Clutch:
Yes! Probably.  Dunno.  Prob not.  No way.

Hustle / Fearless:
Yes! Probably.  Dunno.  Prob not.  No way.

Tough / Heart:
Yes! Probably.  Dunno.  Prob not.  No way.

Playing Smarts:
Super smart.  Smart.  Average.  Not smart.  Stoopid.

Does player get shifted?
All the time.  Alot.  Sometimes.  Rarely.  Never.

Media reports (clubhouse presence, coachability):
Media darling. Positive. Neutral. Negative. Media hates him.

Your overall feelings:
Love. Like. Neutral. Don’tlike. Hate.

================
Enter number
================
Weight:
Height:

================
Multiple Checks (Select 1 or more)
================

Race / ethnicity / skin:
- Black
- White
- Latin / Hispanic
- Asian
- Native / American Indian
- Dark-skinned
- Light-skinned
- Play in pre-Jackie Robinson days

***

I’ll probably create a ballot to guage the interest as to which of these are the most popular.